Journal V19 Book1

ISSN 1310-8271
-
,
19, 1, 2013
OF THE TECHNICAL UNIVERSITY - SOFIA

PLOVDIV BRANCH, BULGARIA
Volume 19, Book 1, 2013
Journal of the Technical University Sofia

Plovdiv branch, Bulgaria
Fundamental Sciences and Applications Vol. 19, Book 1, 2013
International Conference Engineering, Technologies and System
TECHSYS 2013
BULGARIA
EDITORIAL BOARD
:
. ..., ...
EDITOR-in-chief
Prof. Marin Nenchev,
DSc Eng., DSc Phys., PhD

. -
SCIENTIFIC SECRETARY
Assoc. Prof. Bogdan Gargov, PhD
1. . -
2. . -
3. . -
4. . -
5. . -
6. . -
7. . -
EDITORS
1. Prof. Sonia Tabakova, PhD
2. Prof. Michail Petrov, PhD
3. Prof. Angel Vachev, PhD
4. Prof. Andon Topalov, PhD
5. Prof. Dimitar Katsov, PhD
6. Prof. Grisha Spasov, PhD
7. Prof. Angel Zumbilev, PhD

EDITORIAL BOARD
1. . -
1. Prof. Angel Vachev, PhD
2. . . . ... 2. Prof. Venelin Zhivkov, DSc
3. . ...
3. Prof. Georgi Andreev, DSc
4. . ...
4. Prof. Georgi Totkov, DSc
5. . ...
5. Prof. Emil Nikolov, DSc
6. . ...
6. Prof. Ivan Iachev, DSc
7. . -
7. Prof. Marin Hristov, PhD
8. . -
8. Prof. Ognian Nakov, PhD
9. . ...
9. Acad. Nikola Sabotinov DSc
10. . ...
10. Prof. Marc Himbert DSc
11. . ...
11. Prof. Yasser Alayli DSc
12. . ...
12. Prof. Tinko Eftimov DSc
13. . ...
13. Acad. Yuriy Kuznietsov DSc
-3-

Fundamental Sciences and Applications Vol. 19, Book 1, 2013
TECHSYS 2013
BULGARIA
CONTENTS
1 MARIN NENCHEV, MARGARITA DENEVA, YASSER ALAYLI, SUAT TOPSU, LUC
CHASSAGNE, MARC HIMBERT
QUANTUM ELECTRONICS AND OPTICAL TECHNIQUES AND DEVICES FOR
APPLICTIONS
IN
BIOLOGY,
ATMOSPHERE
MONITORING,
OPTICAL
COMMUNICATIONS AND SCIENCES
(PLENARY REPORT-PAPER)
2 ALBENA TANEVA, MICHAIL PETROV, GEORGI SOROKOV ... 23

INDUSTRIAL NETWORK FOR TELECONTROL
3 ALEKSANDAR VUCHEV, YASEN MADANKOV, NIKOLAY BANKOV 27
EXPERIMENTAL DETERMINATION OF THE SELF-CAPACITANCE OF HIGH VOLTAGE
TRANSFORMERS IN THE COMPOSITION OF RESONANT DC-DC CONVERTER
4 ATANAS KOSTADINOV ... 31
DIGITAL DESIGN OPTIMIZATIONS BASED ON QUARTUS II ADVISORS
5 ATANASKA BOSAKOVA-ARDENSKA, NAYDEN VASILEV, IVAN FILLIPOV ... 35
FAST PARALLEL SORTING BASED ON QUICK SORT
6 BORYANA PACHEDJIEVA . 41
INFLUENCE OF THE ATMOSPHERIC TURBULENCE IN POWER DESIGN OF GROUNDTO-GROUND FREESPACE LASER COMMUNICATION SYSTEMS
7 DANIELA ILIEVA, PAVLINA VLADIMIROVA ... 47
AN APPROACH FOR PROCEDURAL TEXTURES GENERATION
8 DIAN MALAMOV, TSVETOMIR STOYANOV 51
QUALITY ASSESSMENT OF WELDED CONTACT CONNECTIONS BY THE METHOD OF
EDDY CURRENTS
9 DIAN MALAMOV ...... 57
INVESTIGATION OF THE PERMISSIBLE OVERLOAD OF THE MAIN CIRCUIT IN AN
ELECTROMAGNETIC CONTACTOR
10 DILYANA BUDAKOVA 63
INVESTIGATION OF THE EFFICIENCY OF GENETIC ALGORITHMS IN OPTIMIZING THE
SEARCH FOR THE BEST QUALITY COMPUTER CONFIGURATION
11 DIMITAR GEORGIEV, GORAN GORANOV ... 69
HALL SENSOR WITH PARALLER PORT OUTPUT
12 DIMITAR SPIROV, NIKOLAY KOMITOV, NANKO BOZUKOV 73
PSPICE MODELIND OF INVERTER INDUCTION MACHINE DRIVE
13 EMIL PENCHEV, TINKO EFTIMOV 77
A SIMPLE LONG-TAPER CORE-CLADDING INTERMODAL MACH-ZEHNDER TYPE
INTERFEROMETER
14 GALINA CHERNEVA, ELENA DIMKINA 81
CUALITY CRITERIA FOR OPTIMAL FUNCTIONING OF THE CIRCUITS
15 GEORGI GANEV 85
DEMAND POWER CONTROL THROUGH POWER STORAGE SYSTEMS
Copyright 2013 by Technical University - Sofia, Plovdiv branch, Bulgaria. ISSN 1310 - 8271
-4-
16 GEORGI IVANOV, DIMITAR SPIROV . 91

ANGULAR VELOCITY SPEED ESTIMATION OF INDUCTION DRIVES BASED ON
NEURAL NETWORK MRAS MODEL
17 GEORGI IVANOV, SEVIL AHMED, IVAN KOSTOV . 97
ANGULAR VELOCITY SPEED ESTIMATION OF INDUCTION DRIVES BASED ON
NEURAL NETWORK MRAS MODEL WITH FUZZY LEARNING RATE ADAPTATION
18 GORAN GORANOV, ISKREN KANDOV, VALERIA AYANOVA . 103
ELECTRONIC MEASUREMENTS OF LIQUIDS
19 GORAN GORANOV, ISKREN KANDOV ... 107
APPLICATION AND CAPABILITY OF ABEL LANGUAGE FOR DESIGNING IN XILINX
CPLD XC9500
20 GRISHA SPASOV ... 111
APPLICATION OF CNDEP - 485 PROTOCOL IN SENSORS NETWORK FOR
TEMPERATURE MEASURING
21 HRISTO GENDOV, PETKO PETKOV ... 115
ROBUST CONTROL OF A QUAD-ROTOR
22 ILKO TARPOV ... 121
HARMONICS AT RECUPERATION
23 IVAN KOSTOV, GEORGI IVANOV ... 125
STUDY ON ELECTROMAGNETIC COMPATIBILITY OF INDUCTION DRIVES WITH
SINUSOIDAL FILTERS IN MATLAB ENVIRONMENT
24 KRASIMIR KRASTEV, RAYCHO ILARIONOV .. 131
MATHEMATICAL MODEL OF COORDINATE TRANSFORMATION FOR 3D SCANNER
WITH 5 DEGREE OF FREEDOM
25 KRASIMIR KRASTEV .. 137
EXPERIMENTAL MODEL OF LASER SCANNING HEAD WORKING BY THE METHOD OF
LASER SECTION
26 KRASSIMIR KOLEV . 143
A COMPUTER SYSTEM FOR HYPERSPECTRAL IMAGE PROCESSING OF READY-TOCOOK MINCED MEAT PRODUCTS
27 LUBOMIR BOGDANOV ... 149
EMBEDDED SYSTEM FOR CONSUMPTION INVESTIGATION OF FUNCTION CALLS
WITH THE OPTIMIZATION PROGRAM POWOT
28 MARGARITA TERZIYSKA YANCHO TODOROV, MICHAIL PETROV ... 155
REAL-TIME SUPERVISORY TUNING OF PREDICTIVE CONTROLLER
29 MARIYA SPASOVA, GEORGE ANGELOV, MARIN HRISTOV ... 161
OVERVIEW OF NANOWIRE FIELD-EFFECT TRANSISTORS
30 MIHAIL KALBANOV, RUMEN POPOV 167
PHOTOVOLTAIC GENERATION PLANT DESIGN
31 MIRCEA WEINGART, OREST VASCAN .. 173
A STUDY OF SOME BLIND IMAGE DECONVOLUTION TECHNIQUES
32 NANKO BOZUKOV, DIMITAR SPIROV, NIKOLAY KOMITOV 181
INFORMATION TECHNOLOGY ASSESSMENT OF ENERGY SAVINGS IN INDUSTRY
33 NEVENKA ROGLEVA, VANGEL FUSTIK, SOFIJA NIKOLOVA-POCEVA, ATANAS
ILIEV, ANTON CAUSEVSKI ... 185
APPLICATION OF NEW INFORMATION TECHNOLOGY IN TRANSMISSION POWER
SYSTEM
34 NIKOLA GEORGIEV, VASILINA ZLATANOVA 189
STRENGTH MEASURING SENSORS OF ELECTRICALLY CONDUCTIVE ELASTOMERS
-5-
OPERATING AT VARIABLE ELECTRICAL MODES

35 NIKOLAY DELIBOZOV, ROSSEN RADONOV, MARIN HRISTOV 193
MEMS DESIGN USING MEMS XPLORER AND SKILL
36 PAVEL DZHUNEV . 197
ANALYSIS OF SOFTWARE DEFINED NETWORKS (SDN). MONITORING OF NETWORKS
WITH CONTEMPORARY STANDARDS AND PROTOCOLS
37 PETYA PAVLOVA, NINA STOILOVA ... 201
TECHNIQUE FOR 3D VISUALIZATION OF OBJECTS OF REAL IMAGES
38 ROMEO ALEXANDROV, ANGEL ZABTCHEV ... 205
ELECTRONIC SYSTEM FOR DEMAGNETIZATION IN MAGNETIC SEPARATION
39 ROSEN SPIRIDONOV, VANIA GARBEVA, NIKOLAI PAUNKOV .. 209
POWER MONITORING OF THE PHOTOVOLTAIC PANEL
40 ROSEN STAMATOV .. 215
AN EFFECTIVE ALGORITHM FOR LOCALIZATION OF SINGLE LINE TO GROUND
FAULTS IN POWER TRANSMISSION LINES THROUGH BILATERAL MEASUREMENTS
41 SOFIJA NIKOLOVA-POCEVA, ANTON CAUSEVSKI, NEVENKA KITEVA ROGLEVA,
VANGEL FUSTIK, ATANAS ILIEV 221
STRATEGY FOR OPERATION OF HYBRID ENERGY SYSTEM WITH INTEGRATED WIND
AND SOLAR POWER PLANTS
42 STEFAN STEFANOV, IVAN PRODANOV . 227
DEVICE FOR MEASURING POTENTIAL, INTENSITY AND CHARGE OF THE FIELD OF
ELECTROSTATICALLY CHARGED MATERIALS
43 TPNKA DVOKOV, FRANTIEK DVOEK, JI SOBOTKA, PAVEL
BRDLK, ZUZANA ANDROV, LUK VOLESK, LENKA DDKOV ... 231
GAUGE BLOCK CALIBRATION BY MEANS OF THE INTERFEROMETRY
44 SVETLANA VASILEVA 237
AN ALGORITHM MODELLING CENTRALIZED TWO-PHASE LOCKING WITH
INTEGRATED MECHANISM OF TIMESTAMPS
45 SVETOSLAV IVANOV, IVAN TANEV ... 243
CONTROL OF DC MOTORS WITH ACTIVE DRIVER CIRCUIT
46 SVETOSLAV IVANOV, YANCA KISSOVA ... 247
ANALYSIS AND SIMULATION INVESTIGATION OF DOUBLE PULSE FLASH-LAMP
PUMP LASER ELECTRICAL SYSTEM
47 TANER DINDAR, N. FUSUN SERTELLER, GOKHAN GOKMEN, TAHIR CETIN AKINCI 251
EXAMINATION OF THE DISCHARGE OF GASES USED IN THE EVENT OF HIGH
VOLTAGE
48 TANYA TITOVA, KAMELIYA VELCHEVA, VESELIN NACHEV, CHAVDAR
DAMYANOV ... 255
ALGORITHM OF RECOGNITION AND VISUALIZATION OF THE RESULTS OF
DIAGNOSTIC TESTS
49 TINKO EFTIMOV .. 261
IDEAL GAS LAWS IN THERMOECONOMICS AND FINANCIAL BUBBLE FORMATION
50 TSONYO SLAVOV, LUBEN MOLLOV, JORDAN KRALEV, PETKO PETKOV ... 267
HARDWARE IN THE LOOP SIMULATION OF ROBUST HELICOPTER CONTROL SYSTEM
51 VALENTINA VOINOHOVSKA, SVETLOZAR TSANKOV, RUMEN RUSEV 275
EDUCATIONAL MULTIMEDIA APPLICATION WITH SOFTWARE SIMULATIONS
52 VANIA VASSILEVA, KATIA VUTOVA . 281
CRITERIA FOR REFINING OF HAFNIUM SCRAP USING ELECTRON BEAM MELTING
-6-
53 VANYA PLACHKOVA, TINKO EFTIMOV ... 287

FIELD MEASUREMENT OF POLARIZATION MODE DISPERSION OF COMMUNICATION
FIBERS USING THE FIXED ANALYZER METHOD WITH POLARIZATION AVERAGING
54 VELIKO DONCHEV, KATIA VUTOVA . 293
APPLICATION SOFTWARE BASED ON NON-STATIONARY HEAT MODEL FOR
SIMUALTION OF THE HEAT PROCESSES DURING ELECTRON BEAM MELTING OF
METALS
55 VESELIN NACHEV, PLAMENA MADZHAROVA, TANYA TITOVA, CHAVDAR
DAMYANOV ... 299
TECHNOLOGY PLATFORM FOR SPECTROMETRIC ANALYSIS OF WHISKEY
56 VESELKA BOEVA . 303
A COMPARATIVE STUDY OF DTW-BASED METHODS FOR MISSING VALUE
ESTIMATION IN GENE EXPRESSION TIME SERIES
57 VLADIMIR DIMITROV 309
WORKABILITY OF MULTIMEDIA PLAYER FOR BUILDING A HOME IPTV NETWORK
58 ZDENEK BRADAC, FRANTISEK ZEZULKA, ZOLTAN SZABO, ZDENEK ROUBAL,
PETR MARCON . 313
DESIGN AND FUNCTIONAL DESCRIPTION OF EXPERIMENTAL SMART GRID
59 ZDENEK PALA, KAMIL KOLAK, NIKOLAJ GANEV ... 319
X-RAY DIFFRACTION ANALYSIS OF RESIDUAL STRESSES IN LASER WELDED STEEL
SHEETS
60 ZEZULKA F., DRAUSCHKE A., BURE Z., KREJ .I, BALCAR J., PROCHZKA M. .. 323
PHYSICAL LUNG SIMULATOR FOR EX VIVO MEASURING OF AEROSOL DEPOSITION
IN LUNGS
61 ZEZULKA FRANTISEK, BRADAC ZDENEK, SAJDL ONDREJ, SIR MICHAL, VESELY
IVO 329
SMART GRID SMART METERING SYSTEM
-7-

Fundamental Sciences and Applications Vol. 19, 2013
TECHSYS 2013
BULGARIA
QUANTUM ELECTRONICS AND OPTICAL

TECHNIQUES AND DEVICES FOR
APPLICTIONS IN BIOLOGY, ATMOSPHERE
MONITORING, OPTICAL COMMUNICATIONS
AND SCIENCES
(PLENARY REPORT-PAPER)
MARIN NENCHEV1, MARGARITA DENEVA1, YASSER ALAYLI2, SUAT TOPSU2, LUC
CHASSAGNE2, MARC HIMBERT3
Abstract: A review of part of the recent our, more important, results, obtained in our quantumelectronics oriented group at Technical University-Sofia, Branch Plovdiv, is presented. The
work is in collaboration with the Universities Sent QuentinVersailles, Paris-Nord and
CNAM Paris, France. On the base of our patented principles the new devices and techniques
and also the new ideas are developed that concern: i) high energetic dual-coaxial-channel and
dual-color laser systems and the principle of the use of their emission in new devices, related
with light pressure manipulation of sub-microns particles and bio-cells (potential for lightpressure nano-rifle and for stopper-operating device for biologic cells; ii) based on proposed
by us composite wedged interference structure new multi-channel WDM system with
independent tuning of each input / output, new continuously tunable, single-mode lasers and
tunable lasers with possibility to fix the generation at desired atomic absorption reference line,
low noise laser emission; iii) linear low-power light injection-locking amplification (~ 107 ),
device for light control by light optical transistor.
Key words: tunable dual-coaxial-channels and dual-color lasers, light pressure manipulation,
submicron particles, high-selectivity interference wedged structures, WDM-system, tunable
laser with fixing of the wavelength, light injection-locking amplifier, optical transistor
The reported results are new developments or parts of the following principal authors publications:
book-monograph (2012) Optical elements and systems in laser technique and metrology Analysis and
modelingEd. Lambert, Academic Publishing, GERMANY (2012) 400pp.; Proc. SPIE, Vol.9770/1A17.(2013) USA; J. OPT. SOC. AMERICA 27 (2010) 58, USA; OPT&LASER TECHNOL., 42(2010)
301,West.Eur.; Proc. SPIE,, Vol.9770/1A1-7.(2013) USA; OPTICAL and QUANTUM ELECTRON,
39(2007),West.Eur.; Proc. SPIE, 5830 (2005) 496 USA; APPL.OPTICS,Vol. 42(27) (2003), 5463 USA;
OPTICS COMMUNICATIONS, Vol.200 (2001) 283 ,West.Eur.; APPL.OPTICS, 40, (2001) 5402 USA,
REV.SCI.INSTRUMENTS, Vol.72, (2001) 164 USA; APPL.PHYS.LETT., 76, (2000) 131, West Eur.;
book-monograph, 2012,,"Recent research developments in quantum electronics", Ed.Transworld Research
Network, India, Belg.,Fr.,ed.in India, (2000), USA; Proc.SPIE, 4397 (2000) 74 USA; ( base for future
developments-OPTICS COMMUNICATIONS,157(1998)343, West Eur.; OPTICS COMMUNICATIONS,
130 (1998) 287, West.Eur; OPTICAL and QUANTUM ELECTRONICS,.27 (1996) 155,West.Eur;
REV.SCI. INSTRUM., 67(4) (1996) 1705, USA; IEEE J of QUANT. ELECTRON., Vol.31, 7(1995) 1253,
USA; INFRARED PHYSICS AND TECHNOLOGY, 36 (1995) 735, West.Eur; INSTR.EXPER.TECHN,
37,5, part 2, 631, (PlenumPubl..Corpor.1995 USA,); OPTICS LETT., 19, 1995 USA. Patents: Multicolour
laser, Bulg. Pat.; No 25954/1978, Tunable two wavelength laser, Bulg. Pat. No 32703 /1981, Spectral selector USA
patent No 4.468.775, 1984; Two-wavelength laser Bulg.Pat. No 96912/1992 Laser, using selectors
No 38474/1995; Device for forming of laser puls,49797/1995; Device for translation measurements, Dem. for
Pat. No 32703 /2010; Composite selector, Dem. for Pat. No 32703I/2012I.
-8-
1. Introduction
The aim of the report is to present the last
our results in the development of original: i)
specialized high-energy output dual-coaxialchannels and tunable dual-colour lasers and the
new principle of their applications for micron and
sub-micron particles (bio-cells) manipulations by
light-pressure; ii) a based of a proposed by us highspectral-selective wedged structures (composite IW,
new methods for utilization) - WDM system with
independent tuning of each input and output;
continuously tunable single-mode lasers; twowavelength tunable lasers all lasers with
independent control of both emissions; iii) laser
system for remote (up to kilometres) measurement
of small (mm) translational stretching - contraction
of an objects, iv) optical transistor; v) new solution
of tunable sub-nanosecond lasers, vi) lasers with
emission, spectrally fixed at reference atomic
absorption line and the optical transistor. An
essential part of the results is obtained in our group
in Technical University of Sofia, Branch Plovdiv
and in close collaborations with Universities ParisSouth, Paris-North, CNAM-Paris, and in the last
time - with University Sent Quentin Versailles,
France. The authors of the report are from the main
and systematic co-authors of the basic works and
have selected and systematized the materials; also,
the essential part is based on their propositions
primarily given in their patents and previous
articles. The report includes also completely new,
non-published results of the authors. The coauthors names and the other corresponding
literature are given in the cited references.
Following the limited pages for the report, we
concern mainly the ideas, principles, the
approaches for realization and short description of
some devices. The necessary physical moments
theory, experiments and the details, are given in
the cited authors works. In the report the main
attention is given on the new and actual ideas
under the development. The objective of the
works is also to establish the developed new
methods and devices as novel and competetive
components in science and practice-in the main
hardware for the indicated areas. Parallelly, new
knowledges in
quantum electronics and
interferometry are presented.
2. Development of high energy output,
coaxial-geometry, dual - channels and dualcolor, flash-lamp pumped Nd:YAG lasers.
Description of some original principles of their
applications in medical-biological investigations
As it is well known, the high-power two
color lasers are of essential interest in the distant
monitoring techniques (LIDAR) of the pollutants
and aerosols of the atmosphere, in differential

absorption spectroscopy, in the level time-life
measurement in spectroscopy, in mixing frequency
generation in nonlinear optics, in holography, in
metrology. As addition, especially the proposed by
us coaxial two-channel lasers and laser systems, are
of direct interest in the development of modern
manipulation techniques for the micro and submicro objects by light pressure (biological cells and
theirs interaction with a nano-particles), as we
discuss below. Generally speaking, many problem
do not needs of high energetic characteristics lasers,
however it is very easy to decrease the output
energy and power in many trivial ways .Contrary, to
increase the energy is complicated question. An
important part of the high energetic lasers are based
on the use of condensed matter active media solid state, liquids and on the flash-lamp or lightemitting diode pumping. The diode pumping
assures more compact technical solutions and
energetically high effectiveness. In the other hand,
the flash-lamp pumping leads easily of higher
output pulse energy and power (~1J10J and ~MWGW) and are essentially chipper and widespread in
many laboratories and users than the diodepumping. Thus, the flash-lamp pumped lasers can
be considered and actually as competitive devices
of the diode-pumped lasers. The presented in the
report methods are equally applicable for flash-lamp
pumped and side-diode-pumped lasers.
The standard way to obtain a two-color,
energetically high laser light, is to combine two
separate lasers. However, such realization is
expensive and complicated, especially for pulsed
flash-lamp pumped lasers, where also the temporal
synchronization is needed.
Simpler and essentially chipper solution to
obtain two-color and high energy output for flashlamp or side-diode pumped lasers, is to use single
active element that operates simultaneously or
during the time of the single pump pulse at two
lasing line co-called two-color lasers [14, 10]. The
development and improvement of such type lasers,
due to the noted above their usefulness, is actual
and now [10,21]. For some realization the two lights
are generate in a single volume. However, the
wavelength competition effect, especially when the
lasing at the two lines start for the common upper
laser level (exactly the case of Nd3+:YAG laser),
strongly limits the tuning range and the stability of
the emission [1,9,11]. This problem can be
eliminated if the dual-color generation is in
separated parts of the active element. The two
beams in such lasers are emitted in two closely
disposed each to other coaxial [14,20] or parallel
[10] beams. If the applications of the described type
of laser need good superposition of the lights, this
-9-
can be obtained on the illuminated object by

focusing of the two beams or at long distance due to
their diffraction.
Earlier, we have patented a flash-lamp
pumped, condensed matter, laser where a single,
active element operates in two coaxially disposed
separate parts, at two different colors [14]. The
three advantages can be marked: i) The known
effect of focusing of the pump light by the external
rods part in the axial part and thus favourising the
lasing of the inner part; ii) Secondly, the beter use
of the working volume when the operation is in
closely attached circular cross-section beams; iii)
third - following our patent and the preliminairy
works, the dual-color coaxial laser can be created
by two coaxially disposed different media the
external is Nd:YAG and the internal dye solution,
or Ruby [20]. In the last case we have more
efficient use of the flash-lamp pump energy due to
the diferent spectral region of the absorption of the
media.
Here, as point 1, we present the actual
development of our patent to created flash-lamp
pumped Nd:YAG laser that generated in two
coaxially disposed different parts of the single
active element two beams with different colors and
independent control of the parameters [14]. The
laser is designed to generate at desired pair of
Nd:YAG lines in the spectral range (1 m
1.45 m), including, in addition to its traditional
line 1.06 m, also lines 1.32 m, 1.34 m and
1.36 m with high efficiency (~30%) and 1.44 m
(however with a low efficency and low
reproducibility). Using the well known techniques
of nonlinear frequency transformation of the
generated lines and their combination, it can be
obtained the series of lines in visible, UV and IR.
The cros-section of the external beam is in the
annular (ring) beam and of the internal coaxial
beam at the radially symmetric full spot. Each
generation can be spectrally, energetically and
temporarly controlled in independent manner (two
emissions simultaneously or in controlled
sequence).
Optical schemes of two variants of the
proposed by us Nd:YAG dual-color coaxial laser
are presented in Fig.1(a) and (b). The lasing of the
external hollow-cylinder part of the active medium
and the coaxial inner part are controlled each by its
own resonator. For the internal part, the resonator
consist of the common for both resonators output
mirror M1, the prism-selective block, formed by the
60o glass prisms Pr1, Pr2, Pr3 and the end totally
reflecting prism EPr1. The mirror M1 has the
reflectivity of ~75-80% for 1.32-1.36 m and ~30%
for 1.06 m. The prisms Pr1,Pr2,Pr3 are disposed in
manner that the resonator axis passed
perpendicularly to their bisectrises, and thus the

beam incident angles are near Brewster value
(Frensels losses for the combination ~ 4 % by
pass). The dispersion of the three-prism block was
measured to be 0.1290 m for the range 1 1.4 m The axis of this resonator passes through the
holle made in the channels deviding rectangular
prism DPr, as it can be seen in the Fig.1(a). The
selected wavelength is tuned by the tilting of the
end prism EPr. The EPr position is marked by the
reflected beam of He-Ne laser on the screen E1
(reflection by the attached to the prism EPr1 mirror
AM). The prism EPr1 with a mirror AM, the He-Ne
laser and the screen E1 form the first tuning
controlled block TCBa). For the external part of the
active medium, the resonator consist of the same
output mirror M1, the rectangular prism-devider
DPr, the prism-selective block, formed by the 60o
glass prism Pr1,Pr2,Pr3 and the end totally
reflecting prism in similar arrangement as TCBa,
(block TCBb). In the picture, the two resonators are
shaded differently. The Nd:YAG crystal used has
diameter 7 mm and length of 80 mm. With precise
translation of the DPr we change the external
resonator ouput spot the near complet ring shaped
(a)
(b)
Fig.1. Schematical presantation of the optical
schemes of the developed tunable dual-color
coaxial geometry Nd:YAG lasers - (a)-using a holeprism HPr or a Brewster cylinder separstor and (b)
using hole-prism separstor and spectral-selective
mirrors.
- 10 -
form to a cickle-shaped one as it is shown below.

In the shemes, E1 and E2 are two screens that can
stop generation into the corresponding channel. The
Nd:YAG crystal and the pumping xenon flas-hlamp
(type IFP-1200, Russian non-transperence scrins,
that enable to stop the lasing in the correspnding
resonator. The separation of the cahnnel was
obtained also using a Brewster-angle cutted of one
end small-diameter of 4 mm glass cylinder, as it is
shown in the incet in Fig.1(a). This arrangement is
very effective and simple, however needs very
precisely prepared separator-cylinder and its
adjustment.
If the user have two sets of spectralselective mirrors for pairs of desired lines in the
noted range, it can use very efficient scheme shown
in Fig.1(b). Its action is clear from the Figure and
the described up schemes. However, evidently such
realisation is very expensive. This scheme was
realised by us only for generation at 1.06 mexternal resonator and 1.36 m - internal one.
In the realistic prism line-tunable system in
Fig.1(a) and prism DPr, which scheme is very easy
for realization and adjustment, we have obtained
lasing in the external resonator at 1.06 m and
efficient lasing at chosen line of 1.32 m,1.34 m
and 1.36 m in the internal part. In our laser we
have obtained up to 300 mJ for 1.06 m in external
resonator and up to ~ 65 mJ for 1.36 m (and for
simple non-selective flat-flat miror resonstor with

reflectivities for 1.06 m of the mirrors 0.99 and
0.4, the outpu energy was approximately 600 mJ).
For the realization with our, imperfect quality
Brewsters cylinder, for the noted lines, the output
was 270 mJ and 35 mJ. Some photograph of
realizations are plotted in Fig.3. Using a passive
Cr4+:YAG Q-switch, we have obtained for 1.06 m
the pulse duration of 100 ns. Simultaneous Qswitching of both generations can be obtained
using a rotating prism Q-switch, described by us in
Ref.[10] for the parallel axis resonator separation.
The other solution of dual-color coaxial
channels lasers, described by us earlier [14.20] and
later by other authors, lies to use combination of
two active media dye solution and Nd:YAG or
two-solid state crystals e.g. Nd:YAG and Ruby (in
development Nd:YAG- Er:YAG laser).
Fig.3. The photograph of the spot of the crosssection of the laser beam. Left, only for the axial
channel open: top- with the centred holle of the
prism; middle with small prism translation;
bottom the two channels are opened. Right spot,
when the system is realized with a Brewster-cilinder
channel separator (thermo-sensible paper).
Fig.2.The actual photograph of the two-coaxial

channels Nd:YAG laser. Top - general view and
bottom the tract with a divaiding prisme with hole.
this order for the other two lines). The electrical
pump energy was 100 J. For the same pumping in
The principle of the applicatin of the developed

two-coaxial-channels laser for the microns and
submicrons objects (e.g. bio-cells) fixing and
treating, is shown in Fig.4.(a),(b). The principle of
the other new application of the two-coaxial beams
(from one laser such as the described up or from
combined pair of lasers) is shown in Fig.3 [14].
This device, action of which is based on the light
pressure, is able to canalize and to accelerate the
canalizing microns or sub-microns particle. The
device uses two coaxial Gaussian laser beams, onethe external with large diameter and with the hole
inside and the second with a small diameter,
propagated in the internal empty cylinder in the
external beam. The internal beam can be nonGaussian only with a parallel propagation. The
action of the devices can be understood from the
Fig.3. The external beam is focalized by the lens L1
into the cuvette with particles, and the internal -
- 11 -
with a short focal length lens L1. The focal point is

very close to the input wall of the cuvette, thus to
produce a near parallel part close to the wall. The
external beam, due to the light pressure and related
Dual-color coaxialbeams laser output
at a
beam
at b
at a Focusing system
Internal coaxial
beam at b
External beam at
Focal spots
Fixed particle
(biological cell)
Solution with free

moved cells
(a)
sequential laser pulses

focalized
beam spot
(d~m)
particle
(biological cell)
transportation of
the micro-particle
(possibility for acceleration)
(b)
Fig.4. Illustration of the principle for application of
the developed coaxial-beam emitting laser for
manipulation of micro-objects by light. (a) Fixing
by the light pressure of the micro-object cell,
using the focalizing external hollow cylinder beam
and intervention with the focused internal beam. (b)
Pushing of the micro-object by the focused semiring cross-section laser beam.
Fig.5. Schematic of the canalizing-accelerating

technique using the light pressure (light pressure
nano-gun).
with its specific Gaussian distribution, collects the
particle inside its hollow cylindrical part as it is well
known. At a controlled moment, the internal beam,
with high pulse intensity, is released and at the focal
point, due to the light pressure, accelerate the
collected particle. The important possibility for the
developed by us coaxial-channel laser is that for the
axial internal beam can be introduced passive Q-
switch between DPr and the first selective prism

Pr1. Thus, the high duration, free generation annular
pulse collects the particles and the appeared after
some delay giant pulse (or CW) pushes the
collected ones. To obtain a Gaussian character
external beam, the end flat mirror M4 is replaced by
the convenient concave mirror M5. We have
calculated that for the parameters of our laser and
the dimension of the particles of order of 1 m with
density of water, this is ~ 106 m/s2 (in water solution
the speed ~ 10-4 m/s2) that is close to obtained such
values in the literature. We have realized parts of
the proposal, the study is in progress. The system is
of interest for study the effects of bombarding the
particles with high accelerated particles. The action
of the system, in principle, is similar to the action of
hunters rifle with the particle being analogical of
the drops in the rifle.
3. Wedged interference structures and
theirs applications
3.1. Wedge-interference structures
some studies by us theory and experiments,
new-properies and new principle of applications
The element, used and studied by us, that is
base of series of new and competitive our proposals,
including for this element a new development
theory, experiment, new basic property and new
ideas for its realization and for practical
applications, is an wedge interference structure
Interference wedge (hard compact version of the
Fizeau Interferometer). [17,18,14,19]. In general,
the Fizeau interferometer or Interferential Wedge
(IW) consists of two reflecting plane surfaces ,
consisting a small angle (~ few rad) and separated
by a gap with linearly increasing thickness to the
opposite side of the wedge angle. In the literature,
except in laser technology, are presented the
applications in metrology and spectroscopy, in
surface topography, in phase-shifting Fizeau
interference microscopy and as a tunable filter in
laser design [18, 19 and the literature sited therein].
As it is discussed in our works
[17,18,19,14,], as a rule, theoretical and
experimental analysis of the IW properties by other
authors were conducted mainly for the case of
infinite plane-wave illumination assuming an
extended wedged structure; different wedge
applications have been also analyzed for this
particular case. Inspired by IW incorporation in
laser design, over the recent years we focused our
efforts on the study of compact wedged interference
structures under illumination with a narrow light
beam of small diameter. We have succeeded to
reveal unique properties of the IW when it is
illuminated by laser light, and to propose various
applications, thus ensuring IW to be a competitive
- 12 -
optical element in laser resonator design

[13,1,18,19,4].
We have patented and developed two new
principles of the IW use: i) on the base of found by
us new property of the IW a spectral selective
non-Snellius reflection use as a tunable by linear
translation high spectral selectivity (~0.01 nm),
near perpendicular disposition, reflector [15,13].
Attractive applications in singlemode laser and in
ring lasers technology are considered in our
earliest articles [13] and in more recent new
solution of continuously tunable single-mode
lasers, of high spectral purity lasers and in lasers
with spectrally fixed emission at reference
absorption lines [13,19]; ii) The principle to use in
combination the three properties of the IW - to be
simultaneously: high selective filter for resonant
wavelengths, near totally reflecting mirror for the
non-resonant wavelength and tunability by linear
translation. We have patented a structure of two
and more IW-s that is competitive element for
tunable laser technology and especially for optical
communications as a new type of WavelengthDivisionMultiplexing (WDM) element with
tunable outputs-inputs [15,14,18].
Below, we have presented the patented and
developed WDM-structure, as an important
possibility for competitive application in the optical
communications and quantum electronics. The
consideration will be in complex with our new ideas
for improving the IW-selectivity and respectively
the selectivity of the IW as a tunable spectral filter
and of the proposed WDM-structure.
Fig.6 represents general scheme and the
principle of the operation of the proposed new
WDM structure for the cases of multi-wavelength
input or output beam. With IW1,2,3 is noted the
sequences of IWs that form the WDM structure. For
resonant wavelength in the incident beam the
corresponding IW is transmissive filter and for offresonant near totally reflecting mirror. The highreflectivity coatings (~95%) IW of thickness of 5300 m acts simultaneously as a spectrally selective
transmissive filter for resonant wavelength and a
channel coupler, being a nearly totally reflecting
mirror for the off-resonant wavelengths. In addition,
tuning of transmission maxima is provided by
simple translation of the IW in its plane and does
not affect the geometry of the system. Respectively,
when we tune one of the channels, this does not
disturb the transmission resonance of the others.
Inversely, the structure gives the possibility to
obtain superposition of individual beams which
have passed through each IW, thus serving as a
multi-beam multiplexing element. Analysis of the
WDM element requires first to analyze the
behaviour of a separate IW for laser (coherent)
beam illumination. On the base of our mathematical

model, adapted for computer simulation of the
Input
Fig.6. Schematic presentation of a basic principle

of a new WDM structure.
transmittance and reflection of the IW for a limited
diameter laser beam (~1-1.5 mm). We have studied
in details the IW spectral behavior. [17,18,19].
The calculations for the transmitted IT and
reflected IR intensities were made for tree types of
interference wedges. The first one is a sandwich
type IW, formed by sequential deposition on the
~1mm silica-plate a dielectric mirror with 90%
reflectivity (this order, for the working range), a
wedged transparent layer with optical thickness of
5 m. and a second, 90 % reflectivity dielectric
mirror. The second type of the IW is the silica
wedge with optical thickness 300 m having
dielectric layers on both surfaces of equal
reflectivity of 90 % in the spectral region of
calculation around = 630 nm. The third type is an
air gap two-mirror composed IW. The apex angle of
the wedges is = 0.04 mrad. One example of the
calculation is shown in Fig.7.
IW thickness
Fig.7. Calculated transmission IT and reflection IR

curves as a function of the wavelength for IW (, air
gap 20 m wedge, beam diameter of 1 mm. ).
The same curves are typical for all type of IW-s.
The sandwich type IW, that is compact hard
- 13 -
version of the Fizeau Interferometer is very

convenient for application in the new WDM
structure due to its compactness.
The experimentally measured transmission of
the 300 m optical thickness IW along the axis,
perpendicular to the apex angle is shown in Fig.9.
The experiment and the theory are in agreement.
IW transmittion, ~
1
0,75
0,5
0,25
0
0
4
6
IW translation, mm
10
- e.g. = /F, where F is the fines factor,

depending on R. Thus the desired low value of
leads to low value of . This limits (to ~ 0.01 nm)
the selectivity of the proposed optical system.
The principle of our solution of this
problem is based on the use of composite wedged
interference structure. It can be understod from
Fig.11 where is given schematically one example of
composed two-component structure. It consist of
one thick wedge e.g. e1= 200 m optical thickness
silica glass wedge with to dielectric mirror at each
wedge plane with reflectivity of R= 0.95%. The
wedge angle of the plate in the example is
1= 200rad. On the one of the mirrors is laid a
transparent wedged layer with optical thickness
Fig.8 The measured transmission of the 5 m,

R1=R2=0.9, =0.05 rad IW for =633 nm.
The study of the IW behavior for illumination with
short laser pulses, including the ~ 0.5 ns pulses,
shows the properties, similar to those in the case of
CW beam illumination. In Fig.10 are presented the
typical computed curves for the 5 m sandwich
type IW irradiated with pulses of duration 0.5 ns
The incident beam is with a Gaussian spatial
distribution as it is shown in Fig.9. In axes x is the
distance along the beam impact area and in axis y
the relative passed intensity. Thus, IW can be used
and for frequency repetition rate to ~10GHz [12].
Fig.9. Transmission and reflectionof

R1=R2=0.9 for 0.5 ns pulse illumination.
Fig. 11. Schematic of the new composite wedged

interference structure.
e2=10m and wedge angle 2. The relatively
simple calculation gives that if the angle 1 and
thickness e1 and the angle 2 and the thickness
e2 are chosen to be in the relation
2 = 1 .e2 / e1
the change of the resonant maximums of both
connected wedges with the translation of the
5um,
For sandwich type structure the thickness is

technologically limited to a few m, and
respectively does not permit to obtain a
transmission line low that few part of nanometers.
Also, there is a general problem for the traditional
IWs to obtain selection of a narrow line in
combination with high separation between the
resonant lines (Fig.8). The calculations shows that
there are a completely similar dependence between
the width of the selected line and the spectral
distance between the lines as this one for FPI
Fig. 12. Computed transmission of the composite

wedged structure, formed by two wedged elements
layer of 10 m and silica glass of 200 m with
convenient wedge angles and the tuning.
composite system in its plane is exactly equal. In
this system the thick wedge gives a very low
spectral width of the transmission of all system
(~0.05 nm) and the thin wedge select only single
- 14 -
resonance of the thick wedge at high spectral

range (~ few nm ). Typical example of computer
calculated resonances at the described system is
given at Fig.12.
We realized experimentally our WDM
structure for the case of CW beam illumination
using a laboratory model of a free-communication
system. We superpose the emission of two He-Ne
lasers - at 0.63 m and at 0.59 m respectively
and of the frequency doubled Nd:YAG laser
(0.53 m). The beam diameter was ~ 1 mm. The
condition of the experiments is noted in Figs.1314, with real laboratory model, shown in Fig.15.
In the start position we measured the resonant

wavelength 1. When the hammer stretches the IW
transmission resonance changed and is for new
wavelength 2 that is obtained by the tuning of the
illuminated tunable laser. By tuning the laser, we
can obtain a new resonant wavelength, 2
corresponding to the new wedge thickness. From 1
and 2 it is easy to obtain the expansion.
Fig.16. Principle of the device, based on IW, for

distant measurement of small translation of a rigid
object in its plane.
Fig.13. The experimental set-up presents the

realized WDM device, composed of 2 sandwich
type, 5 m thickness IWs, each of them to a separate
one channel yellow and red, respectively
Fig.17.Calculated dependence of the resonance

wavelength at different points along the IW.
The other technical realizations are given in [12].
Fig.14. Visualization of the wavelength separation

(using a smog) of the three channel-separation
(colors) by the new WDM-arrangement green,
yellow and red. Sandwich type 5 m IWs. The
laboratory WDM is shown in Fig.15
IW1
IW3
IW2
Fig.15. Realization
of the compact,
three-channel model
(by three sandwich
type IWs) of the
proposed
WDM
element.
We have proposed also a new utilization of the

IW specific properties by designing devices, which
allow distant (from few meters to kilometers) laser
measurement of small (~mm) linear expansion of a
rigid object. The principle can be understand from
Fig.16, which shows as an example a measurement
of small linear expansion of a steel hammer-beam.
Three systems with thick IWs (IW, IWS1

and IWS2, e ~ 300 m, =400rad) are presented
below. They are simple and practical devices with
tunable narrow line (~0.01 nm) transmission in
Fig.17. Selection and tuning of a single

transmission resonance of a high thickness IW.
large range (~few nm) [14]. The selected single
resonance is continuously tuned. In Fig.17, using a
conveniently angle between IWS1 and IWS2
permits to obtain for the combination the effect of
vernier and thus to selects only single transmission
resonance. Effect of vernier can be achieved using
two near equal thickness wedges with conveniently
chosen thicknesses e (e.g. with e and 0.9e), however
the problem is to assure two convenient IWs. These
- 15 -
two systems are convenient for application in the

proposed WDM structure.
In the system in Fig.18 the vernier effects for
the transmission resonances is obtained by chosen
two sequent convenient angles of incident of the
beam, passed two times through the single thick IW.
The system needs of a single IW and is very
convenient for selective resonator applications.
Fig.18. Selection and continuous tuning of a single

transmission resonance using the vernier effect,
obtained by chosen two sequent convenient angles
of incident of the beam, passed two times through
the single thick IW[4].
wedge in its plane [13,12,19]. Calculation of the

reflection for IW with e= 20m, R1=R2= 0.9, is
presented in Fig.20. In the insets are given the
photograph of the reflected non-resonant and
resonant spots and the scheme of single-mode laser,
using this property of the IW.
3.2. Original multi-wavelength lasers
with independent control of each wavelength
based on the application of the developed in the
part 1) WDM structures.
The group has a substantial experience and
results in the development of two-wavelength lasers
[13,1, 1,7,11]. Using original resonators we have
developed with the corresponding theoretical and
experimental background the first in the world two
wavelength
Titan-Sapphire,
F-Centers
and
Yb:YAG lasers. In development of our preliminary
works [15,14] and applying our multi-channel
WDM element, new and simple solutions of multiwavelength lasers with independent tuning of the
wavelengths are proposed. The scheme in Fig.20 on left is with emission at all wavelengths in single
The important, found by us, property of the

IW is to provide the resurgence in a part of the
Fig.20. WDM multi-wavelength laser resonator

schemes with independent tuning at each
wavelength; left with output in a single beam,
right with closely parallel outputs.
Fig.19. An important, found by us, property of the

IW to provide in reflection the resurgence of the
energy in the end (practically out) and at noSnellius angle for a small diameter beam [13].
Calculated curves [13,12]. In the insets - the real
photographs and selective RIW-cavity laser scheme.
energy of incident beam at the end (practically
outside the incident beam) and at non-Snellius
angle. This property permits the use of the IW as a
new spectral selective element a high selectivity
reflection mirror, tunable by translation of the
laser beam. As we have discussed in Sec.2, the

general problem for this case is the wavelength
competition. We have proposed a solution, using an
active mirror [3] hat force the low-gain generation,
however with complication of the lasers. A scheme
in Fig.20 - right, is with a closely disposed parallel
beams, generated in different parts in the active
medium and eliminates the competition effect. If it
is necessary, superposition can be obtained after
focusing or at long distance from the laser.
Actually, we have practically realised
(Fig.21) a two-wavelength semiconductor laser
using red laser diode (max output ~ 1 mW) with
antireflection coated output surface. The laser
operates successively at two wavelengths by
application of the modified architecture, given in
the inset in Fig.21. As a rule the lasing starts firstly
- 16 -
in one of the channels. To obtain lasing also in the

other channel we slowly increase the losses for the
started generation, in practice by misalignment of
the end mirror in its channel. At each tuning of
characteristics theory, experiment, for this laser

are given in Fig.24.
Fig.23.Calculated
two-wavelength
emission
in diode-pumped CW Yb:YAG laser (two pairs of ).
Fig.21. Actual photograph of the operating twowavelength semiconductor red laser. The
generation is at two wavelength 1 and 2 in two
reference outputs and in a single output.
one wavelength, it was necessary to balance the
losses in the channels (tuning range of ~ 2.5 nm
with two lines of FWHM ~ 0.1 nm, Fig.22). The
laser output is ~ 0.2 mW. Nevertheless, the
difficulty for tuning and limitation, such laser can
be useful in many practical applications when the
user dispose with only one laser diode (but high
power type and respectively expensive).
Fig.24. Tuning characteristics of the twowavelength CW Yb:YAG laser. One wavelength- 2

is stopped, the other- 1 is tuned. At each position of
1 the losses in the channels are balanced. The solid
lines are computed curves. The triangles (for 1)
and. circles (2) are experimental point.
By the oscilloscope investigations of the
temporal behavior of the two emissions, (generation
in a single and in two volumes), we have proved
that
the
two
emissions
are
generated
simultaneously. In Fig.25, as example, is shown
two-wavelength operation in single diode-pumped
CW Yb:YAG crystal (near 1.03 m); the same in
ns-scale.
Fig.22. Tuning curves (experimental) for the

described two-wavelength laser. One wavelength2 is stopped, the other- 1 is tuned. At each position
of 1 the losses in the channels are balanced.
We have developed the theoretical models
and computer simulation of the multi-wavelength
laser operation that gives the possibility to predict
and optimize laser characteristics [11]. The base is
conveniently modified rate differential equations
system. Some example of calculated CW twowavelength emission is given for an end- diodepumped CW Yb:YAG laser (Fig.23). The tuning
Fig.25. Two-wavelength emission from diodepumped CW Yb:YAG crystal-oscilloscope traces.

In Fig.26, we show, realized by us, other
type, very simple and effective solution of twowavelength CW laser that emits at two wavelengths
in spatially separated beams, generated in single
- 17 -
pumped volume (application to CW Yb:YAG and

to diode-lasers). The advantage is that such solution
is very congruent for application in small dimension
pumped volume such as an end diode-pumped or
semiconductor lasers.
frequency of the selected mode (due to the change

of the resonator axis length) and this one of the
transmission of the IWHS. The system is very
convenient to be modified for ring laser geometry.
The real spectrogram of tuning is shown in
Fig.27(c).
Fig.26. Schematics of easily realizable twowavelength cw Yb-YAG (or diode) laser. M1 is a

dichroic mirror; M7 and M9 are flat mirrors, M8 is
a concave mirror at distance from M1 equal of its
radius of curvature; FP1,2 FP2- FabryPerot
etalons; IW-Interference wedge, L3 lens.
Fig.27(c). Fabry-Perot spectrograms of a typical

mod-jumping tuning (top) and continuously tunable
single-mode emission (bottom), obtained by the
described schemes. The tuning range is ~ 3GHz.
3.3. Simple continuously tunable singlemode lasers using high spectral selective IW
We present examples of realizations with
dye active medium. The schemes are easily
adaptable also for diode lasers. Two schemes are
shown in Figs.27(a) (development in [13], base
[14]) and 27(b). In Fig.27(a) AM is the active
medium. The PR is isosceles triangle prism, IWHS is
high selective Interference Wedge (Fig.11, or the
shown below complex wedged structures). The
IWLRS is the pre-selector. The apex angle of the
prism and the refractive indices are chosen in this
manner that when the PR is translated in direction
The shown schemes are very simple and cipper. The

limitation for large practical applications is related
with the nedded high quality surfaces of the IW
formed the layers (or mirrors). As example, the part
of the surface of an abitual plate in nanometers
scale (AFM microscopy) is shown in Fig.28. There
are variation of the flatnest that provide during the
tuning mode jumping. The improvement of the
technology actually is in continuous progress.
PR
IWHS
tuning
Me
15 nm
(a)
pump
Gr
IW
AM
RIW
(b)
Fig. 27(a),(b). Schematic of continioously tunable

single-mode lasers based on eual change the mode
frequency by varing the cavity length and aselecting
point of high-selectivity IW.(a)- Using a prism and
IW. b- IW and mirror M is translated jointly in
appropriate angle . In incet- when the RIW is used.
of its bisectrice, the variation of the resonator
optical length to provide equal change of the
Fig.28. The separating layer surface for habitual

silica-glass 200 m IW in nanometers scale
3.4. The use of the RIW gives an effective
solution of low-noise narrow-line (including
continuous tunable single mode) lasers.
The base of one type of solutions, is to
employs the property of the RIW, especially with a
different reflectivity of the consisting mirrors
(layers) to be simultaneously the narrow-line
spectral selective selector (on side the low
reflectivity mirror), narrow-line transmission filter
and high reflectivity mirror at the side of highreflectivity mirror [18]. Also, the RIW works at a
small declination (mrad) with respect to the optical
axis of the system. The described property is shown
by the top picture in Fig.29, and the corresponding
calculated curves at the bottom picture [18,19].
- 18 -
Using an RIW with mirror of R1=0.94 and R2= 0.74

and e=200 m, in the dye laser-amplifier system
scheme, shown in Fig.30 top, we have obtained
narrow-line (~ 0.01 nm), tunable (575 595 nm)
with a noise less than 1%. Note that the use of
standard system with flat output mirror leads to
noise of ~ 5% as it can be seen from Fig.31.
4.1. Scheme with implementation of

original methodology using two-channel WDMsystem based resonator and active mirror
The details are given in [3]. The essence of the
principle is to restrict the starting pulse-like
generation in some lasers with subnanosecond
starting pulsations (spikes) to a single pulsation
by using an active mirror to forces the competitive
damping generation in the second selective channel
(Fig.32). The theorety and experiments show the
improvemen of the shape and shortening the
subnanosecond pulse when use our technique
(Fig.33). The strong spiking provides the pulse
pump dye laser with electro-optical cutting the
relatively long time rising pulse of Nd:YAG
(0.53 m) pump pulse to obtain a very short rising
start front (~1ns). The calculated starting generation
for the last case is given in the inset in Fig.32.
Fig.29. Schematically illustration of the property of

RIW with non-equal mirrors to be simultaneously
spectral-selective reflector for the resonant
wavelength in side of low reflectivity mirror and to
be a simple high reflectivity mirror for side of high
reflectivity mirror and a high selectivity
transmission filter the top picture. The
corresponding calculated curves are given in the
bottom picture [18].
Fig.30. Use of RIW (UMIW in the picture) in the

low-noise dye oscillator-amplifier system with
unequal mirrors as a tunable feed back element and
non-reinjected the luminescence element The
wedge is declined at a small angle (mrad) with
respect to the system axis.[18]
Fig.32. Set-up for selection of a single subnanosecond pulsation by active mirror in twocannel cavity [3]. The top inset spiking with a
cutting of the pump pulse.
Fig. 33. Typical oscilloscope traces (5 ns/div.) of

the optimized selected spike: left for the known
technique of competitive resonators and rightfor the proposed our AMIR- approach [3].
Fig.31. Spectrograms of the laser light after the
amplifier for the case of use of scheme of Fig.29left and for the scheme wit a simple mirror as
output of the oscillator right (the same scale).
4. Generation of subnanosecond pulses
4.2 The other, new our technique uses

pumping by short time superposition of the two
parts, obtained by high speed electro-optically
cutting the laser pump pulse for a dye laser [8].
The system is shown in Fig.34(top). The pump
pulse is typical ~ 30 ns, Q-switching produced pulse
- 19 -
(0.53 um) from Nd: YAG laser that pump a dye

laser. Theory and experiment are shown (Fig.34)
W, s-1
beam, after passing through an optical isolation

system (polarizer and a quarter-wave plate)
3.10 7
1,8.10 12
1.10 7
6.10 11
3,6.10 7
2,4.10 12
1,8.10 7
1,2.10 12
Fig.34. Sub-nanosecond pulse generation by short

time superposition of two parts, obtained by electrooptically cutting
pump pulse (bottom-right and
middle). The short time up-threshold pumping,
limits the spiking generation to the first spike
(~0.2 ns) bottom -left [8].
5. Development of the lasers with fixed
frequency of the emission at reference atomic
absorption line (LFF)-new all optical techniques
5.1. A simple technique for producing of
a pulsed semiconductor laser light, spectrally
narrowed and fixed at a chosen absorption
atomic line [2].
The technique, utilizes conventional narrowline (single mode) diode laser with scanned
frequency of generation around the desired
absorption line by pump current scanning. The
diode laser output is fed to a modified Michelson
Interferometer (MMI). The control of phase and
amplitude correlation between the interfering beams
in its two arms is achieved by using a substance
with desired line of locking, introduced in one of
the arms of the interferometer and the plate glass
lame(s) in the other. The unbalance for the locking
is produced by the absorption and the refractive
index changing throughout the absorption. The
unbalance is varied tilting the glass plate. Theory
and experiment show that under properly chosen
conditions the spectrum of the obtained light
partially overlaps the atomic line and has a linewidth, comparable to this one of the absorption
In Fig.35 a commercial single mode pulsed
diode laser (line-width 100MHz) was used. The
wavelength of the selected mode of the laser
repetitively scanned (forward - backward) within
10 GHz (~ 0.0210 nm) around the 852.1 nm Cs
absorption line by the pump current modulation
within 5 mA around 44.3 mA. The diode laser
Fig.35. Set-up for producing diode laser light

spectrally fixed at the Cs absorption line.
impinged the entrance beam-splitter of a MMI
composed from the beam-splitter and the mirrors
M1 and M2 . The beams reflected from M1 and M2
interfered at the beam splitter and formed the useful
interferometer output (Output 1). A part of the light
beam from the Output 1 illuminates the receiver
PhD1. and part of incident laser beam after splitter
SP and the reference Cs cell - the PD2. Three cases
of control by varying the position of the glass plate
are shown in Fig.36. For the optimized conditions
(Fig.36c), achieved by appropriate tilting of the
glass plate the locked line is practically a single
Fig.36. Spectrograms of the diode laser light

emitted from Output 1 (curve A) and of the
reference Cs cell absorption line ( B; inverted )
line and with a linewidth (~ 1.7 GHz, or 0.0035 nm)
that is comparable to this one of the absorption line
(0.9 GHz, 0.0019 nm) and overlaps the absorption
of approximately 45% (2 mW Output 1, 10 mW
output of diode laser). The reported technique can
be useful in variety of spectroscopic applications
when the target is a single transition which should
be excited .
5.2. The other simple systems of spectral
locking of the laser emission at the reference
absorption line are also developed [5]. Its
principle lies with the disturbance of the
- 20 -
competition between two injection-controlled

generations (Pout2 and Pout1) in two-channel
resonator laser (amplifier) in one active medium
when the injected light frequency is scanned-Fig.37
We describe such amplification with:
i
dN 2
N
= R p N 2 Bi q i + N 2 Bc q c ) 2
dt
dq i
q
Pinj 2
= Bi q i N 2 Va i +
(1 R2 )
dt
h i
c
dq c
q
Pinj 1
= Bc q c N 2 Va c +
(1 R1 )
dt
h c
c
max
TA, P out2/P out2
1
max
P out2/P out2
0.8
Ta
Ta
0.6
0.4
P inj1/P inj2=5
0.2
10
15
20
0
-583
Dl A
583.5
- Dl
A /2
584
0
584.5
Dl
A /2
Dl585
A
Fig.37.Example of an injection-locking system and

the locking curves Ta is the absorption line, Pout2/
Pout1 is the ratio of the two outputs versus .
6. Developing of the described up
principle, we have introduced a new injectionlocking linear amplifier of amplitude modulated
laser light [6]. We considered the case of
amplification of few injected lights at different
wavelengths (multichannel amplification). The light
amplification is of order of ~ 106 and more (from
W to W) with high linearity. To amplify linearly
the injected modulated laser light in the ring laser
(the simplest practical arrangement of the amplifier)
we introduce in opposite side counter-injection that
compete with the modulated light and provide thus
linear amplification. The principle is shown in
Fig.38. The light for the amplification Pinj1 (t) is
injected through partially transparent mirror M1 of
the ring laser (using our WDM system to combine
few different light). In the opposite direction, at
different wavelength, is injected the counterinjection with a power Pinj2 = const.
IW
1
injected beam
...i (~W)
??
i (W)
1 1
cw
c
d
Fig. 39. Example of action of our ring-amplifier. (a)
input signal ,(b) and (c) amplification without and
with counter-injection,(d) Fourier spectrum - c
M1
active
medium
IW
IW
Pump
M2
Here qi, with i=15 and qic is the photon numbers

in the opposite directions, N2 is the inversion
population; Va active volume, Rp pumping rate,
Bi(i) emission cross-section,c,-photon and upperlevel lifetimes. In Fig.39 is selected the calculated
one input and corresponding amplified curve. The
top pair is without counter injection: the input
sinusoidal signal (a) and the amplified signal (b).
The enormous distortion can be seen. The bottom
curves are in the presence of counter injection the
amplified signal (c) and the Fourier-spectrum (d).
The excellent linear amplification can be seen. In
Fig.40 are given the same type of calculated curves
for pulse modulated light for amplification.
IW
CW counterinjection
cc (~ 1mW)
at ?(~1mW)
(
c
ccw
?2
? 3
IW
?4
M3
output ccw
?
1
1 ..
.?
i
?i
output cw
amplified (x10 6 ) light
Fig. 38. New amplifier, based on injection locking

technique with counterinjection.
Fig. 40. Example of action of our ring-amplifier for

the case of pulse modulated light (a) pump light
for the amplifier, (b) input signal, (c) and (d)
amplification without and with counter-injection.
- 21 -
We have shown the ability of our amplifier to

amplify simultaneously and linearly a number of
injected beams with different frequency at a large
distance of ~ 800 GHz. The nonlinear distortions,
defined by the harmonics relative power are lower
than 1%. Such amplification is possible in very
wide range of ~ 2400 GHz. The calculated curves
input- amplified output for 4 wavelengths are
plotted in Fig.41.
Fig.41. The calculated
input-amplified output for
four
wavelengths
in
presence
of
counterinjection for a dye laser.
The linear amplification of
~107times (to ~ 5 kW) can
be seen for low-power
injected light (less than 0.2
mW).
The described technique is of practical interest for

application in optical communications, in different
systems for optical measurements.
7. Device to control light by light (in action
analog to the transistor-optical transistor)
The new interferometer type device for light
control by light (DLCL) uses, on one hand, the high
sensitivity of the Fabry-Perot Interferometer (FPI)
or IW to the losses in the interferometers gap. Our
original idea is to use the possibility to illuminate
the volumes of the end of the gap by two ways: i)
through the interferometer mirrors (beam A - in
Fig.42); ii) directly into the gap (beam B)
IP
Nd :YA
laser
OS
OR
GP
PC
BS
M4
OP
beam B
IFP
4+
Cr :YAG
(IFP plate)
DLCL
M1
PP B
M3
PPA
beam A
M2
IFP mirror
Fig.42. Schematic diagram of a Cr4+:YAG- DCLC

and of the experimental set-up for forming
controlled duration rectangular laser light pulse.
OR optical receiver-synchronizer, PC-Pockels
cell, GP-Glan Prism, MR1R,MR2R- high reflectivity
mirrors. The high speed switching PC (~ 1-2 ns),
activated near the maximum of the input ~ 30 ns
pulse, switches the polarization and the GP forms
two spatially separated pulses that act in the
described manner upon the Cr4+:YAG gap FPI .
If the gap is filled with saturable absorption medium

and the mirrors are high reflective e.g. 0.920.99,
the beam A will affect the saturable absorber
transmission only by transmitted small part through
the mirror and respectively the FPI transmission
will be drastically low for this beam. When the
beam B illuminates directly the saturable absorber
the effect of this illumination is very strong (no
decreasing the illuminated light intensity by the
mirror). Thus, with the low power beam B we can
control in efficient manner (or to open and stop) the
FPI or IW transmissivity for beam A. One first
application of the new optical transistor will be to
form rectangular nano- and sub-nanosecond pulses as
it can be understood from the Fig.42. The optical
transistor is in the dashed ellipse. In Table 1 are
given the calculated transmission T of the Cr4+:YAG
gap filled FPI when only incident beam on the
mirror illuminate the IW and in the bracket - when
the auxiliary (controlling) beam for direct gap
illumination is applied.
Table 1
Example of transmissivity of new DLCL with
Cr4+:YAG as a storable absorber .
(Cr4+:YAG: (0.01-0.5)J/cm2=(2.36-0.6)cm-1).
R of the
mirrors
IFP
Thickness, mm
Illuminating
beam energy
density,
J/cm2
Controlling
beam energy
dens, J/cm2
T
%
0.99
0.99
0.4
0.2
0.5 (0.5)
0.5 (0.5)
0 (0.1)
0 (0.1)
3(8.6)
3( 21)
0..99
0.92
0..52
0.1
0.5 (0.5)
0 (0.1)
9(40)
2.65
20
0.5 (0.5)
0.5 (0.5)
0 (0.1)
0 (0.1)
9(40)
1(10)
8. Conclusion
In the paper are selected and systematized
authors results from the last years, concerned the
development of a quantum-electronics and optical
devices and approaches, directed for application
manly in noted in the title region of the practice and
sciences. The essential part is based on the new
development of the principles, primarily given in
our patents and articles. The report includes also
completely new, non-published results (noted
respectively in the presentation). The objective of
the repport is to establish our resolts as novel and
competetive methods and components in science
and practice-in the main hardware for the indicated
areas.
Acknowledgements
The works is supported by a few contracts
with FSI-Bulg. (804,VUPh-12,RILA 01-7/19/No
25197 VB) and with NIS-TU-Sofia (063-17,08019). M.N. and M.D thank prof. Y.H.Meyer, prof.
J.C.Keller, prof. R.Barbe and prof. E. Stoykova for
their help (except noted in the cited literature
participation) for the realization of the works.
- 22 -
REFERENCES
(the main author works for the report and as ideas
for new developments; cited other authors therein).
1. Gorris-Neveux M., M. Nenchev, R. Barbe,
and.J.-C. Keller, A two-wavelength, passively
self-injection locked, cw Ti 3+Al2O3 laser, IEEE J.
Quantum Electron.Vol. 31, 1995, 1263-1260, USA
2. Deneva M., E.Stoykova, M.Nenchev, R.Barbe,
J.C.Keller, Diode laser emission, spectrally fixed
at atomic absorption line. Optics & Laser
Technology, Vol.42, 2010, 301-307, West. Eur.
3. Deneva M., P. Uzunova, M. Nenchev, Tunable
subnanosecond laser pulse operation using an
active mirror concept. Opt. Quant. Electron.,Vol.
39, 2007, 93-212, West. Eur.
4. Deneva M., E. Stoykova, M. Nenchev, A novel
technique for a narrow-line selection and wideband
tuning of Ti 3+Al2O3 and dye lasers. Rev. Sci.
Instrum. Vol.7, 1996, 1705-1714. USA
5. Deneva M., M. Nenchev, Development of
original, simple quantum electronics device with
emission passively frequency locked at atomic
absorption line, Proc. Intern. Confer.Laser
technology and Lasers, Bulg., 2005, 37-45
6. Deneva M, Sv. Saparev, M. Nenchev, J.-P.
Wallerand, M. Himbert New linear laser
amplifiers of a periodically modulated laser
radiation based on an injection-locked method
Proc. SPIE, Vol. 4397 ,2001,79-84, USA
7.Deneva M., D.Slavov, E.Stoykova, M.Nenchev

Improved Passive Self-Injection Locking Method
for Spectral Control of Dye and Ti:Al2O3 Lasers
Using
Two-Step
Pulse
Pumping
Opt.
Communications,Vol.130,1997,287-298,West Eur
8. Delev A., M.Deneva, M.Nenchev, E.Stoykova,

D.Slavov,Tunable subnanosecond pulse generation
in a dye laser using overlapped pump pulses, Rev.
Sci. Instrum.,Vol. 72, 3, 2001, 164- 168, USA
9. Keller J.C., R.Barbe, M.Deneva, M.Nenchev,
Unidirectional ring Ti3+:Al2O3 laser generation at
the wavelength of an atomic absorption line by
bidirectional passive self-injection locking. Appl.
Phys. Lett., Vol.76, 2000, 131-133 , USA
10. Kissov H., M.Deneva M.Nenchev Development of a tunable, competition less flash-lamp
pumped Nd:YAG laser generated of a chosen pair
of two lines. Proc. SPIE, 8770-54 ,2013,1-7, USA
11. Louyer Y, J. Wallerand, M. Himbert, M.
Deneva, M. Nenchev Two-wavelength passive
self-injection controlled operation of diode-dumped
cw Yb-doped crystal lasers. Appl.Opt,Vol. 42,
2003, 4301-4315,USA
12. Mindizov N, V. Pashova, M. Deneva, E.
Stoykova, M. Nenchev New devices for

applications in lasers and optical communications
based on the wedged interference structures. Proc.
SPIE, 7747-11, 2011, 1-8, USA
13. Nenchev M, Recent research developments in
quantum electronics. , part in the book-monograph,
pp.27-55, written by M.Nenchev, Ed.Transworld
Research Network, S. Pandalai, P. Gallion, I.
Veretennicoff, India,Belg.,Fr., 2000, ed. in India
14. Nenchev M, Multicolour laser, I Bulg. Patent
No 25954/1978); Deneva M., Nenchev M. Coaxial
two-colour Nd:YAG laser, submitted for publication
15. Nenchev M., Two-wavelength tunable laser
Bulg. Patents No 32703/1982 and No742983/1986;
Nenchev M,.Meyer Y,, Optical Selector Devices,
Brevet dInvention No 8021621, France/1980; US
Patent 4,468,775/1984
16. Slavov D, M. Nenchev Study of approaches
for spectral control of Ti:Sapphire lasers. Opt.
Communications,Vol.200,2001,283-301,West Eur
17. Stoykova E.V., M N.Nenchev. Gaussion
Beam Interaction with Air-gap Fizeau Wedge.
J. Opt. Soc. America, Vol.27,2010,58-68; USA
18. Stoykova E., M. Nenchev Fizeau wedge with
unequal mirrors for spectral control and coupling in
a linear laser oscillator-amplifier system.
Appl.Opt.,Vol.40, 27,2001,5402-11, USA
19. Stoykova E., M. Nenchev, V. Saynov Optical
elements and systems in laser technique and
metrology Analysis and modeling. bookmonograph,Ed. Lambert, Academic Publishing,
Germany, 400 pp. (d. 2012, in English)
20. Todorova P., M.Deneva, M. Nenchev A new
flash-lamp pumped and Q-switched Nd:YAG
optical quantum generator with simultaneous lasing
at two colours (1.06 m and 0.94 um). Proc.
Intern.Confer.Electronics ET-2008Bulg., 24-27
21. Uzunova P, S.Arabadgiiska ,Tz.Uzunov, H.
Kissov, N. Kaimakanova, E. Dinkov, M. Deneva,
M. Nenchev Study the penetration of IR laser
radiation in human teeth: determination of the
absorbed and scattered parts. Proc. SPIE, 8770-54,
2013, 8771A-1-7, USA
1) Department of Optoelectronics and Laser
Engineering, Technical University of Sofia, Br.
Plovdiv, 25 Ts. Diustabanov St.,4000 Plovdiv, BG
(marnenchev@yahoo.com; mdeneva@yahoo.com)
2) University St.Quentin, Versailles, France
3) University CNAM, Paris, France
Received 01.04.2013
- 23 -

TECHSYS 2013
BULGARIA
, ,
: (NCS) -
. :
.
- , .

.
, - .
,
, - , .,
.
,
Panasonic.
.
(RTU)
.
,
, .
: , Panasonic
INDUSTRIAL NETWORK FOR TELECONTROL

ALBENA TANEVA, MICHAIL PETROV, GEORGI SOROKOV
Abstract: Networked Control Systems (NCSs) have recently gained the attention on the industry
since they are flexible and re-usable. In this way they can also provide component redundancy,
adaptability and be less invasive by utilizing wireless technologies. The NCSs are attractive
area and many industrial companies (Panasonic) and institutions have shown great interest in
applying various topologies to remote control and manufacturing automation. The industrial
automation systems can be very complex, and they are usually structured into several
hierarchical levels. At the field level NCS, as a distributed topology, includes sensors,
actuators, I/O modules, drive units, which are communicate with automation systems over a
powerful, real-time communication. This paper is focused on a Panasonic laboratory NCS. For
such systems, Panasonic PLCs offer a secure basis for the future as telecontrol main station,
RTU, data collector or complex automation solution. The investigations for network delay are
conducted with the laboratory set up. It is possible to develop an algorithm based on predictive
control strategy to overcome the transmission delay in the NCS.
Key words: Networked Control Systems, Panasonic telecontrol
Copyright 2013 by Technical University Sofia, branch Plovdiv, Plovdiv, BULGARIA. ISSN 1310 - 8271
- 24 -
1. Introduction
A feedback control system wherein the
control loop is closed through a real time network is
known as a Networked Control System (NCS). It
includes fieldbus control systems constructed on the
base of bus technology (e.g., DeviceNet, Modbus
ControlNet and LonWorks) and Internet based
control systems (using general computer networks).
Therefor the servers (e.g. FP Web-Server) are
capable and can work as an Ethernet-to-serial
interface converter. As there are more and more
applications in industry, such as: oil refining,
petrochemicals, central station power generation,
pharmaceuticals, food and beverage manufacturing,
cement production, steelmaking, papermaking and
even spaceflight, more attention in this area has
been paid to design and analysis of NCS.
Generally, there are three types of NCS
methods: Type1scheduling methods that guarantee
network QoS (quality of service); Type2 control
methods that guarantee system QoP (quality of
performance); and Type3 integrated scheduling
and control methods that consider both QoS and
QoP. For Type1, the following scheduling methods
have been developed: scheduling method MEF
(Maximum-Error-First) based on the MATI
(Maximal-Allowable-Transfer-Interval) [5], and a
sampling time scheduling method of network
bandwidth allocation and sampling period decision
for multi-loop NCSs. For Type2, there are many
control methods developed for NCS, for example,
augmented deterministic discrete-time model
method [2], queuing method, optimal stochastic
control method, perturbation method, fuzzy logic
modulation method [1], event-based method and
predictive control [3]. For Type3, the following
problems have been studied: the optimal sampling
period selection problem for a set of digital
controllers, the sampling period optimization
problem under the schedulability constraints, and
the NCS analysis and simulation problem solved by
two MATLAB r_-based toolboxes: Jitterbug and
TrueTime. Internet based control has also been
considered for practical applications: like Internet
based device process and controls [6], Internet
robots and Internet based multimedia education.
Various control approaches have been developed
for NCS. Hence this paper introduces a laboratory
system for telecontrol, which incorporates field bus
(serial communication) and high (information) level
Ethernet protocols in a real time NCS. It is possible
to conduct the measurement of the network delays
and to investigate the different algorithms for
improving the data exchange. The investigations for
network delay are still challenging and unsolved
area. An approach to actively compensate for the
random network delay is not available yet. The very
new way to find the solution is to incorporate the

predictive strategy to compensate the network delay
in NCS, [4]. Hence this is an active way, which is
named as the networked predictive control.
Moreover this new challenge, incorporating the
predictive control strategy in the NCS, requires the
stability analysis of closed-loop systems. Through
laboratory system will be able to conduct many
investigations and experiments.
This paper is organized as follows: Section
2 is devoted to the communication of the NCS with
Modbus-TCP and FP Web Server. In Section 3 the
laboratory NCS is presented. In Section 4 are
presented the developed web application and some
notations for the network investigations. The
conclusions in Section 5 are focused on the future
work and attempting for network operation
improving.
2. Description of the industrial network
Nowadays Ethernet is being touted as the
solution to virtually every network-related
automation problem, for examples: device-level
communication; control; high-level data transfer.
According to a study by Venture Development
Corporation (VDC), Ethernet is projected to take a
significant increase in market share and becoming a
major element in the industrial networks. It is
known that it was not originally designed for realtime control. This is one of the reasons to looking
for better solution to overcome the undesirable
network delays and information gaps in the
industrial application. In this way of improving the
transmission delays can be summarized into three
types:
Constant delay
Random delay, which is independent from
transfer to transfer
Random delay, with probability distributions
For the networked architecture the control
delay k, for the control system can be denoted. The
subscript k is used to indicate a possible time
dependence of the delays.
(1)
k ksc kc kac
where the tsc is the communication delay between
the sensor and the controller; tck denotes the
computational delay in the controller; tca is
communication delay between the controller and the
actuator. According to the network topology and the
communication protocols different delays may
occur. At the present a tendency is observed to
adapt the Ethernet at the field level control. It is
achieved by interface converting. The Panasonic
Web-Server unit provides up to 3 different port
numbers, fig.1. Existing fieldbus and RS485
- 25 -
protocols (PLC-Link, MEWTOCOL, Modbus)

can be used for this multipoint communication via
Ethernet.
3. Panasonic laboratory set up

The laboratory equipment consists of FP
PLC, Asynchronous Motor, Frequency Inverter, FP
Web-Server, GT industrial LCD terminal, and PC.
It provides the ability to manage and motor control.
The inverter in turn is connected to and controlled
by logic controller (PLC). Between the inverter
output and motor the current transformer is
included. In this way is able to monitor the current
that passes through motor windings. The overall
scheme with signal directions is depicted on the
figure 4.
Fig.1. The main 3 different port numbers of the

communication unit
More than the data could be delivered to
multiple receiver stations and used for multipoint
PLC data exchange via Ethernet. An example for
the Modbus-TCP client that communicates with two
remote PLCs (slaves) via FP Web-Server is shown
on fig.2. Any Modbus-TCP client can access the
Modbus-TCP server of the FP Web-Server via
Ethernet. Hence it can communicate with the PLC
connected to: the FP Web-Server's 3-pin RS232C
port; the serial ports RS485 or USB of the FP Web
expansion unit. On the fig.5 is presented the FP
Web-Server and the Web-Expansion units used in
the laboratory networked system.
Fig.2. An example: Modbus-TCP client two PLCs

(slaves) via FP Web-Server
Fig.3. The configuration Modbus-TCP server for

the PLC established for the laboratory system
Fig. 4. The elements of the Panasonic telecontrol

system
The communication network includes the
presented on the fig.5 units. By using the Web
Server system data can be read or modified in the
local network or through the web. The Panasonic
FP-WEB Server can work as an interface between
the LAN or WAN and all PLC family. In this work
it is configured with FP Web-Configurator tool ver.
2.3. The parameters to the MEWTOCOL server
port are: 3-pin RS 485 server port (related to the
additional FP Web expansion unit, fig.5 and 9094
TCP port number. The MEWTOCOL port acts as a
server. It waits to be connected with a client, e.g. a
PC or a transparent client port. A feature here is that
all clients have to use the same 3-pin RS232C
/RS485 serial port to communicate with the PLC.
Due to the limited data bandwidth of this serial port,
the response time can increase drastically.
Expansion unit
1. USB host port
USB 1.1 host
port for
Panasonic
products that are
supported by the
FP-WEB2.
2. RS485 Screw
terminal.
3. High-speed
digital output
Optocoupler,
phototransistor
output.
4. SD card slot
SD/SDHC
memory card
support
Fig.5. The main communication units
- 26 -
5. Conclusions
In this work the laboratory networked
control system with Panasonic: PLC and FP Web
Server was introduced. The basic concept of the
OPC server client with Modbus-TCP protocol was
used to configure the industrial communication. The
NCS is very fruitful and promising research area.
By adding the feedback in developed system can be
conducted the investigations about network delays
in the close loop mode.
Fig.6. The laboratory Panasonic system
On the figure 6 is presented the photo of the
laboratory system. In the PLC run mode can control
the motor either via web or GT panel. Data from
specified devices is collected and stored in the FP
Web-Server unit's RAM and written as log files to
an SD memory card of the FP Web expansion unit
in CSV format when specified.
4. Web applications
The Panasonic FP Web-Server enables
connection of the logic controllers to the global
network through simple web application. To reach
data from or through the web it is necessary to
develop a web application (e.g. an applet). The
Panasonic FP Web Designer software tool is for
industrial web applications development. In this
work it is used for developing the simple web page,
shown on the figure 7. The designet web can be
loaded in the (any) web brouser. The project consist
of a main page and a few sub pages and active links
to the local control system. It is possible from it to
activate the frequency invertor and hence to start the
motor. On the other hand can monitor the motor
consumption. In general this application is made for
telecontrol. Morover the system and the project can
be extended with additional devices. The project for
the laboratory system is necessary to be transferred
to the target system with the FP Web Configurator 2
tool. In this way it is guaranteed the efficiency of
the project.
References
1. Almutairi NB, Chow MY, Tipsuwan Y.
Network-based controlled DC motor with fuzzy
compensation. In: Proc. 27th Annual Conference of
the IEEE, Industrial Electronics Society, Denver,
CO, 3:18441849, 2001
2. Halevi Y, Ray A. Integrated communication and
control systems: Part I Analysis. J Dynamic
Systems Measurement and Control 110:367373,
1988
3. Liu GP, Mu J, Rees D. Networked predictive
control of systems with random communication
delay. In: Proceedings of the UKACC Control,
Bath, UK, 2004
4. Tang P., De Silva C., Compensation for
Transmission Delays in an Ethernet-Based Control
Network Using Variable-Horizon Predictive
Control. IEEE Transactions on control systems
technology, vol. 14, No. 4, July 2006
5. Walsh GC, Ye H, Bushnell L. Stability analysis
of networked control Systems. In: Proceedings of
the 1999 American Control Conference, San Diego,
CA, 4:28762880, 1999
6. Yang SH, Chen X, Edwards DW, Alty JL.
Design issues and implementation of Internet based
process control. Control Engineering Practice
11:709720, 2003
Acknowledgement
The authors would like to acknowledge to
the partners in contract No: I-694/21.12.2012.
Department of Control Systems
Technical UniversitySofia, Branch Plovdiv
25 Tsanko Diustabanov St.
4000 Plovdiv
BULGARIA
E-mail: altaneva@abv.bg
E-mail: mpetrov@abv.bg
E-mail: sorokov@automation.bg
Fig. 7. The developed web application for

laboratory NCS
11.03.2013 .
- 27 -

TECHSYS 2013
BULGARIA

DC-DC
, ,
:
.
DC-DC
CO2 .
.
: , , DCDC
EXPERIMENTAL DETERMINATION OF THE

SELF-CAPACITANCE OF HIGH VOLTAGE
TRANSFORMERS IN THE COMPOSITION OF
RESONANT DC-DC CONVERTER
ALEKSANDAR VUCHEV, YASEN MADANKOV, NIKOLAY BANKOV
Abstract: A method for experimental determination of the self-capacitance of the windings of
high frequency, high voltage transformers is proposed. The transformers are used in a resonant
DC-DC converter for CO2 laser power supply. The determination of the self-capacitance of the
windings is realized during the operation of the resonant converter at no load mode.
Key words: self-capacitance, high voltage transformer, resonant DC-DC converter
1.
DC-DC

,
,
.
.

,

[6].
,

[4, 5].

. .
.
- 28 -

,
-
- ...
. ,

[6].
[5]

.

,

[2].

DC-DC
,
CO2 .
2.
. 1

[7].
- ,

,
L
,
[6].
3. CO2
. 2
,
CO2 [1].
(T1T4), (LS
CS),
(Tr1Tr4)
(Rect1Rect4),
(Tube1Tube4).

,
(C1C4),

(ZVS Zero Voltage
Switching). .

.
Tube
Tube
Tube
Tube
Rect 1
Rect 2
Rect 3
Rect 4
C12
R1
L2
C1
R2
Im
Rc
Lm
C2
. 1.

: R1, R2
, RC
; L1, L2
, Lm
; C1, C2
, C12 .

.

C2
,
,
(. 1).

,

.
TrTr2
TrTr1
LLTt
CCT
t
T1
LLTt
TrTr3
CCT
T4
CCT
C4
LLTt
C1
a
Ud
LLTt
TrTr4
CCT
T2
C2
LS
CS
uab
C3
T3
. 2. DC-DC
LT

LS
.
CT
- 29 -
. ,

.

fS, -,
(LS CS).
4.

(. 2).
,
(
) ,
,
Ud
.
() .

,

.

CS.

.

(. 3).
UCE
T1
L
Ud
CP
T2
CS
UCS
T4
T3
CE
. 3.

L
.
LS
4.LT.
CP = CT / 4
.
,
L
CE:
CE
C S .C P
CS CP
a
a 1
CS
(1)
a

CS:
(2)
a CP CS

(. 3) :
0
1
L .C E
a 1
L .C S
(3)
-
,
S 0 2 f S 0
- .

,
,

. , [3]
DC-DC
,

, :
1 cos
U CEm
cos
2 .
(4)
2 .
,
UCSm CS
UCEm
a:
U CSm
a
a 1
U CEm
(5)

,
CS
,
.

a.

UCSm. (2)
CP.
4.

DC-DC ,

L = 432,6 H,
- 30 -
CS = 34,54 nF.

.

Ud
UCSm
CS.
(1)(5)
UCSm.

a
CT.
1.
1
CP
Ud, [V]
102,0
203,0
300,0
404,0
fs, [kHz]
59,630
59,628
59,624
59,615
UCS_M, [V]
324,0
628,0
915,0
1220,0
UCS_C, [V]
324,1
627,1
914,9
1219,9
1,608
1,638
1,653
1,666
CP, [nF]
55,54
56,58
57,09
57,54
CT, [nF]
222,16
226,32
228,36
230,16
: UCS_M UCS_C

UCSm CS;

,
Ud, a CT
. -

,
. ,
- Ud.
3.

.
DC-DC ,

CO2 .

.

,
DC-DC .
,
.
1. , ., . , . .

CO2 .
12,
2012, , I, 73-78
2. , ., . , . .
LCC
.
2012, 2012,
, 176-181.
3. Al Haddad, K., Y. Cheron, H. Foch, V.
Rajagopalan. Static and Dynamic Analysis of a
Series-Resonant Converter Operating above its
Resonant Frequency. PCI Proceedings, Oct 1986,
55-68.
4. Batarseh I. Resonant Converter Topologies with
Three and Four Energy Storage Elements. IEEE
Transactions on Power Electronics, Vol. 9, No. 1,
January 1994, p 64-73.
5. Biela, J., J. Kolar. Using Transformer Parasitics
for Resonant Converters A Review of the
Calculation of the Stray Capacitance of
Transformers in Resonant Converters. IEEE
Transactions on Industry Applications, Vol. 44, Iss.
1, Jan-Feb 2008, p 223-233.
6. Liu, J., L. Sheng, J. Shi, Z. Zhang, X. He.
Design of High Voltage, High Power and High
Frequency Transformer in LCC Resonant
Converter. APEC 2009, Twenty-Fourth Annual
IEEE, Feb 2009, 1034-1038.
7. McLyman, C. Transformer and Inductor Design
Handbook. Marcel Dekker Inc., New York and
Basel, 2004.
Department of Electrical Engineering and
Electronics, Technical Faculty
University of Food Technologies Plovdiv
26 Maritza Blvd.
4002 Plovdiv
BULGARIA
E-mail: avuchev@yahoo.com
E-mail: yasen.madankov@gmail.com
E-mail: nikolay_bankov@yahoo.com
15.02.2013 .
- 31 -

TECHSYS 2013
BULGARIA

,
QUARTUS II

:
CPLD (Complex
Programmable Logic Devices) FPGA (Field Programmable Gate Array).
,
Quartus II. ,
VHDL (Very High Speed Integrated Circuits Description Language)
(Predicate Logic Processor),
.
.
: ,
DIGITAL DESIGN OPTIMIZATIONS BASED ON QUARTUS II

ADVISORS
ATANAS KOSTADINOV
Abstract: There are analyzed the methods for digital design optimizations applying CPLD
(Complex Programmable Logic Devices) FPGA (Field Programmable Gate Array). Special
attention is paid on Quartus II optimizing programs called advisors. An example is shown
using PLP (Predicate Logic Processor) VHDL (Very High Speed Integrated Circuits
Description Language) description together with Quartus II advisors. The received results are
put in the table.
Key words: digital design, Quartus II advisors
1. Introduction
Optimization techniques employed both at
the design phase as well as the implementation
phase play a significant role in realizing efficient
hardware for computation intensive algorithms. For
time critical applications, timing performance is the
most important requirement as compared to area
and cost. On the other hand, certain applications
require the design to be as compact as possible.
Similarly, portable computing platforms and
battery-powered applications require the design to

be power and energy efficient [11].
Achieving high-speed implementation is an
important consideration in computation intensive
algorithms. Depending on the application and its
requirement, speed of an FPGA design is evaluated
based on the following factors: maximum clock
frequency, throughput and latency (delay).
In FPGA-based design, the maximum clock
frequency specifies how fast the circuit can operate.
It is determined by the maximum delay between any
- 32 -
two sequential elements. Throughput is another

important factor for measuring the timing
performance of FPGA-based designs. Throughput is
defined as the number of data processed per clock
cycle and is measured in bits per second. Latency,
on the other hand, is defined as the time between
data input and processed data output. It is measured
in number of clock cycles.
In FPGA-based design, the hardware area
utilized is provided in terms of look up tables
(LUTs) or configurable logic block (CLB) slices.
However, for comparison of designs based on
similar FPGA devices, all the resources must be
considered. Some of the most commonly used
FPGA resources are:
Number of 4-input LUTs
Number of Slices
Number of Slice Flip Flops
Number of IOBs
A design utilizing dedicated resources of
modern FPGA such as embedded multiplier or DSP
blocks will consume less logic resources (LUTs and
CLB slices) as compared to design that implements
the functionality without using dedicated resources.
Power has become a significant design
constraint due to the demand of battery-powered
devices in the rapid growth of personal wireless
communications and other portable digital
applications. Compared to application-specific
integrated circuits (ASICs), FPGAs are generally
perceived as power inefficient because they use a
large number of transistors to provide
programmability. As multimillion gate FPGAs have
become a reality, its applications in power critical
devices are further limited due to their excessive
energy consumption. Therefore, it is important to
use design techniques that results in power efficient
FPGA designs [9, 11].
The Quartus II design software now
includes optimization advisors, which are tools that
provide you with design-specific advice for
optimizing timing performance and resource usage.
The optimization advisors include recommendations
based on your current project settings and
assignments. The Compilation Time Advisor,
Incremental
Compilation
Advisor,
Timing
Optimization Advisor, Resource Optimization
Advisor, and Power Optimization Advisor use
current project information and settings to make
recommendations of project settings and
assignments, individual entity assignments, and
design changes for partitioning a design or
optimizing a project for power, resource usage, or
timing.
The recommendations are organized into
general recommendations and more specific
recommendations for each category. Where
applicable, the categories are organized into a series

of stages, presented by complexity. Each
Recommendation provides a Description, Summary,
and Action.
Description - Describes the reason for the
recommendation and information on the settings
and assignments for the recommendation. The
description may also include links to the appropriate
topic in Quartus II Help or chapter in the Quartus II
Handbook.
Summary - Describes the net effect of the
recommendation on the fMAX, logic element
usage, compilation time, or power consumption for
the design. The effects of the recommendation for
the design are indicated by the following symbols:
( + ) - The recommendation has a net
positive effect on the design.
( - ) - The recommendation has a net
negative effect on the design.
( = ) - The recommendation does not have
an effect on the design.
Action - Describes the options for the
recommendation, the specific steps necessary to
implement it, and the current status of the settings
or assignments for the recommendation. If
applicable, the action also links to the appropriate
Quartus II dialog box, page in the Settings dialog
box, or category in the Assignment Editor [4, 7].
The Resource Optimization Advisor
provides guidance in determining settings that
optimize
resource
usage.
The
Resource
Optimization Advisor provides step-by-step advice
about how to optimize resource usage (logic
element, memory block, DSP block, I/O, and
routing) of your design. Some of the
recommendations in these categories might conflict
with each other. Altera recommends evaluating the
options and choosing the settings that best suit your
requirements [5].
The Timing Optimization Advisor guides
you in making settings that optimize your design to
meet your timing requirements. When you open the
Timing Optimization Advisor after compilation,
you can find recommendations to improve the
timing performance of your design. Some of the
recommendations in these advisors can contradict
each other. Altera recommends evaluating these
options and choosing the settings that best suit the
given requirements.
The Power Optimization Advisor provides
guidance for reducing power consumption. In
addition, the Incremental Compilation Advisor
provides suggestions to improve your results when
partitioning your design for a hierarchical or teambased design flow using the Quartus II incremental
compilation feature [6].
- 33 -
2. Predicate Logic Processor (PLP)

The predicate logic processor (PLP) has
been designed and verified during the postdoctoral
fellowship in NTNU (Norwegian University of
Science and Technology) [3, 8]. The simplified
processor architecture is shown in Fig. 1.
Control Unit
Datapath
Control Logic and

State Register
ACCA
PLU
IR
PC
Program Memory
Data Memory
Fig. 1. Simplified PLP architecture

The proposed PLP belongs to the
application-specific instruction set (ASIP) type of
processors optimized for the solution of artificial
intelligence problems, and it consists of a datapath
and control unit.
The
datapath
is
responsible
for
manipulation of data. It has a register named
accumulator and is denoted as ACCA in Fig. 1.
Another part is a predicate-logic unit (PLU) which
performs the predicate logic operations AND, OR,
and NOT. Additionally, the RAM module (Program
Memory and Data Memory in Fig. 1) is used to
keep the program code (instructions) and
corresponding data.
The control unit of the microprocessor is
realized as a finite-state machine (FSM). By
stepping through a sequence of states, the control
unit controls the operations of the datapath. For
each state that the control unit is in, the control unit
output logic generates the appropriate control signal
for the datapath to perform one data operation [2]
The control unit consists of logic circuit and
state register together with the instruction register
(IR) and program counter (PC). IR contains the
code of the currently executed operation, and the
PC is used to access the memory cells in RAM.
The design of the PLP starts with the
datapath synthesis. Then, simulation of the
proposed solution is performed to check the
correctness of the unit design. After these steps, the
synthesis of the control unit is realized, and the
datapath and control unit are connected together.
They also have a connection with the RAM module
containing the test program, which consists of all

implemented instructions for the predicate
processor.
The synthesis of the whole circuit is
performed with a DE2 FPGA board using
Quartus II both manufactured by Altera
Corporation. To verify the design, a control
program is coded and loaded into RAM. The first
part of PLP debugging is that all introduced
predicate logical instructions are verified one by
one using the SignalTap II Embedded Logic
Analyzer which belongs to the Quartus II program
[1, 10, 12]. The last stage of PLP verification is
execution of some limited number of testing
programs. The data captured by the SignalTap II
Embedded Logic Analyzer is compared with the
one based on theoretical calculations.
3. Results
It has been synthesized the three versions of
PLP using Quartus II 11.1sp2 Web edition
program and applying different types of advisors.
The results are presented in Table 1.
Results after synthesis
Total
Total
Dedicated
logic combinational logic
elements functions
registers
toa
171
138
111
roa
154
126
106
poa
161
134
108
Table 1
Fmax,
MHz
260,96
177,27
246,91
The abbreviation toa stands for realized PLP

architecture using Timing Optimization Advisor,
roa - using Resource Optimization Advisor and poa
- using Power Optimization Advisor.
It is shown that toa processor architecture
consists of the largest number of total logic
elements, total combinational functions and
dedicated logic registers. It is achieved the maximal
clock frequency.
The second realized PLP architecture using
Resource Optimization Advisor has employed the
smallest number of hardware resources but it is
achieved the lowest clock frequency.
Applying Power Optimization Advisor, it
leads to the third PLP architecture which is
something between toa and roa processors.
4. Conclusions
Three different processor architectures has
been realized using Quartus II optimization
advisors in easy way.
Quartus II optimization advisors are
applicable both in the research as well in the
teaching process.
- 34 -
It is suitable to be presented to the students

and to be used in VLSI (Very-Large-Scale
Integration) design laboratory exercises.
References
1. Corno F., G. Cumani, M. Reorda, G. Squillero
Evolutionary test
program induction for
microprocessor design verification, Proceedings of
IEEE Asian Test Symposium, Guam, USA,
November 18-20, 2002, 368-373.
2. Hwang E. Digital Logic and Microprocessor
Design with VHDL, Thomson, Riverside, 2006.
3. https://fellowship.ercim.eu/documents/old_report
/038_Kostadinov_FP_2006_2007_SR_NTNU.pdf
4. http://quartushelp.altera.com/10.1/mergedProject
s /report/oaw/oaw_view_using_oaw.htm
5. http://www.altera.com/literature/hb/qts
/qts_qii52005.pdf
6. http://www.altera.com/literature/hb/qts
/qts_qii52016.pdf
7. http://www.altera.com/support/kdb/optimization/
oa-index.html
8. Kouzaev G., A. Kostadinov Predicate Gates,
Components and a Processor for Spatial Logic,
Journal of Circuits, Systems, and Computers,
Vol. 19, No. 7, May 2010, 1517-1541.
9. Li F., D. Chen, L. He, J. Cong Architecture
evaluation for powerefficient FPGAs, Proceedings
of the ACM/SIGDA International. Symposium on
Field Programmable Gate Arrays, Monterey, USA,
February 23-25, 2003, 175-184.
10. Pastor J., I. Gonzalez, J. Lopez,
F. Gomez-Arribas, J. Martinez A remote
laboratory
for
debugging
FPGA-based
microprocessor prototypes, Proceedings of the 4th
IEEE International Conference on Advance
Learning Technology, Joensuu, Finland, August 30
- September 1, 2004, 86-90.
11. Qasim S., S. Abbasi, B. Almashary A review
of FPGA-based design methodology and
optimization techniques for efficient hardware
realization of computation intensive algorithms,
Proceedings of Multimedia, Signal Processing and
Communication Technologies Conference, Aligarh,
India, March 14-16, 2009, 313-316.
12. Von Kaenel P. Designing and testing a control
unit, Journal of Computing Science in Colleges,
Vol. 19, Issue 5, May 2004, 228-237.
Department of Computer Systems and

Technologies
25 Tsanko Dystabanov St.
4000 Plovdiv
BULGARIA
E-mail: kostadat@tu-plovdiv.bg
15.02.2013 .
- 35 -

TECHSYS 2013
BULGARIA

-, ,
:

. MPI
.
10 000 000
(2,65 16 )
(1,42 16 ).
: , , MPI, Blue gene/P
FAST PARALLEL SORTING BASED ON QUICK

SORT
ATANASKA BOSAKOVA-ARDENSKA, NAYDEN VASILEV, IVAN FILLIPOV
Abstract: This paper presents a parallel sorting algorithm which uses dividing of given row to
sub-rows. The sub-rows do not overlap one another. The algorithm is implemented in C with
using of MPI functions. For the experiments is used Bulgarian supercomputer Blue Gene/P. The
results show that increase the number of parallel processors lead to acceleration of time for
processing. When are sorted 10 000 000 integers the parallel sorting algorithm which uses subrows is 2,65 times faster than sequential quick sort and 1,42 times faster than classic parallel
quick sort.
Key words: parallel algorithm, quick sort, MPI, Blue Gene/P
1.

.
, 25% ,
[3].
,

.
[1, 2, 3, 6].

[1]:
-
(
,
.);
- ()
( ,
, , ,
.).
-
,
, ..
.

- T=f(n).
[1, 2, 3], -
..
,
. (n)
- 36 -
-
f(n).

O(n2).

O(n lg n).
2.

.
,
.

.
:
1)
;
2)
( i),
-
.
( i
);
3)
( j),
-
.
( j
);
4) 2)
3).
5)
i >= j;
6) 1) 5)
;
7) 1) 5)
;
8)

.

. 1
,
.
(.1 ()) ,
,
(
) .
1()
( ),

.
1() -
,

.
33 55 77 11 66 88 22 44
11 22 33 44 55 66 77 88
33 11 22
11 22 33 44 55 66 77
11
55 77 66 88
33
55 77 66
11 22 33 44 55 66
55
11 22 33 44 55
77
11 22 33 44 55 66 77 88
11 22 33 44
11 22 33
11 22 33
11
33
55 66 77 88
55 66 77
11 22
55
11
77
()
()
. 1.
3.

[7]
.

.
(
,
, - -
).
, ,
,
..

(.2).
P0
P0
P0
P0
P1
P2
P4
P1
P3
P3
P5
. 2.

- 37 -
[7]

hypercube [5, 7].

,
(
).
[4]
Parallel Two-Dimensional
Quicksort (PTSA).
,

.
- -
. ,

-
.

. 3

33.
, ,
, :
- ()
;
- ;
- .
aj, j=1,2,,n.
aj A={a1, a2,, an}. :
A1={
},
A2={
},,
Ak={
};
A t = 1,2,k
At
a:

() ,

.
.
k
.

1
k ( ). 4
.
P0 -

1
P1
P2
PK
. 3. PTSA
4.

,
,
.
()
.
().
. 4.
5.

- 38 -
MPI.

. Wilkinson . Allen
(. .2) [11]. 5

.
MPI
myrank == MASTER

,
]max/np[ - . (
, max.)
np ,
]...[ , ,
max np .
40 3
13.

.

,

.

.

Blue Gene/P [9, 10].
VN,

. 1, 2 3
1 000 000

4, 8 16 .
,

.
1
4 1 000 000
quicksort
quicksort
2,256913
2,258248
3
4
slaves
master
MPI_Finalize()
. 5.

Master/Slave

,
(master) ,
(slaves).

, .

(max). (
.)
0,644886
0,192773
0,411344
0,6041
0,645611
0,181269
0,402716
0,5839
2,252437
0,642398
0,198139
0,414439
0,6125
2,253359
0,641176
0,192317
0,418164
0,6104
2,253014
0,642576
0,193370
0,419722
0,6130
2,258851
0,644491
0,196138
0,414815
0,6109
2,257624
0,641958
0,192576
0,413977
0,6065
2,252398
0,647423
0,199921
0,415126
0,6150
2,256617
0,646600
0,192054
0,419746
0,6118
10 2,257512
0,643716
0,196137
0,410519
0,6066
.
- 2,2556973
0,6440835
0,193469
0,414057
0,6075
4, 5 6
10 000 000
.
6 ,

, 7
.
- 39 -
2
8 1 000 000
quicksort
quicksort
5
8 10 000 000
quicksort
quicksort
2,256913
0,451764
0,156674
0,278319
0,4349
1 68,260787
47,57321
11,28697
26,64193
37,928
2,258248
0,452897
0,146349
0,266716
0,4130
2 68,264618
47,12694
11,76145
26,97164
38,733
2,252437
0,453914
0,151838
0,275311
0,4271
3 68,262374
47,55183
11,57316
26,56713
38,140
2,253359
0,455561
0,151546
0,273164
0,4247
4 68,264415
47,97164
11,77195
26,69741
38,469
2,253014
0,459632
0,157921
0,272267
0,4301
5 68,261796
47,27164
11,82272
26,77617
38,598
2,258851
0,456712
0,155561
0,271389
0,4269
6 68,266512
47,32169
11,64155
26,69138
38,332
2,257624
0,456348
0,156973
0,277961
0,4349
7 68,269713
47,37165
11,36172
26,69447
38,056
2,252398
0,455514
0,152367
0,273584
0,4259
8 68,265476
47,69132
11,16797
26,23648
37,404
2,256617
0,458237
0,158126
0,274691
0,4328
9 68,264432
47,91864
11,71333
26,67144
38,384
10 2,257512
0,459456
0,156614
0,272315
0,4289
10 68,267612
47,77114
11,85167
26,66971
38,521
0,4279
.
- 68,26477
47,503827
11,59525
26,66177
38,257
.
- 2,2556973
0,4560035
0,154397
0,273572
3
16 1 000 000
quicksort
quicksort
2,256913
2,258248
6
16 10 000 000
quicksort
0,2973371
0,101564
0,168319
0,2698
1 68,260787
0,2913642
0,106674
0,157164
0,2638
2 68,264618
2,252437
0,2999183
0,106674
0,161389
0,2680
2,253359
0,2926472
0,106674
0,165569
2,253014
0,2956123
0,106674
0,162711
2,258851
0,2966479
0,106674
2,257624
0,2927156
0,106674
2,252398
0,2944138
2,256617
10 2,257512
36,149955
9,130397
16,66987
25,800
36,649214
9,130473
16,47012
25,600
3 68,262374
36,479832
9,138716
16,06974
25,208
0,2722
4 68,264415
36,914673
9,339871
16,44698
25,786
0,2693
5 68,261796
36,647892
9,674136
16,10397
25,778
0,163916
0,2705
6 68,266512
36,237496
9,449873
16,55479
26,004
0,164561
0,2712
7 68,269713
36,149873
9,564129
16,01469
25,578
0,106674
0,162254
0,2689
8 68,265476
36,971643
9,364812
16,69713
26,061
0,2926941
0,106674
0,161769
0,2684
9 68,264432
36,274936
9,664779
16,03975
25,704
0,2918234
0,106674
0,163715
0,2703
10 68,267612
36,913684
9,446985
16,66470
26,111
0,2692
.
- 68,26477 36,5389198
9,390417
16,373174
25,763
0,106163
0,1631367
4
4 10 000 000

59,136413
15,08641
32,09554
47,181
2 68,264618
59,130397
14,37512
31,59126
45,966
3 68,262374
59,131576
15,07641
32,09812
47,174
4 68,264415
59,132283
15,07217
32,09541
47,167
5 68,261796
59,136419
15,07552
32,09341
47,168
6 68,266512
59,135917
15,07731
32,09268
47,170
7 68,269713
59,139932
15,07629
32,09238
47,168
8 68,265476
59,136488
15,07227
32,09537
47,167
9 68,264432
59,135427
15,07671
32,09641
47,173
10 68,267612
59,136118
15,07554
32,09367
47,169
15,07638
32,04442
47,05
59,135097
2,5
quicksort
1 68,260787
.
- 68,26477
.
- 2,2556973 0,29451739
quicksort
quicksort
1,5
0,5
0
4
16
. 6.

- 40 -
2. Cormen T., Ch. Leiserson, R. Rivest, C. Stein,

Introduction to Algorithms, Second Edition, The
MIT Press and McGraw-Hill, 2001.
1,45
1,4
3. Knuth D., The art of computer programming,

V3. Sorting and Searching, Addison Wesley
Publishing Company, 1973.
1,35
1,3
1,25
4. Rjoub H. Al, A. Odat, A. Audat, Parallel TwoDimensional Quicksort Algorithm (PTSA), Journal
of Computer Science, vol. 4, no. 1, 2008.
1,2
1,15
4
16
. 7.

6.

.

MPI.

1 000 000 10 000 000 . ,

. ( 4
10 000 000
1,45, 16 2,65.)

, [5]. ( 4
10 000 000
1,25, 16
1,42.)

.

PTSA,

[4].

.
1. ., ,
, : , 2005.
5. Roosta S. H., Parallel Processing and Parallel

Algorithms- Theory and Computation, 1999.
6. Sedgewick R., Algorithms in C, AddisonWesley, 1990.
7. Wilkinson B. and Allen M., Sorting Algorithms,
Parallel Programming: Techniques and Applications
Using Networked Workstations and Parallel
Computers, Prentice-Hall, 1999.
8. http://www.mpi-forum.org/
9. http://www.redbooks.ibm.com/abstracts/sg2472
87.html
10. http://www.scc.acad.bg/ncsa/index.php/bg/-blue-gene-p
11. http://www.macs.hw.ac.uk/
technologies
University of Food Technologies, Plovdiv
26 Maritza Blvd.
4000 Plovdiv
BULGARIA
E-mail: a_bosakova@uft-plovdiv.bg
technologies
4000 Plovdiv
BULGARIA
E-mail: mnvasilev@yahoo.com
E-mail: i.fillipoff@gmail.com
13.02.2013 .
- 41 -

TECHSYS 2013
BULGARIA

: a
(FreeSpace
Laser Communications FreeSpace Optics FSO)
a
. . a

FSO.
: FreeSpace Optics, BitError Rate,
INFLUENCE OF THE ATMOSPHERIC

TURBULENCE IN POWER DESIGN OF
GROUND-TO-GROUND FREESPACE LASER
COMMUNICATION SYSTEMS
BORYANA PACHEDJIEVA
Abstract: An modification of algorithm for power design of the ground-to-ground FreeSpace
Laser Communication Systems (FreeSpace Optics (FSO)) based on the random laser beam
shift, caused by random character of the Atmospheric turbulent heterogeneities has been
suggested. he definition area of the solution has been estimated. An analysis of the Influence
of the Atmospheric turbulence on power parameters of the FSO has been made.
Key words: FreeSpace Optics, BitError Rate, atmospheric turbulence
1.

(Free Space Laser
Communications, Free Space Optics FSO)

[4,5,6,8,9,11,13].
, -
FSO
,
.
, -
,
- 42 -
.
,
FSO [4,5,6,7,8,9,10,11,12,13].
FSO, ,
.

,
,
,

.

,
FSO.

[1]

FSO
-
,
,
.
2.
FSO (.1),
-
L, TA
1,
r0, RA
R2
2,
IF
PhD.

,

,
.
, Bit
Error Rate (BER)
,
L,
r0 R2.
,
:
f ;
;
1; r0 ;
H;
;
Z.
,
:
SM,

Cn2 ,
.
,
:
.

iS Z ,
BER.
BER [3]
/
Q erfcinv2.BER
(1)
,
-
[2] /
Q( Z )
. 1.

,
[1]
,

,
.
2j
iS ( Z )
2 2 i2S ( Z ) 2jS ( Z )
(2)
, i2S
,
- 43 -
iS Z
16.e.f .Q 2
,
1 BZ .Q 2
(3)
e 1,6.10 19 C
BZ
8. 2 Z
(4)
K ,

,

.

is Z

[3]
Z exp .Z ,km
(5)
-
S i ,

[3]
3,91 , m
,km
S M ,km 0 ,55
1
0 ,585 S M ,km 1 3
. (6)
Z , ,

,
1
(7)
Z
4. x2 Z
1 2
r Z
1
2
(8)
2 Z
Z ,
2
8. x Z
1 2
r Z
x ,
.
x,

[3]
Z C .Z r
2
n
3 1 / 3
0
(9)
(7) (8) r Z

[1]
K . 2
.Z 2 ,
r Z r 1
2
.r0
2
0
Si
e. .
,
h.c
(12)

Z 2 Z
2
x
2. 1 . 2 .Si .R22 . L
. Z . Z , (11)
.r 2 Z
(10)
1 exp 2 0,865 .
(3) (11)
L .R2 2 . 2
2
(13)
8. .e.f .erfcinv( 2.BER ) .r 2 Z
1 B.erfcinv( 2.BER ) . .S . Z . Z
2

.r 2
, .. Z 0 ,
K .
(12)
r 2 Z
K 2 .2 .Z 2
.
2 .r02
(14)
(14)
(13),
L .R2 2 . 2
8. .e.K 2 .2 .Z 2 . f .erfcinv2.BER
(15)
1 B.erfcinv2.BER .r . 1 .Si . Z . Z
2
2
0
(4)
(15)

a
FSO
.
2.2.
FSO

[1]
- 44 -
1 B.erfcinv2.BER >0
(16)
BZ .erfcinv2.BERmin 1 ,
(17)
K 10 ;
Z 6 km ;
f 1GHz ;
0 ,7 ;
2
n
SM 10 km ; C var ; r0 var
BERmin

,
(15).
(4) (17)
BERmin
BERmin Z
Z
1
erfc
8. 2 Z
2
(18)
(18) , BERmin

. ,
,

,

,
,
BER BERmin .
(19)
3.
3.1.

FSO

,

r Z ,
r0 .

FSO
,

e.

r0

2
: 1,55 m ; 1 0,7 ;
. 2. BERmin
r0

C n2 , SM 10 km.
. 2

()

FSO

,
.

Cn2 ,

-
FSO.
-
( - ),

,
-
.
3.2.

FSO
(4) (19),

- 45 -
,

L Z
BERmin Z
FSO : 1,55 m ; r0 3 mm ;
K 10 ;
Z var ;
SM 10 km;
1 0,7 ;
Cn2 var ; f 1GHz ;

R2 10 cm ; 2 0,6 .
0 ,7 ;
. ,

L
Z.
BER 10 6 ;
. 3 . 4
BERmin Z
L Z Cn2 var .
. 4.
L Z Cn2 .
. 3.
BERmin( Z ) Cn2 .
. 3
BERmin Z

C n2 . ,
BERmin( Z )
( Cn2 1012 m2 / 3 ).
,
Cn2 ,
,
10 -6 ,
6 km.
L Z ,
. 4
.
BER 10 6

L 10 100 mW
15 20 km
1. Bonev, B., B. Pachedjieva, E. Ferdinandov.

Influence of the Atmospheric transparency
fluctuation on solving the reverse task of the
analysis and power design of ground-to-ground
FreeSpace Laser communication systems. Journal
of the Technical University Sofia, branch Plovdiv,
Fundamental Sciences and Applications, Vol. 18,
2012, pp. 7-12, ISSN 1310-82712.
2. Bonev B., Relative Influence of Some
Stochastic Factors on Bit-Error Rate of Groundto-Ground Free Space Optics, XLII International
Scientific
Conference
on
Information,
Communication and Energy Systems and
Technologies, ICEST 2007, Vol. 1, pp. 203
206, June 2007, Ohrid, Macedonia.
3. Ferdinandov E., B. Pachedjieva, B. Bonev,
Sl. Saparev, Jointly Influence of Heterogeneous
Stochastic Factors on Bit-Error Rate in groundto-ground FreeSpace Laser Communication
Systems, Optics Communications 207, p.121127, 2007.
4. Hecht, J. Laser Focus World 11 2001.
5. Hranilovic, S., Wireless Optical Communication
Systems, Springer Science, Boston, 2005.
6. Killinger, D. Optics and Photonics News 13 (10)
(2002) 36-42.
- 46 -
7. Shlomi, A., Effects of Atmospheric Turbulence

and Building Sway on Optical Wireless
Communication System, Optics Letters, Vol.28,
2, 129131, 2003.
8. Korevaar, E.J. (Ed.), Optical Wireless
Communications III, SPIE, vol. 4214, 2000.
9. Mecherle, G.S. (Ed.), Free-Space Laser
Communication Technologies XII, SPIE, vol. 3932,
2000
10. Mitsev Ts., N. Kolev, Hr. Ivanov, K.
Dimitrov, Optimum Divergence of the Transmitter
Optical Radiation in FSO Systems, XLVII Intern.
Scientific Conf. on Inform., Communication and
Energy Systems and Technol. (ICEST 2012), June
28 to 30, Veliko Tarnovo, Bulgaria, 2012.
11. Willebrand, H., B. Ghuman, Free-Space
Optics: Enabling Optical Connectivity in Todays
Networks, SAMS Publ., Indianapolis, 2002.
12. Xiaoming, Z., J.M. Kahn, FreeSpace Optical
Communication Through Atmospheric turbulent
Channels, IEEE Trans. On Commun., Vol.50, 8,
12931300, 2002.
13. Zaatari, M.O. The
Review (2003) 49-56..
Telecommunications
Department of Optoelectronics and Laser

Engineering
25 Tsanko Diustabanov St.
4000 Plovdiv
BULGARIA
E-mail: pachedjieva@yahoo.com
26.02.2013 .
- 47 -

TECHSYS 2013
BULGARIA

,
:
.

.
,
.
: , ,
AN APPROACH FOR PROCEDURAL TEXTURES

GENERATION
DANIELA ILIEVA, PAVLINA VLADIMIROVA
Abstract: In this paper is proposed an algorithm for creating procedural texture images based
on Perlin noise. The generation of texture images is a process of creating digital images with
necessary size from a given noise functions. The new generated texture has a great quality and
desired size and can be applied in Computer Graphics, Computer Vision and Computer games.
Key words: procedural texture, Perlin noise, texture generation
1.

,
, , , , .
,
, , ,
.

, , , ..
, , , , ,
, .

.
,
2D 3D

2.

, , .

,

. ,

, .

- 48 -
-
. -
, , ,

.

[1, 2, 3].
3.

[1].

.
2D:
Persistence
= 1/2
Amplitude
Persistenc
e=1/
root2
/2
/4
Amplitude
/8
/ 1.414
/16
/2
/2.82
/32
result
/4
/5.656
result
Persisten
ce = 1
Amplitud
e
result
. 1.

, -
- 2D 3D (.2).
.2.

,
.
frequency = 2i
amplitude = persistencei
i i- ,
.

.

,

. +
Fre
quen
cy
Persist
ence=
1/4
Ampli
tude
+
1
+
1
/4
+
1
/16
16
+
1
/64
32
+
1
/256
=
1
/1024
resul
4.
function Noise1(integer x, integer y)

n = x + y * 57
n = (n<<13) ^ n;
return ( 1.0 - ( (n * (n * n * 15731 + 789221) +
1376312589) & 7fffffff) / 1073741824.0);
end function
function SmoothNoise_1(float x, float y)
corners = ( Noise(x-1, y-1) +Noise(x+1, y-1) +
Noise (x-1, y+1)+Noise(x+1, y+1) ) / 16
sides = ( Noise(x-1, y) +Noise(x+1, y) +Noise(x,
y-1) +Noise(x, y+1) ) / 8
center = Noise(x, y) / 4
return corners + sides + center
end function
function InterpolatedNoise_1(float x, float y)
integer_X = int(x)
fractional_X = x - integer_X
integer_Y = int(y)
- 49 -
fractional_Y = y - integer_Y
v1 = SmoothedNoise1(integer_X, integer_Y)
v2 = SmoothedNoise1(integer_X + 1, integer_Y)
v3 = SmoothedNoise1(integer_X,
integer_Y +
1)
v4 = SmoothedNoise1(integer_X + 1, integer_Y +
1)
i1 = Interpolate(v1 , v2 , fractional_X)
i2 = Interpolate(v3 , v4 , fractional_X)
return Interpolate(i1 , i2 , fractional_Y)
end function
function PerlinNoise_2D(float x, float y)
total = 0
p = persistence
n = Number_Of_Octaves - 1
loop i from 0 to n
frequency = 2i
amplitude = pi
total = total + InterpolatedNoisei(x * frequency, y
* frequency) * amplitude
end of i loop
return total
end function
3. 2D
: 2D .
(
subdivision),
,
.
, ,
.

.
.
:
.
:

.
++,
.
, - ,
.
Base frequency=5
Octaves=10
Function: x
Base frequency=60
Octaves=5
Function: Misc
5.
Base frequency=50
Octaves=5
Function: Misc
- 50 -
Base frequency=5
Octaves=6
Function: 1/x
Base frequency=5
Octaves=6
Function: 1/cos(x)
,
.
.

(
).

,
.

.

.

:
.

.

, .
1. D. Ebert, K. Musgrave, D. Peachey, K. Perlin,
S. Worley Texturing and modeling. A procedural

approach. Third edition. Morgan Kaufman
Publishers. 2003. ISBN: 1558608486
2. P. Schneider, D. Eberly "Geometric Tools for

Computer Graphics". Third edition. Morgan
Kaufman Publishers. 2002.
3. R. Goldman Pyramid Algorithms: A Dynamic
Programming Approach to Curves and Surfaces for
Geometric Modeling Second edition. Morgan
Kaufman Publishers. 2007.
Base frequency=5
Octaves=10
Function: Wood
5.

:
Department of Computer Sciences

Technical UniversityVarna
1 Studentska St.
9009 Varna
BULGARIA
E-mail: ilievadaniela@mail.bg
E-mail: pav_varna@yahoo.com
28.02.2013 .
- 51 -

TECHSYS 2013
BULGARIA

,
:

.

.
.
: , ,
.
QUALITY ASSESSMENT OF WELDED CONTACT

CONNECTIONS BY THE METHOD OF EDDY
CURRENTS
DIAN MALAMOV, TSVETOMIR STOYANOV
Abstract: The paper investigates the possibility to use the method of eddy currents for quality
assessment of welded contact connections in a miniature automatic circuit-breaker.
Correlational dependence between the power of disconnecting the contact connection and the
output signal of the device for eddy current control has been experimentally established. The
method of eddy currents is suggested for putting into practice of production.
Key words: contact connection, electric-contact welding, eddy-current control
1. .

,

.

( ),
(, , .).

,
,
.
,
-

, -
.

,
:
-
;
- 52 -
-
;
-
.

,
:
-
;
-
;
-
;
-
.

:
- :

;
-
:
,
,
,
,
.

,
. 1.
2. .
[1], [2]

.2. (1)
(3),
(4) (2).
(5),
.

,

(6) .
. 2.
.

,
, :
Z 0 R0 jX 0 ,
(1)
Z0
:
; R0
; X 0
.
Z B RB jX B
. 1.
() (a),
(b)
.
(2)

.

( ).
- 53 -

, ,
, .
[1] j
:
j
(3)
e x f ,
j0
j0

ELOTEST 3 [3].

. 1,5 mm.

.

.
; f
;

;
.

e .
x0
1
f
(4)

()
.

:
RB R0
,
X0
X
Rcn B .
X0
Rcn
. 3.
ELOTEST 3
(5)
(6)

:
-
;
-

,
, ;
-
.
. 4.
a),
b).
. 5.

.
- 54 -
"-"
71%
Mean = 9,3633; StdDv = 0,1202; SW-W = 0,7531;

p = 0,0000001;D = 0,3751; p < 0,0100; Lilliefors-p < 0,01
61%
51%
41%
31%
20%
10%
0%
9,00
9,15
9,30
9,45
9,60
" Y"
. 6.

.
3. .

:
-

;
-

.

,
.
STATISTICA
7, . 7 . 8.
"-"
Mean = 9,054; StdDv = 0,1388; D = 0,3514; p < 0,0100; Lilliefors-p < 0,01;
SW-W = 0,7916; p = 0,0000006.
70%
60%
50%
40%
30%
20%
10%
0%
8,80
8,95
9,10
9,25
9,40
" Y"
. 7.

() .
. 8.

() .

.

(
,
). ,

.
:
-
;
-
.

STATISTICA 7.

.9 .10.

,

.
- 55 -
4. .

" Y"
"-"
10,0
9,5
9,0
8,5
8,0
7,5
7,0
6,5
- F, dN
. 9.

()

" Y"
"-"
11
10
1.

.

.
2.
,

0,997,
4% ,
5% .
3.

.

.
1. . . , . .,

, ., ,
1978, 216 .
10
12
14
16
- F, dN
. 10.

()
.

,
.
:
- - :
Y 6,93 O,146F 0,046F 2 ;
(7)
- - :
Y 6,05 O,02F 0,024F 2
(8)
: Y
; F
dN.
STATISTICA.
3. Eddy Current Instruments and Systems,

Rohmann GmbH, Germany (www.rohmann.de).
2. Javier Garca-Martn,
Jaime Gmez-Gil,
Ernesto
Vzquez-Snchez,
Non-Destructive
Techniques Based on Eddy Current Testing,
Sensors 2011, 11, pp 2525-2565.
Department of Electrical Engineering

4000 Plovdiv
BULGARIA
E-mail: deanmalamov@abv.bg
Schneider Electric
Industrialna Zona Plovdiv
4202 Radinovo
-mail: fidel_kastro@abv.bg
03.03.2013 .
- 56 -
- 57 -

TECHSYS 2013
BULGARIA

:
. ANSYS,
.

.

,
.
: , ,
INVESTIGATION OF THE PERMISSIBLE

OVERLOAD OF THE MAIN CIRCUIT IN AN
ELECTROMAGNETIC CONTACTOR
DIAN MALAMOV
Abstract: A computer model is proposed to analyze the thermal field of the main circuit in
transient mode of heating. The model has been developed in ANSYS and the related problems
concerning both the electric and the thermal fields have been solved. The level of heating in
characteristic points from the main circuit at loading with normed short withstand current has
been defined. Graphical and functional dependencies are proposed, concerning the influence of
the contact resistance in the switching contact and the contact area between the contact body
and the contact base over the heating in the main circuit.
Key words: short withstand current, contact resistance, contactor
1. .

,

[1]. ,

[3,5]. [3]
,

.
[7] 2D

.
- 58 -

. [5,6] 3D

.
.
2. 3D .

25
3.
.

.
3D

. 1.
-
.
.
. 1.
25 : 1- ; 2-; 3-; 4 ; 5-
.

. 2.
. 2.
: -

; -
; c-
.

:
DZ
(1)
D
:
DZ

; D
.
kz

.

3D

[2,8].
d 1 dV
dx T dx
d 1 dV

dy T dy
d 1 dV

0,
dz T dz
T 0 1 T ,
(2)
(3)
E grad V
(4)
:
V
; ( T )

-
T;
.

,

.
p T T J 2 T E 2
(5)
: p T - .

:
-
;
-
;
- :
V=0
(5)
-
:
V
(6)
0
n
n e
.
- 59 -

.
2T 2T 2T
T
(7)
2 2 p T c
2
y
z
t
x
:
;
; .

:
-
;
-

[3].
-
:
T
0
n
(8)

(3).
3D

[6],
. 2 (1).
ANSYS

.

.

. . 3

.
. 3.
3D- .
. 4

,
[1].
. 4.
200 A 10 s
0,4 mohm .
. 5

(a) (b)
.
a)
b)
. 5.
200 : akS 0,4 b - kS 1 .
. 6

- 60 -
, :
,
, .
,
.

,
(1),
.

.
. 6.
(1),
(2) (3) 200 A
10 s 0,4 mohm .
3. .

:
1)

[6].
200 A 10 s.
2)

kS 0,1 1 .
3)

[3].
RK 0.1 1 mohm.

0,05 mohm.
4)

[3].

STATISTICA. . 7, .
8 . 9

. 7.
RK
kS .
. 8.
RK
. kS .
- 61 -
. 9.

RK
kS .

STATISTICA

,

.
:
- :
T 61,4 25kS 132RK
12,6kS 2 1,7kS RK 23,4RK 2
(8)

:
T 102,1 60x 402,2 y
30x 2 78xy 76 ,2 y 2
110 C .
( , .),
[1]

. [3]
,

.

. [3]
325 C .
[5]

200 C .

200 C . [3]
,

-
.

250 C .
(8), (9) (10)

RK
kS , . 10.
(9)

:
T 159 130,1x 279 y
62x 2 31,6 xy 21,7 y 2
(10)

[1].
,
. 10.
RK
kS : 1-; 2-; 3 .
- 62 -
4. .
1.

-
,
.
kS 1

RK (0,2 0,8 )mohm ,

: : 87%, 113%
93%.
RK 0,2mohm kS (1 0,1)
: -16%, 14% -14%.
2.
,

RK 0,2 mohm.
15%.
, 51 C .
-
RK 0,2mohm
k S
,
.
3.

,
, kS 0,5
.
1. EN 60947-4-1
.
.
2. ., , , , 1991.
3. . .

,
, 2009 .
4. ., , I ,
2004.
5. .,

,
IV- 2012,
2, , 28 .-1 . 2012 . . (360367).
6. Dian Malamov, Ivan P. Georgiev, Modeling of
the Thermal Field in the Main Circuit of an
Electromagnetic Contactor, Proceedings of 17th
International Symposium of Electrical apparatus
and Technologies SIELA Vol. I, 28-30 May 2012,
Bulgaria, pp.194-200.
7. Herman A. Nied, The Thermostructurate
Respounse of Electric Contacts Having Brasse
Defects, IEEE Trans. on Components, Packaging,
and Manufacturing Technology, Vol. CHMT-8,
1, March 1985, Page(s):187196.
8. Popa Ioan C., Cautil Ioan, Modeling of High
Currents Dismountable Contacts, Proceedings
of 15th International Symposium of Electrical
apparatus and Technologies, SIELA 2007,
Volume I, Page(s): 150-157.
4000 Plovdiv
BULGARIA
E-mail: deanmalamov@abv.bg
03.03.2013 .
- 63 -

TECHSYS 2013
BULGARIA

:
(CPU, HDD, Mother
Board, RAM, Video Card and Mark) -
. , .
, .

.
: , , ,
,
INVESTIGATION OF THE EFFICIENCY OF

GENETIC ALGORITHMS IN OPTIMIZING THE
SEARCH FOR THE BEST QUALITY COMPUTER
CONFIGURATION
DILYANA BUDAKOVA
Abstract: The aim of this paper is to investigate the efficiency of using a genetic algorithm in
optimizing the selection of computer components (CPU, HDD, Mother Board, RAM, Video Card
and Mark) for constructing the best quality computer configuration. A model, appropriate for
simulation of evolution has been proposed. A software system, based on this model, has been
developed. The experiments, conducted with the developed software system, confirm the
efficiency of implementation of genetic algorithms for solving problems of this order.
Key words: genetic algorithms, optimization, search methods, machine learning, computer
configuration
1. Introduction
Genetic algorithms (GAs), introduced by
Holland in 1975 [4], are inspired by natural
evolution and the magnum opus The Origin of
Species, published in 1859 by Charles Darwin.
They are search methods based on the evolutionary
concept of natural mutation and the survival of the

fittest individuals. Given a well-defined search
space they apply three different genetic search
operations, namely, selection, crossover, and
mutation, to transform an initial population of
- 64 -
chromosomes, with the objective to improve their

quality.
Experimental analysis illustrated that the
GAs design constantly outperforms the greedy
method in terms of solution quality.
GAs have been used for problem-solving
and for modeling[9]. GAs are applied to many
scientific, engineering problems, in business and
entertainment, including [2,3,6,9]:
Optimization: numerical optimization,
combinatorial optimization problems such as
traveling salesman problem (TSP).
Machine and robot learning: including
classification and prediction, and protein structure
prediction. GAs have also been used to design
neural networks, to evolve rules for learning
classifier systems or symbolic production systems,
and to design and control robots.
Economic models. Immune system
models. Ecological models: GAs have been used to
model ecological phenomena such as biological
arms races, host-parasite co-evolutions, symbiosis
and resource flow in ecologies.
Models of social systems: GAs have been
used to study evolutionary aspects of social
systems, such as the evolution of cooperation [2],
the evolution of communication, and trail-following
behaviour in ants.
In this paper the efficiency of a genetic
algorithm in optimizing the selection of computer
components (CPU, HDD, Mother Board, RAM,
Video Card and Mark) for constructing the best
quality computer
configuration is under
investigation. A model, suitable for evolution
simulation is proposed. A programming system,
using this model, is developed. The experiments,
conducted with the developed programming system
the efficiency of genetic algorithms in solving
problems of this order.
2. Examples of evolution simulation
In GAs the following specifications are
adhered [7]: A chromosome is a representation in
which: There is a list of elements called genes. The
chromosome determines the overall fitness
manifested by some mechanism that uses the
chromosomes genes as a sort of blueprint.
Create a chromosome from a given list of
elements in this case the constructor might be
called the genesis constructor. The multiplicity of
candidate solutions, processed by the genetic
algorithms at each step is called population. Mutate
one or more genes in one or more of the current
chromosomes, producing one new offspring for
each chromosome mutated. Mate one or more pairs
of chromosomes. Add the mutated and offspring
chromosomes to the current population.
Create a new generation by keeping the best

of the current populations chromosomes, along
with other chromosomes selected randomly from
the current population. Bias the random selection
according to assessed fitness.
Here we have a number of exemplary
models, used for evolution simulation in accordance
with this terminology.
The model, developed for solving the
traveling salesman problem (TSP) with the help of
a genetic algorithm is of special interest. This is a
typical optimization problem, aimed at finding a
Hamilton cycle with minimum length at a given
weighed complete graph G(V,E) with weight of the
edges real (positive) numbers. It is NP-complete
and in solving it by full running of a complete graph
with n vertices, n! Hamilton cycles must be
checked, which is unacceptable for big graphs with
more than 50 vertices.
For solving this problem with GAs [1][9] an
initial population with arbitrary Hamilton cycles
chromosomes, as well as with genes the vertices
of the graph under consideration, is built. They are
stored as permutations of the numbers from 1 to n,
according to the order in which the vertices are
visited in any Hamilton cycle. Population
development is guided by the objective function, by
which only a number of the generated Hamilton
cycles with the smallest length survive to the next
generation, and those with greater length drop out.
The genetic algorithm builds a Hamilton cycle with
optimal or close to the optimal length for only a few
steps, avoiding the necessity of considering all
solutions [1][9].
In http://www.boxcar2d.com/index.html [8]
a model is developed with the terms of GAs, aiming
at evolution simulation in order to study a
programming system for constructing high quality
2D cars.
Each car is a set of 8 randomly chosen
vectors: direction and magnitude. All the vectors
radiate from a central point (0,0) and are connected
with triangles. For each wheel it randomly chooses
a vertex to put the axle on, and picks an axle angle
from 0 to 2 pi.
For the purposes of GAs the following
model is developed [8]. Each car represents one
chromosome and has 22 variables such as: vertex,
axle angle, and radius, each represented as a real
number (or integer) with varying ranges. For the
selection process two algorithms are

implemented: Roulette-Wheel Selection and
Tournament Selection. The authors use two point
crossover, which means two random points along

the chromosome are selected and everything in
between is swapped. In addition to the crossover, in
each generation the chromosomes go through
- 65 -
distributed system so that the network traffic is

minimized. The developed in [5] Genetic
replication algorithm (GRA) and Adaptive GRA are
a good example. In their model [5] a chromosome
encoding a replication scheme is a bitstring
consisting of M genes (one for each site). Every
gene is composed of N bits (one for each object). A
1 value in the kth bit of the ith gene, denotes that ith
site holds a replica of kth object, otherwise it is 0.
Using this encoding the total length of a
chromosome is MN bits. [5] The main merits of
using a genetic algorithm approach in the dynamic
case lies in the proposed adaptive GA that uses
existing knowledge about replica distribution in
order to quickly define a new scheme.[5].
mutation. This means there is a probability that each

aspect of the car (or variable in the chromosome)
will change, as determined by the mutation rate
slider set by the user. When a variable mutates, a
new value is randomly chosen in the desired
range.[8]
Another example is the model developed
for solving the problem of minimizing network
traffic for fast dissemination and access of
information in large distributed systems, such as the
Internet with GAs [5]. The decision of what to
replicate where, requires solving a constraint
optimization problem, which is NP-complete in
general. The replication in large static distributed
systems is considered and it is aimed at finding the
appropriate allocation of the replicas in the
Fig. 1. A chromosome in the computer configuration world consists of six numbers, which act as gene
analogs. They determine quality(from 1 to 5) of CPU, HDD, Mark, Memory, Mother Board, and Video Card
to use.
Fig.2 A chromosome undergoing a series of mutations, each of which changes genes by adding or
subtracting 1. The original chromosome is 1-3-3-5-4-2 chromosome which produces a computer
configuration of quality 18. The final chromosome is a 5-5-4-5-4-4 chromosome, yielding a quality 27
computer configuration.
- 66 -
3. Model and procedures constructing

The design of the chromosome (or
individual) is probably the most important step in
making a successful genetic algorithm.
In the computer configuration world each
computer configuration is an individual. The
chromosome consists of six genes each of which
is a data structure with a number, quality from 1 to
5, description and a picture (Fig.1) The genes are as
it follows: CPU, HDD, Mother Board, RAM, Video
Card and Mark. The quality of the computer
configuration ranges from 5 to 30.
The fitness of a chromosome is the
probability that the chromosome survives to the
next generation. Accordingly a formula is required
to relate the fitness of the ith chromosome, fi, a
probability ranging from 0 to 1 to the quality of the
corresponding computer configuration, qi, a number
ranging from 5 to 30. The following formula, in
which the sum is over all candidates, is one
possibility[7]:
qi
fi
(1)
jq j
An example of calculated quality and
fitness of the individuals from one population by ten
computer configurations is given in Table 1.
N:
1
2
3
4
5
6
7
8
9
10
Table 1
The fitness of a ten chromosome population
CHROMOSOMES
QUALITY
STANDARD
FITNESS
1-3-3-5-4-2
2-5-5-3-4-5
5-5-4-5-4-4
3-4-5-1-1-1
3-3-3-2-4-5
4-5-2-2-3-4
5-5-3-2-5-5
5-2-3-4-4-5
1-1-4-4-5-4
4-3-2-5-5-5
18
24
27
15
20
20
25
23
19
24
0.08
0.11
0.13
0.07
0.09
0.09
0.12
0.11
0.09
0.11
Only half of the most adapted individuals

survive to the next generation (computer
configurations CC). As it is seen from Table 1,
these are the CCs with numbers 2,3,7,8, and 10.
These are the CCs which will participate in
mutation and crossover for obtaining new CCs,
whose quality will be evaluated again etc. up to
reaching the best quality CC.
Fig. 2 shows the process of mutation. A

chromosome is undergoing a series of mutations,
each of which changes genes by adding or
subtracting 1. The original chromosome is 1-3-3-54-2 chromosome, which produces a quality 18
computer configuration. The final chromosome is a
5-5-4-5-4-4 chromosome, yielding a quality 27
In Fig. 3 two chromosomes are undergoing
crossover, each of which is cut in the 2,4, and 6
genes and reattached to the other chromosome. One
of the original chromosomes is a 2-5-1-5-4-5
chromosome and the other is a 5-2-4-1-5-2
chromosome. One of the two new chromosomes is a
5-5-4-5-5-4 chromosome, which yields quality 28
For the work of the GAs a number of basic
functions of the programming system are developed
in order to realize the following possibilities:
for
random
computer
- function
configuration creation;
- function for selection of the 5 or 10 or 15
of the best generated computer configurations of the
generating population;
- function for viewing the best generated
computer configuration;
- function for realization of the mutation of
the selected genes. The step of mutation is +1 or -1;
- funciton for realization of the crossbreed
of the genes of the chromosomes;
- function for realization of the crossbreed
between 1-2-3 genes from one of the chromosomes
and 4-5-6 genes from another chromosome;
- function for realization of the crossbreed
between 2-4-6 genes from one of the chromosomes
and 1-3-5 genes from another chromosome.
The software system is implemented using
Visual Studio .NET, ASP.NET - server- side Web
application framework designed for Web
development to produce dynamic Web pages;
AJAX (Asynchronous JavaScript and XML) - a
group of interrelated web development techniques
used on the client-side to create asynchronous web
applications and the programming language C#.
A number of experiments are conducted
with the help of this system and the results from
them will be presented in the next section.
4. The experimental results
Genetic algorithms generally involve many
choices [7]:
How many chromosomes are to be in the
population? If the number is too low, all
chromosomes will soon have identical traits and
crossover will do nothing; if the number is too high
computation time will be unnecessarily excessive.
- 67 -
1
9
2
2
Quality
28
4
Quality
12
2
Fig. 3. Two chromosomes undergoing crossover,

each of which is cut in the 2,4, and 6 gene and
reattached to the other chromosome. One of the
original chromosomes is a 2-5-1-5-4-5 chromosome
and the other is a 5-2-4-1-5-2 chromosome. One of
the two new chromosomes is a 5-5-4-5-5-4
chromosome, which yields quality 28 computer
configuration.
What is the mutation rate? If the rate is too
low, new traits will appear too slowly in the
population; if the rate is too high, each generation
will be unrelated to the previous generation.
Is mating allowed? If so, how are mating
pairs selected, and how are crossover points
determined?
Can any chromosome appear more than
once in a population?
Based on these questions, a number of
experiments are conducted to investigate the
efficiency of the developed model:
- The search for the best quality computer
configuration is realized by creating populations
with different number of individuals, e.g. 10, 20 or
30 CC;
- The mechanisms of mutation and crossover
are used in some of the experiments, while in others
only mutation or only crossover is implemented.
Then the procedures for quality assessment and
natural selection are applied;
- When implementing crossover, experiments

are conducted in which different positions for
cutting the chromosomes are chosen, e.g. after the
2nd and the 4th gene or after the 1st, 3rd and 5th
gene or after the 3rd one. It is investigated how the
different choices influence finding the best solution.
The results show that, when crossover is
used along with mutation, the best quality computer
configuration is found much faster.
The small number of individuals in one
generation and the use of only crossover (with no
mutations) quickly leads to obtaining repeating
individuals and cannot result in achieving the best
solution.
The mutation leads to obtaining new genes,
not seen before and correspondingly to appearance
of new individuals with new quality. Therefore, the
use of mutation is compulsory. For solving the
problem in which it is impossible to achieve a
generation with zero quality (0) and when there will
be no surviving CCs to the next generation, the use
of only mutation (with no crossover) and selection
is sufficient for obtaining the best solution.
The efficiency of using GAs can be
illustrated by calculating the total number of CCs,
which can be built in solving the problem with full
running,
In the model considered here 6 components
are used for building a computer configuration, and
copies from each component with quality 1, 2, 3, 4
and 5; then 15625 computer confirgurations can be
made out of these components. These are all
of n=5 elements of
variations with repetition
k=6 class
(2)
rVnk 56 15625
variations with repetition.
In order to obtain the best quality computer
configuration with the help of the genetic algorithm
it is enough to build 50-100 computer
configurations.
The genetic algorithm is efficient for
building qualitative computer configurations
because without building all 15625 computer
configurations we manage to quickly build the best
and approximately the most qualitative one.
5. Conclusion
The efficiency of using a genetic algorithm
in optimizing the selection of computer components
(CPU, HDD, Mother Board, RAM, Video Card and
Mark) for building the best quality computer
configuration is investigated in this paper.
A model is proposed, in which the
computer components are analogues to the genes,
and the computer configurations themselves are the
analogue to the chromosomes. All genes can have
- 68 -
quality from 1 to 5 and, consequently, the quality of

the computer configurations can vary from 5 to 30.
A formula for quality assessment of each
computer configuration is suggested.
Procedures, analogous to mutation,
crossover and natural selection are developed. By
means of them and based on the quality assessment,
the process of evolution is modeled and the best
solution is searched for.
The experiments, conducted with the
developed software system confirm the efficiency
and the benefits from implementing genetic
algorithms for solving this type of problems.
The results from the conducted experiments
show that the GAs are efficient in solving this type
of problems and allow for avoiding a great number
of poor quality computer configurations.
The results also show that when along with
mutation crossover is also used, the best quality
computer configuration is found much faster.
On the one hand, the small number of
individuals in one generation and the use of only
crossover (with no mutation) quickly lead to
appearance of repeating individuals and cannot
result in the best solution. On the other hand, the
mutation leads to new genes, not seen before, and,
consequently, to appearance of new individuals
with new quality. Therefore, the use of mutation is
required.
The efficiency of implementation of genetic
algorithms is also confirmed by the fact that without
being needed to go through full running (i.e.,
through
building
all
15645
computer
configurations), it is sufficient to build only 5-6
generations of computer configurations with 10,20
or 30 individuals in order to find the best quality
3. Goldstein M. Jonathan, Genetic Algorithm

Simulation of the SHOP Scheduling Problem,
published by An ICMS/Shell Oil Business
Consultancy; Central Library of Imperial College (4
Management Thesis) theses, September 1991
4. Holland J.H., Adaptation in natural and artificial
systems, University of Michigan Press, Ann Arbor,
MI, 1975.
5. Loukopoulos Thanasis, Ishfaq Ahmad, Static
and adaptive distributed data replication using
genetic algorithms, Journal of parallel and
distributed computing; Elsevier, 0743-7315, 2004.
6. Schultz C. Alan, Learning Robot Behaviours
using Genetic Algorithms , by. Navy Center for
Applied Research in Artificial Intellignece. Central
Library of Imperial, Obtained : web page - theses
College, 2 Info. Desk MSc 1995.
7. Winston P., Artificial Intelligence, AddisonWesley, 1992
8. http://www.boxcar2d.com/index.html
9. http://www.doc.ic.ac.uk/~nd/surprise_96/journal/
vol4/tcw2/report.html#TSP

Technologies
4000 Plovdiv
BULGARIA
E-mail: dilyana_budakova@yahoo.com
Ref erences
1. ,., . , =
++, TopTeam Co., ISBN: 954-8905-06X, 2003.
2. Chughtai M., Determining Economic Equilibria
using Genetic Algorithms, published by Imperial
College theses, September 1995
13.02.2013 .
- 69 -

TECHSYS 2013
BULGARIA

,
:
.

.

.
: , , , ,
HALL SENSOR WITH PARALLER PORT

OUTPUT
DIMITAR GEORGIEV, GORAN GORANOV
Abstract: Analog to digital converting ant their further processing is integral part of ,modern
engineering. Sensors as sensing of modern automated systems must offer high accuracy and
reliability in their work. The constant pursuit of new, better approaches and methods for
measuring and achieving high accuracy is the primary objective in the development of sensors
and sensor systems.
Key words: sensor, sensor systems, engeneering, ADC, DAC
1.

,
.

.
,
.

.
.

()
,
,
.
,
.

,
,
.

,
,
.

.

.

.
- 70 -

8
.
2.

1
8
.
. 1.

8
.
CYL3503
( ), 8 ADC0808 (
),
555 ( ),

8,561KHz.
5V,
LM7805.[4, 5]
CYL3503
.
2.
200m 200m.
1,4V 4,9V.

.
5V.
. 2.
CYL3503
ADC0808,
,
, .
8
- . [1,
3]
3.
. 3. ADC0808
()
,
()
().

UK.

UK = UX.

N
. ADC0808,
R ,
. [2, 3, 4]
4
,
8
.

Lalim Parallel
Port Control Basic version 3.6.0,
5.[6]
- 71 -
3.

Lalim Parallel Port Control Basic

Tektronix TDS 1002
DM27.
6.

.

, V
3
2,5
DM27
Tektronix TDSA 1002
SOFTWARE
1,5
1
0,5
0
12,5 22,7
34
45,6 58,1 68,7 80,2 91,2
103
116 126 137 151 160

, mT
174
184
222
)

5
4,5
, V
4
3,5
3
2,5
2
1,5
DM27
Tektronix TDSA 1002
SOFTWARE
0,5
0
12,5
22,7
34
45,6
58,1
68,7
80,2
91,2 103,2 115,6 126,2 137,2 151
160 173,5 184,2 222
, mT
)
. 6.
. 4.
. 5. Lalim Parallel Port Control Basic

. 3,
LXI 2012

,

.[2]
4.

:
-

;
-

;
- ;
- 72 -
-
, ;
-
.
Department of Electronics
Technical University Gabrovo
4 Hadzi Dimitar St.
5800 Gabrovo
BULGARIA
E-mail: dimitar.g.georgiev@abv.bg
E-mail: g_goranov@bitex.bg
1. ., . , .
, , , 1983.
2. ., ,
,
,
. 3,
LXI 2012, .
3. ., .
,
, , 1988.
4. http://www.national.com/ds/AD/ADC0808.pdf.
5. http://www.fairchildsemi.com/ds/LM/LM7805.p
df.
6. http://www.softpedia.com/get/System/Hard-DiskUtils/Lalim-Parallel-Port-Control.shtml.
01.03.2013 .
- 73 -

TECHSYS 2013
BULGARIA

PSPICE
, ,
: OrCAD/PSpice
,
.
,
, .

.
.
: , , PSpice
PSPICE MODELING OF INVERTER INDUCTION

MACHINE DRIVE
DIMITAR SPIROV, NIKOLAY KOMITOV, NANKO BOZUKOV
Abstract: The software product OrCAD/PSpice allows examining simulations of electrical and
electronic systems, and the use of real models of electronic components. In this paper
simulation models have been developed of control block, connection block and power inverter
system for control of induction machine, and the machine model. Simulation studies were
conducted and received performance of the studied system. The reliability of the developed
model is validated with experimental studies of the motor.
Key words: inverter drive, induction machine, PSpice
1.
OrCAD/PSpice
,
-
: , ,
,
. Spice

,
, ,
- [1].
,
-,
,
[1]. PSpice
,
[3, 4, 5].
- 74 -

,

,

PSpice.
.
2.

PSpice
()

, .
. 1.
U DC
uabc , iabc
rTG
. 1.

[2]:
, .

,

[2].

,
Vtr
120
V1, V2, V3.
. 2.
, (S1 S6).
,
().

.
N- MOSFET IRF 150.
. 3 ,
.
Vdc1 Vdc2.
. 3.

()
,
:
dis
di
+ Lm r ;
dt
dt
dis
dir
u s = Rs is + Ls
+ Lm
;
dt
dt
(1)
dis
dir
+ Lr
+ r Lmis + r Lr ir ;
0 = Rr ir + Lm
dt
dt
di
di
0 = Rr ir + Lm s + Lr r r Lmis r Lr ir .
dt
dt
u s = Rs is + Ls

. 4 . 5.
. 4. -

:
. 2.
. PSpice
J m dr
= Me Mc ,
p p dt
(2)
- 75 -
= 1,5 p p Lm (is ir is ir )
.
PSpice
, . 6.
3.
AO-90S-4,

(Mc=Mc0=0,9735Nm).

.

.

.
. 5. -

. 8. us, us =f(t)
. 6.

,

:
. 7.
u s =
2
1
1
u sa u sb u sc ;
3
2
2
u s =
2 3
3
.
u
u
sb
sc
3 2
2
(3)
. 9. is,is=f(t)
. 7.

. 10. , r=f(t)

,
- 76 -

, Hall-

NI USB-6009
National
Instruments
,
.

MATLAB/Simulink.
,
iA
isa .
11.
25
iA , isa, A
iA
20
isa
15
10
1. , ., . , . ,

PSpice, CIELA,
, 2002.
2. , ., . , . ,
, ,
, 2006.
5
0
-5
3. Rashid, M., Power

Academic Press, 2001.
-10
-15
-20
-25
1.6
1.65
1.7
1.75
t, s
1.85
1.8
. 11. iA, isa=f(t)
,
rTG
r=r/pp . 12,
pp .
160
4.

,

,

PSpice.

,
.
-1
rTG, r, s
140
120
100
80
60
40
rTG
20
0
1.95
r
2
2.05
2.1
2.15
2.2
2.25
t, s
electronics
4. Reynaud, J., P. Pillay, Modeling of saturation in

induction machines using EMTP, PSpice and a
dedicated computer program, Electric Power
Systems Research 30, 1994, 279- 285.
5. Rinkeviien, R., A. Petrovas, Modelling AC
Induction Drive in Pspice, ELECTRONICS AND
ELECTRICAL ENGINEERING, No. 1, 29-32.

AO-90S-4:
PN=1,1kW; UN=380V; IN=2,8A; f=50Hz; pp=2;
MN=7,45Nm;
cos=0,8;
nN=1410min-1;
Jm=0,0014kgm2.
Rs=7,3;
Rr=5,0026;
Ls=0,0184H;
Lr=0,0184H; Lm=0,3347H.

26 Maritza Blvd.
4002 Plovdiv
BULGARIA
E-mail: dimitar_spirov@abv.bg
E-mail: nikkomitov@abv.bg
. 12. rTG, r=f(t)
,
,
handbook,
28.02.2013 .
- 77 -

TECHSYS 2013
BULGARIA
-
-
1, 2
: -
.
SMF-28 .
- .
.
.
: , , -
A SIMPLE LONG-TAPER CORE-CLADDING

INTERMODAL MACH-ZEHNDER TYPE
INTERFEROMETER
EMIL PENCHEV1, TINKO EFTIMOV2
Abstract: We report a novel and simple construction of a fiber-optic core-cladding intermodal
interferometer based on long fiber tapers. A standard SMF-28 communication fiber is used. The
equivalent optical scheme is that of a Mach-Zehnder interferometer. The spectral responses and
the dependence on the fiber length between the two tapered regions are presented. Possible
applications are commented.
Key words: fiber tapers, core-cladding interference, fiber-optic interferometers
1. Introduction
Core-cladding intermodal interferometers
have been proposed several years ago and have
caused a significant interest because of their
capability to sense surrounding refractive index
(SRI) changes [3]. The reason for this increased
sensitivity is the fact that interference is between the
fundamental core mode which is insensitive to SRI
changes and serves as a reference arm, and a
cladding mode of an effective glass-air waveguide
whose propagation constant and attenuation are
sensitive to SRI changes. To excite the cladding
mode a section of the fiber must be stripped off to
allow a glass-air waveguide to be formed. Then a
perturbation must be imposed prior to and after the
stripped section which couples a part of the core

mode to a cladding mode and another perturbation
of the same type recombines both modes. In the
subsequent section of the non-stripped fiber, the
cladding mode attenuates. There are several
methods to excite the cladding mode: long period
gratings (LPG), abrupt tapers, short tapers, off-axis
splicing, core mismatch, multimode fiber segment,
air-hole collapsing in PCF fibers [1], [2], [5]. In the
present paper we report the construction of a simple
core-cladding intermodal interferometer of a MachZehnder type in which the splitter/recombiner is a
long taper. This constriction has specific properties
which are discussed in this paper.
- 78 -
2. The
long-taper
core-cladding
interferometer
The long-taper core-cladding interferometer
is a fiber optic structure that contains two tapered
regions with similar profiles and length. The first
one couples part of the core mode in the cladding
and the second recombines the core and cladding
modes (fig .1). That leads to interference between
the core mode and cladding modes.
Fig. 1. Core-cladding interferometer of a MachZehnder type.
Fig. 3 Block diagram of the experimental set-up.

The sensing fiber is a Corning standard single mode
communication fiber SMF-28 with 9/125 m
core/cladding diameter. The fibers which couple the
sensing fiber to the light source and the OSA are
also SMF-28fibers.
The phase difference m between the core

and the cladding modes is given by (1)
2n m eff L
(1)
where nmeff is the effective RI difference between

the core mode and the m-th cladding mode, is the
input wavelength and L is the interaction
(interferometer) length between the two tapered
regions [4]. As shown on fig. 2 part of the cladding
mode interacts with the surrounding area thus
change in m is expected if SRI changes. That
makes the long-taper core-cladding interferometer
sensitive to SRI changes. The dependence between
L and the period is given as [5].
n m eff L
(2)
Fig. 4 The spectrum of the light source.

4. Results and discussions
In our experiments we used long-tapered
core-cladding interferometers with different
interaction length L between the two tapers. As a
result we demonstrated that the dependence
between L and is similar to that in (2). Increase
of L leads to a decrease of the spectral period, i.e.
the longer interaction section, the more interference
peaks appear as evident from the plots in Fig. 5
It can be seen that the spectral period will increase

if interferometer length decreases [4].
-4
-5
Loss (dB)
-6
-7
-8
-9
-10
-11
1520 1530 1540 1550 1560 1570 1580 1590 1600 1610 1620
Wavelength (nm)
a)
Fig. 2. Mode field in fiber tapers.
-10
-12
-14
-16
Loss (dB)
3. Experimental set-up
The experimental set-up is shown in Fig.3.
As a light source in our experiments we use
JOINWIT Optical Light Source. The spectrum of
emission of the light source is shown in Fig. 4. The
interrogation unit is an ANDO AQ6331 Optical
Spectrum Analyzer (OSA).
-18
-20
-22
-24
-26
-28
-30
1520
1530
1540
1550
1560
1570
1580
Wavelength (nm)
b)
1590
1600
1610
1620
-6
-7
-2
Wavelength sift
Loss (dB)
- 79 -
-8
-9
-10
-11
-4
-6
-8
-10
-12
1520
1530
1540
1550
1560
1570
1580
1590
1600
1610
-12
1620
1,1
1,2
Wavelength (nm)
1,3
1,4
1,5
RIU
c)
Fig. 5 Dependence between L and : a) L=10mm,
b) L=25mm and c) L=40mm.
Fig.6. represents the inverse length dependence of
the spectral period and confirms eqn. (2)
Fig.8 Peak shift with increasing SRI.

The wavelength shift vs. SRI diagram is
shown in Fig. 7.
4.2. Temperature response.
The core-cladding interferometer with
L=25mm was tested for a temperature response.
The experimental results for the temperature
response of the interferometer are shown in fig. 9.
-11
Loss (dB)
-13
-15
-17
-19
-21
-23
-25
18 C
-27
-29
-31
1520
Fig.6 Dependence between L and .
-10
1560
1580
1600
1620
Fig. 9 Temperature sensitivity of the interferometer.

The wavelength shift of the peaks was
estimated to be 46.92pm/oC.
5. Conclusions
-15
Loss (dB)
1540
Wavelength (nm)
4.1. SRI response.

The SRI response of the core-cladding
interferometer was studied. The observed
wavelength shifts of the peaks are illustrated in Fig.
7.
The experiments performed allow us to

formulate the following conclusions.
-20
-25
-30
-35
1520
37 C
1530
1540
1.333
1.3475
1.3609
1.3748
1.3896
1.4022
1.4176
1.4312
1.4464
1550
1560
Wavelength (nm)
First, we have shown that the simple long-taper

core-cladding interferometer can be used for
surrounding area refractive index measurements.
Second, the temperature response of the
interferometer shows that it is practically insensitive
to changes in the temperature of the environment.
Fig. 7. Wavelength shift.

With increasing the SRI the peaks shift to
shorter wavelengths. In our experiments as a
surrounding medium we use ddWater - Glycerol
solution.
REFERENCES
1. Benye Li , Lan Jiang, Sumei Wang, Lanying
Zhou, Hai Xiao and Hai-Lung Tsai, UltraAbrupt Tapered Fiber Mach-Zehnder Interferometer
Sensors, Sensors 2011, 11, 5729-5739.
- 80 -
2. Byeong Ha Lee, Young Ho Kim, Kwan Seob

Park, Joo Beom Eom, Myoung Jin Kim, Byung
Sup Rho and Hae Young Choi, Interferometric
Fiber Optic Sensors, Sensors 2012, 12, 2467-2486
3. Caldas P., P.A.S. Jorge, F. Arajo, L. A.
Ferreira, G. Rego and J. L. Santos. Effect of
fiber tapering in LPG-based Mach-Zehnder modal
interferometers for refractive-index sensing Proc.
of SPIE OFS 20, Vol. 7503 750349-1 to 4
4. Di Wu, Tao zhu, Member, Ieee, Member,
OSA, Kin Seng Chiang, Member, IEEE, Fellow,
OSA, and Ming Deng All Single-Mode Fiber
Mach-Zehnder Interferometer Based on Two
Peanut-Shaped Structure
5. W. J. Bock, T. Eftimov, P. Mikulic and J. Chen
(2009), "An in-line core-cladding intermodal
interferometer using a photonic crystal fiber." , J.
Lightwave Technol, vol. 27, issue 17, pp. 3933-3939
1
Department of Physics and Mathematics,

University of Food Technologies, Plovdiv
2
Faculty of Physics and Engineering
Technologies, PU P. Hilendarski, Plovdiv
E-mail: emil.penchev@abv.bg
E-mail: teftimov@abv.bg
28.02.2013 .
- 81 -

TECHSYS 2013
BULGARIA
,
:
.
,
,
.

.
: , , ,
CUALITY CRITERIA FOR OPTIMAL

FUNCTIONING OF THE CIRCUITS
GALINA CHERNEVA, ELENA DIMKINA
Abstract: Analysis and diagnostics are close connected and mutually complementing problems
in the theory of electrical circuits. With the analysis the characteristics of circuits with known
structure and parameters are defined, while the diagnostics is connected with a study on the
definite variables of the circuit, it are the custom to diagnostic indicators. The paper presents a
functional criteria for an evaluation of the optimum functioning of the circuits on the base of the
sensitivity theory.
Key words: Diagnostics of electrical circuits, sensitivity, optimization, functional
1.

-
[1,2,4].

,
,

,
.
[3].
:

( )
[2,3,4].

.
-
[2],
Z- Y- [3],
[5] .
- 82 -

.

[1]
,

.

(),

(
).

[1].

,

.
2.
,
,
, .1, :
X=(x1, x2,,xn )
,
Y=(y1, y2,,yl) ,
Z=(z1,z2,,zr )
.
=[pij], i 1...l , j 1...n .
pij (2)
,
.
- ,
, pij
(2)
.

(1)
. pij
.
,
.
- (1)
- ,

pij
.
, , pij

.
3.

H H1 ,......... .., H k ,
Z=(z1,z2,,zr )
X=(x1, x2,,xn )
EB
Y=(y1, y2,,yl )
.1.

Y :
Y = P(, Z) ,
(1)
,

,
.
P :
(2)
(3)

.
Y,
( ) .
.

.
H i H :
Hi , Hi 1 ,.... m , , i 1..k ,
(4)
1 ,......... .., m R
(5)
, R ,
- 83 -
(,
, ).
,
j
j 1...m ,
j , j 1...m ,
1 ,......... .., m
(6)
.
(3)
k- :
k
d H i
d 1 .....
d m H i
m
1
H i , H i , H i ,
1
dH i , d 2 H i , ...
2!
1
(8)
d n1 H i , ) ,
n 1!
[1]
H i ,
S j ,
, j 1...m ,
j
(9)
j .

1 ,......... .., m

,
,
[4].
,
,
, :
(10)
D
.

0 ,

,
(10) .
H i
, (10)
:
(7)
1 ,..., m
(k+1)- .
H i , ,
[4]:
I H i , D j d
I S j j D j d .
(11)

(12)

.

[4].

,
,

, .
(8) (9)

1 ,......... .., m .
, (11)
, ,
, :
C P ,
:
P
(12)
DHd ,
(13)
DS i dt .
(14)
[C]
Sj Si
m - .
- 84 -
C P , detC 0
1
(15)
.
:
k Ck Pk
1
(16)

, :
0 i ,
(17)
i 1
(+1)-
:
I p 1 I p Pp C p
1. ..
.
. . 2002.
2. .., .. .

. . .1988.
3. .., .. , .. .
.
., , 1983.
4. ., ., .
. . 1987.
5. Constantinesu F., C.V. Marin, M. Nitescu, D.
Marin. A new approach to parameter identification
of linear circuits. IEEE conference Signals,
Circuits and Systems, Romania, 2003, p 457
460.
, :
I 0 I1 .... I p I p1 ....
(18)
..
,
,
(10) .

Todor
Kableshkov
University
TransportSofia.
158 Geo Milev St.
1574 Sofia
BULGARIA
E-mail: cherneva@vtu.bg
E-mail: e.dimkina@abv.bg
4.

,

. ,

. ,

,

.
13.02.2013 .
- 85 -

TECHSYS 2013
BULGARIA

: ,

.
.

.
:
DEMAND POWER CONTROL THROUGH

POWER STORAGE SYSTEMS
GEORGI GANEV
Abstract: The power storage systems development are studied in this paper, a comparative
analysis of their technical parameters is made and the areas of their application are marked.
Demand power consumed by one real object is proposed. The perspective of local power
storage systems usage for local consumers demand power control is proved.
Key words: power storage systems, demand power control
1.

:
/
,
;
/
-
();
/
;
/

. ,
2003.
..
.
,

. -
SCADA-

.

.

,
- 86 -
2.
2.1. .
,

() [1,3]:
2.1.1. -
()

,
.
,
;

;

(.1).
,

; ,
. ,
( 1,4GW);

38,3GW, 21,5GW.
2.1.2. ()
,
.
,

.
,
-.

(.1).

-
,
.
2.1.3. ()

.

.
,
,

; /

/ (.1).

.

.
(()),
6000 min-1.
,
, ...
-
(()).

,
.
, () -
.
2.1.4. ()

,
.

.
,
,
,
(.1).

( 9 [3]).

,
.
2.1.5. ()
.
(LA),
- (NiCd), - .

(.1).
,
,
... ,
.
-

(UPS),

/
.

- 87 -
(flow batteries).

,
.
,

polysulphide-bromide
batteries
(PSB,
RegenesysTM),
zinc-bromine
batteries (ZBB) vanadium-redox batteries (VRB).
,
, . ,
,
( , .)
, ,
.

,
.
2.1.6. ()

1.
[1]
.
.
.
20
36
15 7
18
0,25
MW
2100
350
<1,6 1
<0,7 2
kWh/kW
>5
~5
0.001
kW/kg
kW/m3
3
1700
530
2,5
170 000
kWh/kg
530
0,6 6
kW/m3
0,4
7 000
50 3
75 4
240 5
247
53
75
70
>95
80
<70
>95
60s
24h
10s
5ms
30ms
1000 3
2500 4,5
2-10
70-4000
1000
300 3
600 4
800 5
400
10000 6
2000 8
15 6
5ms
12h
24h
30ms
15min
424
90 1
93 2
120s 1
1h 2
5ms
..
1000000
.
/kW
/kW
60
35-70
450
40
10-70
500
30
140-350
30
200-500
1000
,%
$/kW
600
480
$/kWh
300
350 2
300 1
20000 2
300
120
1000
1830 6
1100 7
640 8
1000 6
>30
70-400
300
13000 28000
: , : 1 / 2
; 3LA; 4NiCd; 5NaS; 6VRB; 7PSB; 8ZBB.
2.2.

.
,
,
, ,
.,

,
(.1).

.

,
[1]:
/ ( 1 );
/ (
);
- 88 -
/ (
).
-,
:
/

. ,
,

( ).

. ,

.

;
/

.
,
-
-

( 2. 6.) .

.. ,
;
/

.. ,

.

,
.
.1

.

,
[1].
. 1.
3.

,
, , .

,
,
(
),
,
.
- 89 -
.2, 3 4
,
. .2

,
.3 - .4. - .

Power Logic Schneider Electric.
,
.

,
( ), (
).
/
. 2. (, 2010.)
. 3.
(, 2013.)

2010-2013 , :
/
/ - 2 3 ;
/

(.2 3);
/
9 12
14 17 (.4);
/ 22 6

15kW o
30kW (.2, 3 4);
/ , 1 min
15min
15% ( .4 4).
/
. 4. (,
2013.)
4.

.

,
[2].

.

.
,
15min,
- 90 -
.

.

.

/;
, .

(.1)
(.5).
.
~
~
~
.
. 5.
W

I :
1
1
(1)
W .I.2 .k .m.r 2 .2
2
2
m ; r
; k
.
.6

9MJ (.. 10kW
15min), , k =1.
r=0,5m
r=1,0m
r=1,5m
;
-
,

;
-

;
-

SCADA;
- ...
.
5.

.
,

,
.

(IV
- , )
,
.
1. ., ..
, -,
60, .1, .229-238, 2010, ISSN 1311-0829.
2. .
, , 2012.
100
3. Carbone R. Energy storage in the emerging era

of smart grids, 2011.
m, t
4. Kampf T. High speed flywheel design using

advanced composite materials, sept., 2012.
5. Ostergard R. Flywheel energy storage a
conceptual study, dec., 2011.
10
1
2000
4000
6000
-1
8000
, min
.6.
1 10

[3,4,5]:
- /
/ -

4000 Plovdiv
BULGARIA
E-mail: gganev@tu-plovdiv.bg
01.03.2013 .
- 91 Journal of the Technical University Sofia

TECHSYS 2013
BULGARIA

MRAS
,
: (MRAS)
.
MRAS ,
. ,

(backpropagation).
MATLAB/Simulink
Texas Instruments TMS320C2000 .

.
: , , , , , ,
,
ANGULAR VELOCITY SPEED ESTIMATION OF

INDUCTION DRIVES BASED
ON NEURAL NETWORK MRAS MODEL
GEORGI IVANOV, DIMITAR SPIROV
Abstract: Model reference adaptive systems (MRAS) is widely applied in systems for sensorless
speed control. The article analyzes evaluation method based on the speed of neural MRAS
model oriented on rotor flux linkage. Adjustable model is implemented as a neural network, and
the speed is adjusted by gradient algorithm (backpropagation). Developed model was tested in
simulation environment MATLAB/Simulink and real-time microprocessor platform of Texas
Instruments - TMS320C2000. Conclusions are made regarding the accuracy of the different
operating modes of the presented algorithm.
Key words: sensorless, speed, estimation, induction, motor, artificial, neural, network
1.
,
-

[22].

[7, 12, 13, 18, 19, 20,
24, 25],
- ,
,

,

- 92 [17].
(, )
( )
,
[2]. -

[3, 10, 11]. MRAS ,

-...
[1, 6, 8, 14].

,

(),

,

- 1Hz [1, 10, 23].

-
.

[5, 15, 21].
MRAS

[9, 16, 22].
,

,
[20].

,

.
[4]

TMS320C2000.

MRAS

C28xx.
2. ,
MRAS

MRAS ,

,

(.1).
r
Us
r
Is
r
r
r
r
w2
Z 1
. 1. MRAS

[1],
(1),
r =
r
Lr
( s Ls .is )
Lm
L
= r ( s Ls .i s )
Lm
(1)

[1]:
d r Lm
1
i s r r r
=
dt
Tr
Tr
d r
dt
L
1
= m i s + r r r
Tr
Tr
(2)

(
)
:
r (k ) = w1 r (k 1) w2T r (k 1) + w3is (k 1)
r (k ) = w1 r (k 1) + w2T r (k 1) + w3is (k 1)
(3)
c=T/Tr ,
(Tr) ,
:
w1 = 1 c
w2 = r cTr = rT
(4)
w3 = cLm
(4) , w1 w3
, w2
.

, .2.
4 .

- 93 , ()
[ r ( k 1), r ( k 1) ]
,

[is(k-1), is(k-1)].
2

[ r ( k ), r ( k ) ].
.
()

. -
r (k 1)
W1
r (k 1)
-W2
is (k 1)
W3
is (k 1)
W3
W2
r (k )
W1
r (k )
w2 (k ) =
E
E r (k )
=
w
r (k ) w2
2
,
E / r (k ) ,
(6), :
E
1 {[ r ( k ) r (k )]}
=
= T ( k )
r ( k ) 2
r (k )
r (k ) / w2 ,
(6)
(3)
w2:
r (k )
= [ r (k 1), (k 1)]T
w2
,
E=(1/2)2(k) , :
(8)
(7) (8)
(6) , (k ) = [ ( k ), ( k )]T
:
w2 (k ) = T (k )[ r (k 1), r (k 1)]T =
.2

(6)
= [ (k ) r (k 1) + (k ) r (k 1)]
(9)
(k ) = r (k ) r (k )
.
(9)
( k ) = r ( k ) r ( k )
r ( k ) = [ r (k ), r (k )]T ;
, .
(, )
(9):
r (k ) = [ r (k ), r (k )]T .
w2 (k ) = {[ r (k ) r (k )] r (k 1)
(k ) = r (k ) r (k ) ;
,

E/w2, ,
..
..
,
- -
,
,

.
-,
w2 :
E
w2 (k ) =
w2
(5)
(5)
r / r . ,
[ r (k ) r (k )] r (k 1)}
(10)
. 2
(11)
w2 = w2 (k 1) + w2 (k ) =
= w2 (k 1) + {[ r (k ) r (k )] r (k 1) (11)
[ r (k ) r (k )] r (k 1)}
- -
,

.
w2 (6),

:
r (k ) = r (k 1) +
w2 (k )
=
T
= r (k 1) + {[ r (k ) r (k )] r (k 1)
T
[ r (k ) r (k )] r (k 1)}
(12)
- 94 2.

AO-90S-4,
.
. eZdsp F2808
()
TMS320F2808, Simulink.

Hall- LTS 15-NP
Hall- LV 25-P
,
[4].

.

.
400
1.5
300
0.5
-0.5
-1
1.6
1.65
1.7
1.75
1.8
.5 r, r=f(t)
,
rTG
rMRAS=r/pp
. 6, pp
.
-1
r, s
140
200
120
100
100
80
-100
60
40
-200
20
-300
t, s
-400
1.6
1.65
1.7
1.75
1.8
.3 us, us=f(t)
25
t, s
-1.5
160
us
us ,us , V
us
r, r , Wb
-20
rTG
1.6
1.65
1.7
1.75
t, s
1.8
.6 rMRAS, rTG=f(t)
is, is , A
,
,

.
is
20
rMRAS
is
15
10
5
0
-5
-10
-15
-20
-25
t, s
1.6
1.65
1.7
1.75
1.8
.4 is, is=f(t)
4.

MRAS ,

.
,
.

matlab/simulink
Texas
Instruments TMS320C2000.
- 95 ,
1. ., . , . ,
MRAS
, Journal of the Technical

University Sofia Plovdiv branch, Bulgaria,
Fundamental Sciences and Applications Vol. 18,
2012, ISSN 1310 827, .33-42.
2. ., . , .,
, 160., ISBN-10
954-323-233-4,
ISBN-13
978-954-323-233-8,
, , 2006.
3. ., .,

,
, 2008.
4. ., . ,

, UNITECH 12, Gabrovo.
Vol. 1, pp.104-109.
5. Akatsu K., and A. Kawamura, Sensorless very
low-speed and zero-speed estimations with online
rotor resistance estimation of induction motor
without signal injection, IEEE Trans. on Industry
Applications, vol. 36, No. 3, pp. 764-771, 2000.
6. Akin B., State Estimation Techniques For Speed
Sensorless Field Oriented Control Of Induction
Motors, Middle East Technical University, Ankara,
2003.
7. Bose B., Power Eectronics and Motor Drives Advances and Trends, Academic press, 2006.
8. Filippich M., Digital Control of a Three Phase
Induction Motor, The University of Queensland,
2002.
9. Gadoue S.,D. Giaoris, J. Finch, Low Speed
Operation Improvement of MRAS Sensorless
Vector Control Induction Motor Drive Using Neural
Network Flux Observers, IEEE IECON 2006, Paris,
pp. 1212-1217.
10. Holtz, J., Sensorless Control of Induction
Motor Drives. Proceedings of IEEE, Vol. 90, 8,
pp. 1359-1394, August 2002.
11. Ilas, C., A. Bettini, L. Ferraris, G. Griva, F.
Profumo, Comparison of Different Schemes
without Shaft Sensors for Field Oriented Control
Drives, IEEE IECON, 1994, pp. 1579-1588.
12. Kar B. N., Indirect Vector Control of Induction
Motor using Sliding-Mode Controller, IEEE
SEISCON 2011, Chennai, pp. 507-511.
13. Krishnan R., Electric Motor Drives. Modeling,
Analysis, and Control, Singapore, 2003.
14. Ma Chaozheng, Speed Sensorless Control Of

3-phase Induction Motor Using MRAS Speed
Estimator, Ryerson University, Toronto, 2003.
15. Marino R., S. Peresada, and P. Tomei, Online stator and rotor resistance estimation for
induction motors, IEEE Trans. on Control
Systems Technology, vol. 8, no. 3, pp. 570-579,
2000.
16.Nandhini Gayathri M., S. Himavathri,
Sankaran, Neural Learning Algorithm Based
MRAS Rotor Resistance Estimator Using Reactive
Power Technique For Vector Controlled Induction
Motor Drive, Journal Of Theoretical And Applied
Information Technology 2012, Vol. 45 No.1, pp.
356-364.
17. Raghu T.,J. Srinivas Rao, S. Chandra
Sekhar, Simulation of Sensorless Speed Control Of
Induction Motor Using APFO Technique Drives,
International Journal of Electrical and Electronics
Engineering, vol.4, No. 4, pp 440-444, eISSN 20103972, 2012.
18. Szabat K., M. Kaminski, T. OrlowskaKowalska. Robust Control of an Electrical Drive
using Adaptive Fuzzy Logic Control Structure with
Sliding-Mode Compensator, The International
Conference on Computer as a Tool EUROCON
2007, Warsaw, pp. 1706-1711.
19. Tomar A., Yog Raj.Various Control
Techniques for Induction Motor Drive: A Brief
Insight, International Journal of Engineering
research & Technology (IJERT), 2012, vol.1 Issue
5.
20. Vas, P. Sensorless Vector and Direct Torque
Control. Oxford University Press, 1998.
21. Vasic V., S.N.Vukosavic, E.Levi, A stator
resistance estimation scheme for speed sensorless
rotor flux oriented induction motor drives, IEEE
Trans. on Energy Conversion, vol. 18, no. 4, 2003,
pp. 476-483.
22. Venkadasen A., S. Himavathi. Novel SNCNN-MRAS Based speed Estimator for Sensorless
Vector Controlled IM Drives, International Journal
of Electrical and Electronics Engineering, eISSN
2010-3972, 2011.
23. Vonkomer J. Induction Motor Sensorless
Vector Control for very wide Speed Range of
Operation, IEEE Carpathian Control Conference
(ICCC), Velke Karlovice, 2011, pp. 437-442.
24. Yang L., A novel Fuzzy Logic Controller for
Indirect Vector Control Induction Motor Drive,
IEEE Intelligent Control and Automation, 2008,
Chongqing, pp. 24-28.
25. Yongdong Li. Predicitve Control of Torque and
Flux of Induction Motor Drives, IEEE Power
Electronics and Drives Systems, 2006, Kuala
Lumpur, pp. 67-71.
- 96

AO-90S-4:
PN= 1,1kW; UN= 380V; IN= 2,8A; f= 50Hz; pp=2;
=
MN 7,45Nm;
cos=0,8;
nN=1410min-1;
Jm=0,001kgm2.
Rr=5.00255;
Ls=0.01839H;
Rs=7,45;
Lr=0.01839H; Lm=0.33475H.

- 1220031-19
09.05.2012
.
25 Tsanko Dyustabanov St.
4000 Plovdiv
BULGARIA
E-mail: georgi.iwanow@gmail.com
26 Maritza Blvd.
4002 Plovdiv
BULGARIA
22.02.2013 .

TECHSYS 2013
BULGARIA

MRAS

, ,
: (MRAS)
,
.
MRAS ,
.
(backpropagation),
.
MATLAB/Simulink
Fuzzy Logic Toolbox.
,
.
: , , , , , ,
,
ANGULAR VELOCITY SPEED ESTIMATION OF

INDUCTION DRIVES BASED ON NEURAL
NETWORK MRAS MODEL WITH FUZZY
LEARNING RATE ADAPTATION
GEORGI IVANOV, SEVIL AHMED, IVAN KOSTOV
Abstract: Model reference adaptive systems (MRAS) for sensorless speed estimation are
characterized with robustness and wide operating range, relatively simple implementation and
high dynamic and static performance. The article analyzes evaluation method based on the
speed of neural MRAS model oriented on rotor flux linkage. Adjustable model is implemented as
a neural network and the speed is adjusted by gradient algorithm (backpropagation) and
learning rate is determined by fuzzy logic controller (Mamdani). Developed model was tested in
simulation environment of MATLAB/Simulink and instruments of Fuzzy Logic Toolbox.
Conclusions are made regarding the accuracy in tracking mode and compensation of
disturbances and also during operation of electric power in low speed region.
Key words: sensorless, speed, estimation, induction, motor, artificial, neural, network
- 98 1.

,
(, )
(,
..), (
) (
) [5].

, ..,

.

,
,

[2].

, -
, ,

, ,
[10].

[2].
,

[10, 11].
MRAS
, ,
, [10, 11]
,

(sliding mode control) [4].

[6, 11].

,

[13].

,

.

[7, 8].

[12].
-

.

MRAS,
,
.
2. ,
MRAS

MRAS ,

,
,
.
MRAS
.1:
r
Us
r
Is
r
r
r
r
w2
Z
. 1. MRAS

[11],

[1]:
r (k ) = w1 r (k 1) w2T r (k 1) + w3is (k 1)
r (k ) = w1 r (k 1) + w2T r (k 1) + w3is (k 1)
(1)
- 99
r (k )
w1 = 1 c
w2 = r cTr = rT
(2)
w3 = cLm
r ( k )
r (k )
r ( k )
w1 w3 ,
w2,

.

(backpropagation) [3]

.
- k-
(3):
E
w2
(3)
E=(1/2)2(k).
(3)
r / r . ,
:
w2 (k ) =
E
E r (k )
=
w2
r (k ) w2
Z-1
(k )
r ( k )
. 2.

[9], :
(k ) = [ r (k ) r (k )] r (k 1)
[ r (k ) r (k )] r (k 1)
(6)
(k ) = (k ) (k 1)
.
,
. 3

MRAS .
(4)

,
(3) (4),
(2)
:
r (k ) = r (k 1) +
E
w2
Z-1
Z-1
c=T/Tr.
w2 (k ) =
(k )
w2 (k )
=
T
= r (k 1) + {[ r (k ) r (k )] r (k 1) (5)
T
[ r (k ) r (k )] r (k 1)}

[1, 11]. -
-
,
,

.

.

.2:
. 3.

.

(. 4).
a)
- 100 3) if (k ) is small and (k ) is big

then is slow
M
9) if (k ) is big and (k ) is big
then is medium
)
. 4.
) ) )

small,
medium big, ,
(k ) (k ) .
,
,
. 1.
1

(k )
small
medium

Fuzzy Logic Toolbox
(FLT) MATLAB/Simulink.

. 4. . 1
FLT,
(AND), (OR), ,

(. 2):
2

min
max
prod
sum
centroid
3.

MATLAB/Simulink
:
MRAS Speed Estmator by Neural Rotor Flux Model and fuzzy Learning Adaptation
wm
v s_abc
us_abc
3-phase
Voltage Supply
big
wk
us_v u
Tm
wm
is_ba
us_ba
Mechanical Angular
Velocity Estimation
wm
MRAS_NNFLRA
Induction Motor
Model
wk
(k )
is_v u
Compare Estimation
and Actual Velocity
Reference
Frame
small
medium
fast
fast
medium
slow
medium
fast
big
slow
slow
medium
,
. 1, ,
- (if-then,
-) -
[3]:
1) if (k ) is small and (k ) is small
then is medium
2) if (k ) is small and (k ) is medium

then is slow
Load
Clock
To Workspace1
. 5.

w2
.6:
Product3
1
yr_ba
1
Product1
1
2
z
Unit Delay1
yr^_ba
Product4
z
Unit Delay
Unit Delay2
Fuzzy Logic
Controller
dw2
Scope3
. 6. w2

- 101
.10
.
156.6
156
155.8
0.5
1.5
t, s
2.5
3.5
160
MRASNNFLRA
140
120
157.1
100
157.05
-1
, s
157
156.95
156.9
80
60
40
156.85
actual speed
20
MRAS NNFLRA
156.8
0.2
0.3
0.4
t, s
0.5
0.6
0.7
0.8
. 7.
0
-20
actual speed
1.5
t, s
2.5
3.5
4.5
MRASNNFLRA
85
, s
0.5
MRASNN
86
-1
. 10.

87
3.2
84
3.1
83
3
-1
, s
82
81
80
. 9.
157.15
-1
, s
156.4
156.2
MRASNN
157.2
MRASNNFLRA
156.8
actual speed
157.25
actual speed
157
, s -1
.
-90S-4,
, [1]
U/f n,

. .7 .8

,
actual speed, MRAS
MRASNN,
MRAS

MRASNNFLRA. ,
MRASNN
0.01,
MRASNNFLRA
=0.000010.1
2.9
2.8
0.067
0.068
0.069
0.07
t, s
0.071
0.072
. 8.

, .9.

t= 1.5s
(= ),
= 0.5.

- 380V/50Hz 190V/25Hz.
actual speed
2.7
MRASNNFLRA
2.6
0.5
1.5
t, s
2.5
3.5
4.5
. 11.

MRAS

1Hz. . 11

3.14 rad/s
.
- 102

( mi ) (mse)
(max) (.3)
:
n
mse =
(
i =1
mi
mi ) 2
, i =1 n
(7)
max = max(| mi mi |)
n .
3
MRAS
MRASNN
MRASNNFLRA
mse
0.632
0.0013
0.00014
max
4.156
0.5628
0.2429
4.

.
MRAS ,

.
,

MRAS.

.

.
1. ., . ,

MRAS , Journal of the
Technical University Sofia Plovdiv branch,
Bulgaria, Fundamental Sciences and Applications
Vol. 19, 2013, ICETS TECHSYS 2013.
2. . .,
,
-,
,
2009.
3. ., . , . , .
,
, , 2009.
4. Boldea I, C. Lascu, F. Blaabjerg, A Class of
Speed-Sensorless Sliding-Mode Observers for
High-Performance Induction Motor Drives, IEEE

Trans. on Industrial Electronics, vol. 56, no. 9,
September 2009, pp.3394-340
5. Bose B., Power Eectronics and Motor Drives Advances and Trends, Academic press, 2006.
6. Bose B., Neural Network Applications in Power
Electronics and Motor Drives An Introduction and
Perspective. IEEE Transactions on Industrial
Electronics, Vol. 54, No.1, pp.14-33, 2007.
7. Funahashi K., On the approximate realisation of
continuous mapping by neural networks. Neural
Networks 2, 1989, pp. 183192.
8. Hornik K., M. Stinchcombe, H. White,
Multilayer feedforward networks are universal
approximators, Neural Networks, Volume 2, Issue
5, 1989, pp. 359-366, ISSN 0893-6080,
10.1016/0893-6080(89)90020-8.
9. Mamdani E. H., Application of fuzzy logic to
approximate reasoning using linguistic synthesis,
IEEE Transactions on Computers 26(12): pp. 11821191, 1977.
10. Tarchala G., M. Dybkowski, T. OrlowskaKowalska, Analysis of the chosen speed and flux
estimators for sensorless induction motor drive,
International Symposium on Industrial Electronics
(ISIE), pp. 525-530, 2011.
11. Vas P., Sensorless Vector and Direct Torque
Control. Oxford University Press, 1998.
12. Zadeh L., Fuzzy sets. Information and Control,
8: pp. 338-353, 1965.
13. Zerikat M., Fuzzy-neural networks controllerbased adaptation mechanism for sensorless
induction motor drives. IEEE Transactions on
Industrial Electronics, Vol. 54, No.1, pp.14-33,
2007.

- 1220031-19
09.05.2012
.
4000 Plovdiv
BULGARIA
E-mail: sevil.ahmed@tu-plovdiv.bg
E-mail: ijk@tu-plovdiv.bg
04.03.2013 .
- 103 -

TECHSYS 2013
BULGARIA

2D

, ,
: ,
,
. ,
, - .
: ,
ELECTRONIC MEASUREMENTS OF LIQUIDS

GORAN GORANOV, ISKREN KANDOV, VALERIA AYANOVA
Abstract: Approach which is available for measuring the amount of liquid is combined and is based on
the use of pressure sensors and sensor to measure rpm turbine. The basis of the system is the possibility of
remote sensing data and a more precise measurement.
Key words: water measurement, complex method of measuring liquids
1.

.

.

-
.

.

.
2. .

,
.
,

.[4,5].

( ),
,
.
,

.
- ,
,
.[5,6]

,
, . -

.
- 104 -
.
,
,

. ,

,
.[4,5].
. 1.

.
,
.

, .

,
, -
.
3.

,

. ,

, -
.

.3.

- ,
- .
,
.2
. 3.

. 2.

.
,
,
.

.

, ,
.
(),

.

,
.
CPU ,
.
CPU ,

.
PIC16F874,
8-
- ,
- 105 -
. 4.
,

- 106 -
. [3,4]

, LCD .

. ,

,

,
.[1,2,4,6]

,
,
.
,

.

.

LCD .
,

,
,

.

,
,

.
,

.

, .
4.

,
, -

,
.

,
,
,
.
1. , . -
, --2012.
2. , . -
,- 2007.
3. , . - I,
-2003.
4. Wiley, J. - Sensors and signal conditioning- 2001
.
5. Wilson, J. - Sensor Technology Handbook 2004.
6. www.autoshop101.com/forms/h35.pdf - sensors
and pressure sensing 2013.
Technical UniversityGabrovo,
4 Hadji Dimityr St.
5300 Gabrovo
BULGARIA
E-mail: iskren.kandov@gmail.com
E-mail: val_gb12@mail.bg
01.03.2013 .
- 107 -

TECHSYS 2013
BULGARIA

ABEL
XILINX CPLD XC9500
,
:

.
, ,

.

.
Xilinx XC9500
Advanced Boolean Expression Language (ABEL),

,

. ,
ABEL
-
.
.
: VHDL, VERILOG, ABEL, JHDL, OPENVERA, CPLD, Xilinx
APPLICATION AND CAPABILITY

CAPABILITY OF ABEL
LANGUAGE FOR DESIGNING IN XILINX CPLD
XC9500
GORAN GORANOV, ISKREN KANDOV
Abstract: The use of different languages and approaches in programming is individual for each
programmer. It depends of the programmers skills. In fact the programmers cannot know all possible
languages, but nevertheless can easily teach languages which are closer in structure and logical order.
Advanced Boolean Expression Language (ABEL) is a language used for design in Xilinx XC9500. The
language includes as logical formats for equations and truth tables also and formats for partially state
description with graphs. In addition, ABEL can also be used for description of the test-vector patterns
for inputs and expected outputs. As a structure the ABEL language is similar to assembler.
Key words: VHDL, VERILOG, ABEL, JHDL, OPENVERA, CPLD, Xilinx
1.

(HDL),
(
).
,

. HDL
-

.

, HDL
,

.

1983. Data-I/O ABEL[6].

- 108 -
.
1987.,
VHDL (VHSIC
Hardware Description Language), VHSIC
Very High Speed Integrated.
VHDL Verilog
HDL
,
- HDL.
VHDL Verilog
[5,7]: HDL
.

.

,

.
ABEL
VHDL
Verilog. HDL
PLD
. , ABEL
Xilinx Inc.
2. ABEL
ABEL-HDL
,
.
ABEL

PLDs FPGAs. ABEL-HDL

.
ABEL-HDL
[4]:
- 150 ;
-
(hex 0A),
(hex 0B), (hex 0C);
-
.
,
,
;
-
,
. ,
MODULE
MOD_ULE,
, MOD UL;

,
(keys sensitive);

.
: a - z (
), A - Z ( ), 0 - 9
(), <space}, <tab>, ! @ # $ ? + & * (
) - = + [ ] { } ; : ' " ` \ | , < > . / ^ %.
(),
.
, ,
(1).

, ,
.

.1.
1
ABEL
async_reset,
library
case,
macro
declarations,
module
device,
node
else,
option
enable (obsolete)
pin
end
property
endcase
state
endwith
state_diagram
equations
state_register
external
sync_reset
flag (obsolete)
test_vectors
functional_block
then
fuses
title
goto
trace
if
truth_table
in
when
interface
with
istype

,
, ,
.
.2,
, .
4
X = x3x2x1x0 Y = y3y2y1y0,

8
, .. =X.Y.
- 109 -
,
.
XILINX Xilinx Fundation[1,2],

(
,
, , .) CPLD

XC95108PC84-15 - . 3.
,
Xilinx Fundation .4.

x3-x0 y3-y0 DEC 8x8, BIN 1000
1000. P = DEC 64, BIN
01000000.
.
. 1.
. 2. .

,
.
,

N- . 4 ,

8 .
. 3. .
Module mul4x4
title '4x4 Combinational Multiplier'
X3..X0, Y3..Y0 pin;
P7..P0 pin istype 'com';
P = [P7..P0];
PC1 = Y0 & [0, 0, 0, 0,X3,X2,X1,X0];
PC2 = Y1 & [0, 0, 0,X3,X2,X1,X0, 0];
PC3 = Y2 & [0, 0,X3,X2,X1,X0, 0, 0];
PC4 = Y3 & [0,X3,X2,X1,X0, 0, 0, 0];
equations
P = PC1 + PC2 + PC3 + PC4;
End mul4x4

ABEL.

. 4. - 4
.
- 110 -
3.
ABEL
XILINX Lattice [1].

.

.

.
Xilinx [4]:
( VHDL, ABEL - );

*.jed
.
1. ABEL design manual - Lattice Semiconductor

Corporation, 5555 NE Moore Ct. Hillsboro, OR
97124 (503) 268-8000, March 2003
2. Barbacci, M., Grout S., Lindstrom, G.,
Maloney, M.P. - "Ada as a hardware description
language : an initial report," Carnegie-Mellon
Univ., Dept. of Computer Science, 1984.
3. D. Van den Bout, "The Practical Xilinx
Designers Lab Book 1.5", Prentice Hall, Upper
Saddle River, 1999.
4. Foundation Series 2.1i User Guide , Chapter 5,
Design Methodologies - HDL Flow, available on
the Xilinx website.
5. J. Wakerly, - "Digital Design", 2nd Edition,
Prentice Hall, Upper Saddle River, 2000.
6. J. Mermet - Fundamentals and Standards in
Hardware Description Languages (Springer Verlag,
1993).
7. Using Verilog to Create CPLD Designs
XILINX XApp 143 2001.

Technical UniversityGabrovo,
4 Hadji Dimitar St.
5300 Gabrovo BULGARIA
E-mail: iskren.kandov@gmail.com
01.03.2013 .
- 111 -

TECHSYS 2013
BULGARIA
CNDEP 485

: CNDEP-485,

. /
. ,
.
: , ,
APPLICATION OF CNDEP - 485 PROTOCOL IN

SENSORS NETWORK FOR TEMPERATURE
MEASURING
GRISHA SPASOV
Abstract: The paper presents a custom-developed protocol (CNDEP-485 - Controller Network
Data Extracting Protocol for RS 485) for data access in distributed embedded systems and its
application for temperature measurement in sensor network. The protocol is designed as a part
of Multi-tier Client/Server System for Distributed Measurement and Control. The Message
formats, the protocol vocabulary and the communication rules are described.
Key words: Distributed Embedded Systems, Sensor Networks, Protocol Design and
Implementation
1.

RS-485.

EIA (Electronics Industries Association EIA 485)

TIA (Telecommunications Industry Association),
,
,
[1, 3].
RS-485

( 1200)

[1].

,
( )

1.

- 112 -
,
. . ,
,
,

(master).
(slave) .
32
(
12) [2].
. 1. (a) . (b)
. ()

.

,

[2].
2. CNDEP-485
CNDEP-485
CNDEP
RS-485. CNDEP

(Eternet, WiFi)
TCP/IP
[7, 8], RS485 . CNDEP-485

,
,
(Gateway) 2 [4, 5, 6].

() .

,
8 -
.
2
(VIA EPIA Mini-ITX)
,
master CNDEP-485
-
TCP/IP .
RS-485,
RS-232 (com ) RS
-485, -
.
. 2.
(Gateway).

CNDEP-485,
()
(master )
(slave ).

( 3),
/.
. 3. (a) /. (b)
.
- 113 -
: ;
,
; .

,
, ,
,
, .
3. (b).

ASCII STX (start of text - 02), a
- ETX (end of text - 03).

,
,
, . command-ID
CNDEP
, 0 255 TEST
( OK, ERROR).
1 127 GET ,
128 254 SET .
-

.
,
1.
1. CNDEP-485
Command-ID
Meaning
0
255
TEST
TEST (OK)
TEST (ERROR)
GET
1
2
3
4
5
6
7
8
9
10
11
122-126
127
Get Temperature
Get Humidity
Get Temperature Options
Get Humidity Options
Get Heater Options
Get Cooler Options
Get Image
Get Image Options
Get Lights Options
Get Security Alarm Status
Get Fire Alarm Status
Get User Data
Reserved
128
Reserved
SET
Test:
.
GetTemperature:
.
.
GetHumidity:
.
.
,

. (Master)
,

. 4

.
. 4. CNDEP 485 Master.

3.

RS-485
( 2)

RS-232 <> RS- 485
( )
. 5
RS-232 <> RS485.
RS-232
RS-485,

RS-232 RS-485
.

.
. 5. RS-232 <> RS-485.

-
PIC18F252 TCN75
Microchip 6.
-
USART,
(RS-485<->RS-232)
- 114 -
MAX485.
-
I2C.
-55 +125.
,
,
INT
( D7
).
CNDEP-485
,

.
1. Igor Blai, Peter Draho, THE INDUSTRIAL

COMMUNICATION SYSTEMS PROFIBUS AND
PROFInet, Applied Natural Sciences, 2009, pp.
329 336.
2. DJIEV,
S.
Industrial
Networks
for
Communication and Control, http://anp.tusofia.
. 6.
.

. 7

.
: ,
,
,

.
bg/djiev/PDF%20files/Industrial%20Networks.pdf.
Viewed on July 2009.
3. I. Akyildiz, W. Su, Y. Sankarasubramaniam, and
E. Cayirci, A survey on sensor networks, IEEE
Communication Magazine, Vol. 40, No. 8, August
2002, pp. 102-114.
4. Kwang-il Hwang, Jeongsik In, NhoKyung Park,
Doo-seop Eom, A Design and Implementation of
Wireless Sensor Gateway for Efficient Querying
and Managing through World Wide Web, IEEE
Transactions on Consumer Electronics, Vol. 49, No.
4, NOVEMBER 2003, pp. 1090 1096.
5. Koopman, P., "Embedded system security,"
Computer , vol.37, no.7, pp. 95-97, July 2004.
6. Michael BOROVICKA, Design of a Gateway
for
the
Interconnection
of
Real-Time
Communication
Hierarchies,
Technischen
Universit at Wien, Institut fur Technische
Informatik, Treitlstr, 2003.
7. N. Kakanakov, I. Stankov, M. Shopov, and G.
Spasov, "Controller Network Data Extracting
Protocol design and implementation," Proc.
CompSysTech, 2006, pp.III-A.14-1-6.
. 7.
.
4.
CNDEP485,
,

(Gateway).

,
.

8. N. Kakanakov and M. Shopov, "Evaluating

controller network data extracting protocol for
embedded devices," in Proc. ELECTRONICS, Sept.
2006, Book 3, pp. 15-20, ISBN: 954-438-566-6.
Technologies.
4000 Plovdiv
BULGARIA
E-mail: gvs@tu-plovdiv.bg
05.03.2013 .
- 115 -

TECHSYS 2013
BULGARIA
,
:
.
, .
,
. ,
.
,
.
: , , -
ROBUST CONTROL OF A QUAD-ROTOR

HRISTO GENDOV, PETKO PETKOV
Abstract: In this work, the design of a robust control system for attitude stabilization of a
miniature quad-rotor is presented. A linearized model of the quad-rotor with input
multiplicative uncertainty is derived for which a two-degree-of-freedom -controller is
determined. The closed-loop frequency responses confirm that the desired performance of the
system is achieved. A 3D PID-controller is used to achieve the desired position of the quadrotor in the space. Simulation results for the nonlinear closed-loop system are shown which
confirm the efficient work of the robust controller under the presence of disturbance torques
along the three rotation quad-rotor axes.
Key words: quad-rotor, robust control, -synthesis
1.

,

.

,

,
(
, ..).

,

,
.
,

,
. [1], [2], [3], [5], [6]. . 1
- 116 -
, .
.
=
=
+ sin
tan
= cos
'() *
+ /
+ !/
(2)
+ "/
+ cos
sin
tan
(3)
+,' *
% = &+,' - . + & +,' - .
. 1.

.
,
.
,

.
-,

.

.
2.

. -, ,

,
-,
-.
=
=
+ sin
+ cos
sin
cos
cos
+ / (1)
+ /
u, v, w
X, Y Z
, p, q r
, , , , X, Y, Z
, L, M, N
, Ixx, Iyy, Izz
, m .

/ = /012 =
012
012
Xfus, Yfus, Zfus

,
3 = 34 + 35 + 36 + 37
(4)
.

,
34 = 8945 , 35 = 8955 , 36 = 8965 , 37 = 8975 (5)
b ,
,
. ,

, ..
;4 = <= 945 , ;5 = <= 955 ;6 = <= 965 ;7 = <7 975 (6)

-
,

,
94 = > ?4 95 = > ?5
96 = > ?6 97 = > ?7
(7)
- 117 -

-

.

= @AB 94 95 + 96 97
+< 35 + 37 + !CD
(8)
! = @AB 94 95 + 96 97
+< 34 36 + !CE
" = ;4 + ;5 ;6 + ;7 + !CF
@AB
, d

, a !CD , !CG , !CF
.
,
Simulink
linmod MATLAB.

,
.

,

.

.
9-
=
=
+IJ K O
<LMN
= P = +QJ K O
<LMN
=H
= [ , , , , , , , , %]
K = [?4 , ?5 , ?6 , ?7 ]
<LMN = [!CD , !CG , !CF ]
A, B, C, D
.
(9)

=
M = T=1UC M V
=T M J K O
<LMN
(11)
G = [Gu Gd ] , (11)

= T1
+ TC <LMN
(12)
Gu Gd
.
3. -
(9)

(1 + YZ ), |YZ | < 0.1, L = 1, . . . ,4

10 %.

T M = T=1UC ` + <La , 6
= <La Y4 , Y5 , Y6 , Y7
.

M
W (10)
<LMN M
K

T=1UC M H, I, P, Q.

. (10)
.

. 2.

,
, . 2.

K ,

M
dist
.
,

! = <La
4 , c 5, c 6
(13)
- 118 -
=c
=c
= 6d 2d e56g.`2e4 (14)
ch M =
4gjd 2e4
0.9 4gjd 2e4gjk 66
Singular value plot of the output sensitivity function
(15)
,

,
.

5.4gjn 2e4
c1 M = 2.10m6 6.4gjo 2e4
50
0
Singular Values (dB)
(

). . 4 ,

35 dB, .., 50 .
50
77
(16)
,
.
-
dksyn
MATLAB.

0.740,
..,
.
100
2
10
10
10
10
10
10
Frequency (rad/s)
. 4.

Singular value plot of the input sensitivity function
80
70
60
4.
50
Closedloop singular value plot
20
40
30
20
0
10
0
20
10
40
20
2
10
10
10
10
10
10
Frequency (rad/s)
. 5.
60
80
100
2
10
10
10
Frequency (rad/s)
10
10
. 3.

. 3

( )

10
rad/s,
.
. 4 5

5.
,
,

,

.

. ,

Vx ,V y ,Vz ,[m / s ]

- 119 -
pD = qrM qrM%
pG = qrM%
pF = qrM qrM
(17)
+ qrM qrM
(18)
[ / M 5 ]
.
Z ,

3 ["]
3 .
Mdz Z

0.08 Nm
0.025 Hz. ,
,
,

( ).

ode23tb
.
-3
(19)
(19) (18) ,
/, ,

. (17)

?F = u>h4 v
?D = u>h5 v
@w0
(rad)
-1
-2
-3
x >C4 pF y / qrM qrM
-4
@w0 x >C5 pD y / qrM qrM% (20)
?G = u>h6 v
@w0
Roll angle
x 10
10
20
30
40
2.5
, >h4 , >h5 , >h6
>C4 , >C5 , >C6
. (20)

,

/, , .
?F , ?D , ?G
-.
(20)
>h4 = 0.4, >h5 = 0.1, >h6 = 0.3
>C4 = 0.3, >C5 = 0.5, >C6 = 0.6
,
.
6.

-
Simulink
.
X Y
!CD = 0.1 sin 0.2N , !CG = 0.05sin 0.16N
70
80
90
100
Pitch angle
x 10
2
1.5
1
(rad)
xref , yref , zref
60
. 6.
x >C6 pG y/qrM
-3
50
t (sec)
0.5
0
-0.5
-1
-1.5
-2
10
20
30
40
50
t (sec)
60
70
80
90
100
. 7.
. 6, 7 8
,

5 m. ,
,
. 4,
50
.
.
,

,
.
- 120 -
-3
1.5
Yaw angle
x 10
1. Boubdallah, S., Siegwart R. Design and control

of a miniature quadrotor. In: Valavanis, K.P. (Ed.):
Advances in Unmanned Aerial Vehicles, pp. 171210. Springer , 2007.
(rad)
0.5
2. Kim, J., Kang, M.-S., Park, S. Accurate

modeling and robust hovering control for a quadrotor VTOL aircraft. J Intell Robot Syst, Vol. 57,
2010, 9-26.
-0.5
-1
-1.5
10
20
30
40
50
t (sec)
60
70
80
90
100
. 8.
3. Nonami, K., Kendoul, F., Suzuki, S., Wang,

W., Nakazawa, D. Autonomous Flying Robots.
Unmanned Aerial Vehicles and Micro Aerial
Vehicles. Springer, Tokyo , 2010.
X-Y motion
0.2
4. Gu, D.-W., Petkov, P.Hr., Konstantinov, M.M.

Robust Control Design with MATLAB .
Springer-Verlag, London , 2005.
0.15
0.1
5. Shakev, N.G., Topalov, A.V., Kaynak, O.,

Shiev, K.B. Comparative results on stabilization of
the quad-rotor rotorcraft using bounded feedback
controllers. J Intell Robot Syst, Vol. 65, 2012, 389408.
y (m)
0.05
-0.05
-0.1
-0.15
-0.2
-0.15
-0.1
-0.05
0.05
0.1
0.15
x (m)
6. Yu, Y., Ding, X. A quadrotor test bench for six

degree of freedom flight. J. Intell Robot Syst, Vol.
68, 2012, 323-338.
. 9.

. 9

.
0.2 m
.
6.

.

,

.

.
. ,
,
1210085-08
.
,
.
Department of Systems and Control

Technical University of Sofia,
8 Kliment Ohridski Blvd.
1000 Sofia
BULGARIA
E-mail: h_gendov@tu-sofia.bg
E-mail: php@tu-sofia.bg
12.03.2013 .
- 121 -

TECHSYS 2013
BULGARIA

:

Simens . -

.
: , , ,
HARMONICS AT RECUPERATION
ILKO TARPOV
Abstract: In the report are considered and analyzed harmonics of current and voltage at
recuperation of the Simens company`s electric motor trains "Desiro" operated in the park of the
Bulgarian Railways. Also, there are some of the most important results of the measuring and
recommendations to optimize electricity costs in the company.
Key words: recovery, harmonic analysis, BDZ, power factor
1.
2008 .

.

.

()
.
,
.
.

.

.
Uh
Zh
Ih (1).
U h Zh .I h
Uh
;
Zh -
;
Ih - .
(1)
- 122 -
,
.

:
-
;
-
;
- ;
-
.
, 10%
U 18%

5% ,
6% [2].

-

"".
2.

, , ,
01.2012 .
. 01.2013 .

UMG604 JANITZA
Elektrik.

40 .

,
.

()

.

Gridvis.
, 60 .

.
(PF), -,
.
power
factor (PF) (2).
KM
P
cos 1
S
(2)
;
S - ;
-
;
1-
.
os1 ,

,

. . 1

os1.
. 1. os1

- .
1.
I1
I1
2
1
I I 2
(3)
I ;
I1 -
;
I -
.
I
:
I I a 2 I r 2 I 2
Ia ;
(4)
- 123 -
Ir - ;
I - .
I a I1 cos
P
, I r I1 sin
U
(5)
cos

,
,
.
- cos, ,
.

,
- ,

1.

,
,
1 [1].
. 2
100 0 km/h
33 .
(7). ,

.
I h % 100
I
U, ,
50 Hz
-
. -

(h),
:
I
I
h 1
2
h
U
h 1
2
h
U h % 100
Uh
U1
(7)
I U

. 3.
. 3. U
I

Imax Umax
-
- 2 , ,
.. .
-.
I
. 2.
Ih
,
I1
I max ,
2
U max
2
(8)
Imax Umax
;
I U
.
1,5 2,

5. . 4.
(6)
Ih Uh
.

h,
. 4. -(-) (-)
,
-
- 124 -
4 5
5 .

,
.

,

. ,

.

..
D.
. . 5
1 THDi
.

(11).
2
I
THDi 1 ,
I1
I1
1
I
1 THDi2
(11)
(3) (2)
(12).
KM
cos 1
1 THDi2
(12)
(9)
(12)
-
THD.
.

.

THD (Total Harmonic
Distortion),
(10),
-
1.
3.
,

.

.
,

D S (P Q )
THDi
I
h2
I1
2
h
, THDu
U
h2
2
h
U1
(10)

.
%.
1. .

,
. ,
, 2006.
2. Schneider Electric
.
. 5. THDi

2 5%
250%
VTU "T.Kableshkov" Sofia,

158 Geo Milev St.
1574 Sofia
BULGARIA
E-mail: stsb_plovdiv@abv.bg
15.02.2013 .

TECHSYS 2013
BULGARIA

,
:

.

.

.
. .
: , , .
STUDY ON ELECTROMAGNETIC
COMPATIBILITY OF INDUCTION DRIVES
WITH SINUSOIDAL FILTERS IN MATLAB
ENVIRONMENT
IVAN KOSTOV, GEORGI IVANOV
Abstract: The article shows the results of the evaluation of the harmonic composition of the
output voltage of the frequency converter in the frequency controlled induction motor drives.
Obtained and analyzed the harmonic composition of the three-phase AC voltage frequency
converter with sinusoidal PWM and power sine-wave filter. Accuracy of the results is confirmed
with simulation verification procedures by synthesis and experimental way. Experimentally
obtained results were compared with results from simulations. Here are directions for future
research.
Key words: induction, motor, sine-wave filters, EMC.
1.
(),
()
,

[3,4,6].

[5]. ,
,
.

.
- 126
,
15kV/ms 2kA/ms. -
, -

. -
-
, .

,

.
. ,
, EN55022,
CISPR22.
IEEE519,

, ,

.

,

.
,
.

:
- ;
-
;
- ;
;
-
;
- ;
- ;
- ;
- .

, :
-
,
,
;
- -
,
(5-30%) ;
-
.

5 kHz,
,

50 Hz 120 Hz.

.

[3],

.
.
2.
()
. 1,

,
SF,
NI6809
[1].

,
.
,
fc = 5 kHz
UDC = 310V.
: RL =
= 0.15 ; L= 30 mH; C= 1 F.

124 V 50 Hz. A PN = 750 W;
UN = 380 V; f = 50 Hz; Zp= 2
-
: Rs = 10.2 ; Ls = 0.061 H; Rr = 10.52 ;
Lr = 0.061 H; Lm = 0.457 H.

- 127 MATLAB/Simulink
13- .
. 1.

.

10 s.

[2]:
u( t ) =
= 13
Um sin(t + )
3.
.

.
(250 s
)

13- .

f = 50 Hz
.
SF

(1) (3),

, [1].
3.1.
.2

-
. .2
.
(1)
=1
Um = U A2 m + U B2m
= arctg (U Am U Bm )
(2)
(3)

[2]
(4) (5):
kTHD = U 11
k HVF = U 11
U2
(4)
=2
(5)
=2
:
- U Am ,U Bm - [1];
- Um - - ; -
- ; = 2f -
; U
; U1
.
. 2.

(); ().
U,V; t,s
- 128
: KTHD =
0.0236 KHVF = 0.0099.
,
f = 60 Hz.
3.2. .
.3,

, .
. 4. a
SF
(); (). U,V; t,s

SF: KTHD = 0.02796; KHVF =
0.011283.
. 3.

SF ();
(). U,V; t,s
4.2.
f = 100 Hz.
.5,

,
f = 100 Hz.
4.

.
4.1.
f = 60 Hz.
.4,

SF: KTHD = 0.022021; KHVF =
0.008861.
- 129 ,
f = 120 Hz.

SF: KTHD = 0.158529; KHVF =
0.070854.
. 5. a
SF
(); (). U,V; t,s

SF: KTHD = 0.056453; KHVF =
0.023466.
4.3.
f = 120 Hz.
. 6. a
SF
(); (). U,V; t,s
.6,

5.

,
.

.

.

.

,
.

,
,

.

,
[1],

, 100 Hz.

,
- 100 Hz.

,
,

.

, -

.

, :
- 130 -

;
-

.

(
)
;
- (
,
,
);
- ,
, ;
-

.

-
1220031-19 09.05.2012 .

.
1. ., . , . , .
,

,
,
- , .9, , 2012 .,
. 136-140.
2. ., . ,

,
2012, , 1-4 , 2012, ,
, -, 62,
2, 2012 48, 429-437, ISSN
1311-0829.
3. Das J., M. Dekker, Power System Analysis:
Short Circuit Analysis and Harmonics, New York,
2002.
4. Lundquist J., On Harmonic Distortion in Power
Systems,
Department
of
Electric
Power
Engineering, Chalmers University of Technology,
Sweden, 2001.
5. Oraee H., S. Filizadeh, The Impact of Harmonic
Order on Insulation Aging in Electric Motors, 36th
Universities Power Engineering Conference, UPEC
2001, UK, Sept. 2001.
6. Watson N. R.,
B. C. Smith,
J. Arrillaga,
A. R. Wood, Power System Harmonic Analysis,
University of Canterbury, Christchurch, New
Zealand, John Wiley & Sons, New York, 2000,
ISBN 0 471 97548 6.
Control Systems Department
4000 Plovdiv
BULGARIA
E-mail: ijk@tu-plovdiv.bg
11.02.2013 .
- 131 -

TECHSYS 2013
BULGARIA

3D 5
.
,
:
3D 5
. ,
,
x, y z .
, X, Y, Z,
.
: , 3D ,
MATHEMATICAL MODEL OF COORDINATE

TRANSFORMATION FOR 3D SCANNER WITH 5
DEGREE OF FREEDOM
KRASIMIR KRASTEV, RAYCHO ILARIONOV
Abstract: This article proposes a mathematical model of coordinate transformation for
automatic 3D scanner with 5 degree of freedom and laser distance sensor. As result from this
transformation is obtained point cloud data, where each point is represented in the Cartesian
coordinate system as combination of x, y and z absolute values. Input data are the scanned
points from the object surface, as each point is presented in scanner coordinate system - the
position of X,Y,Z, A and B driven axes and measured distance to the object surface.
Key words: coordinate transformations, 3D laser scanner, point cloud
1.
,
, 3D
.

,
.

x, y, z.
P
[4]
(1).
x1 y1
x y
2
P 2
x n yn
z1
z 2
zn
(1)
- 132 -
3D
,

.
3D

[2,5].

[3,6].

3D
5
()
Wolf&Beck.
2.
3D . 1,
5 .

X, Y, Z - A B .
5
,

.

()
Wolf&Beck,
[1].

,

.

.
X, Y Z
,

.
.

.
X.

Ps (2),
3D .

5-

.
X 1 Y1 Z1 A1 B1 l1
X Y Z A B l
2
2
2
2
Ps 2 2
X n Yn Z n An Bn ln
(2)
Xi, Yi, Zi, Ai Bi

, i=1n. l
e a,
b c.
li
(3).
li ai bi ci
(3)
,
X, Y, Z
Ps(4), P(1).
X 1 Y1
X Y
Ps ' 2 2
X n Yn
. 1. 3D
3D

.
Z1
Z 2
Zn
(4)

,
[inc],

[mm] (5).
Xi
X ai
Rt
(5)
- 133 -
R
, t e
, Xai
.

Y Z.
,
,
X
, . 2.

,
. i

(6).
Ai 360
Ri
(6)
R
, Aai

(i=1n), i
.
. 2.

li

, i X.
Pl Ps
x,y z

.

Y
Y (. 3. . 4.).
Y
.
,

i
- ri
.

x y
z ,
Z
(4) Z(7).
. 3. - Y

x z,
X
Z Pl (7), Y=const.
X '1 0 Z '1 l1 cos1 0 l1 sin 1
X ' 0 Z ' l cos 0 l sin
2
2
2
2
(7)
Pl ' 2
2
X 'n 0 Z 'n ln cos n 0 ln sin n
. 4. - Y
- 134 -

(x, y, z)
P (8).
r1 cos 1 r1 sin 1 Z1 l1 sin 1
r2 cos 2 r2 sin 2 Z1 l2 sin 2
(8)
P
rn cos n rn sin n Z n ln sin n

i
( ), ri

(i=1n).
i
(9),

.
Bai 360
i
Ri
(9)
R
, Bai

(i=1n) , i
.

Y ( ) ri
,
x1 y1
x y
2
P 2
x n yn
ri X i X 'i X i li cos i
( X i li cos i
(12),
Yi
(13)
(10)
, Y
(. 4.), ri
. ri
OXY
(11).
ri ( X i li cos i ) 2 Yi 2
(11)
ri (11)

- Y
.
Y, ri
i .
i (13).

(8)
i i,

.

3D
(12)
.
j1
j1
r1 cos(1 arctan Y ) r1 sin(1 arctan Y ) Z1 l1 sin 1

1
1
z1
j
j
z 2 r2 cos( 2 arctan 2 ) r2 sin( 2 arctan 2 ) Z 2 l2 sin 2
Y2
Y2

zn
j
j
r cos( arctan n ) r sin( arctan n ) Z l sin
n
n
n
n
n
n
n
Yn
Yn

.

(Xi)
Xi(i=1n).
Y.
i arctan
X (Xi)
Xi.
(12)

jn.
ji X i li cosi
(14)

(12)
3D (Xi, Yi, Zi, Ai, Bi, li)
(x, y,
z)
,
- 135 -
,

3D (X, Y, Z, A B)
(l).

,
(x, y z)
.
A
(12)
. 5.

.
,

3D

.
(pt),
(
)
.
.

X,Y Z , .

,
.

l
, r

(x, y z)
(12).

.
.

(n
).

.
,
,

.
. 5.
- 136 -
3.

3D
.
,
3D .

,
.

, *.xyz, *.pcd,
*.ptx, 3D
(PolyWorks,
GeoMatic, MeshLab ..).

,

,
, .

,
,
.

().
1. Dr. Wolf&Beck GmbH. Integration Handbook

for Optoelectronic Distance Measuring System,
2005.
2. Ilarionov, R., Model approach in the design of
device for 3D information input into computing
environment, CompSyst Tech 09, ACM, New York,
USA
3. Shell, R., Hall, E. (2000). Handbook of
Industrial Automation, Marcel Dekker Inc, New
York, USA.
4. http://renderman.pixar.com/view/point-clouds
5. http://www.freepatentsonline.com/8249823.pdf
6. http://ocw.mit.edu/courses/aeronautics-andastronautics/16-07-dynamics-fall-2009/lecturenotes/MIT16_07F09_Lec03.pdf

Technical University Gabrovo
4 H. Dimitar St.
5300 Gabrovo
BULGARIA
E-mail: krasimir.krustev@yahoo.com
E-mail: ilar@tugab.bg
14.02.2013 .
- 137 -

TECHSYS 2013
BULGARIA

.

:
,
. ,
,
.
.
, 3D .
: 3D , ,
EXPERIMENTAL MODEL OF LASER SCANNING

HEAD WORKING BY THE METHOD OF LASER
SECTION
KRASIMIR KRASTEV
Abstract: This paper proposes an experimental model of laser scanning head, working on the
method of laser section, as a variety of laser triangulation. Over the scanned object, laser light
is projected in the form of a thin line, the reflected light is captured by a digital camera, and is
separated from the rest of the image. Then the coordinates of the object points from the
intersection plan are calculated. The developed model will serve as a basis for studying the
processes of 3D laser scanning.
Key words: 3D laser scanning, laser triangulation, laser section
1.
3D CAD
,
3D [1].
3D
,
.

,

.

.

3D .
,
,
-
- 138 -
.
,

.

(CCD/CMOS).
,

[5].
().
3D
.

.

3D .

.

:
- 1. CCD QuickCam PRO 9000;
- 1. LC650-5-5-F;
2.
2.1.
-
10 m (. 1.).
,
.

( ),
.

.

. 1.
. 1.

C 1
F.
( C C ),
F F .

0,01% .

.

,

.
2.2

.

, ,
.
. 2.

.
. 2. ;
,

.

.

, ,
- 139 -

.

(),
.
( ),
(CCD ).

[3].

(n
) .

.

[2].

.
,

.

,
.
3D
,

( )
.

3D
.3.
(x,y,z) 3D
,
(x, y),
(1), (2) (3), [1].
. 3.
.
x=
b x'
f cot x '
(1)
y=
b y'
f cot x '
(2)
z=
bf
f cot x '
(3)
, b (),
,
f .

, .

.
a
,

(. 4.)
[2].
. 4.

2.3.
,

(. 5.).
,

( ). ,
.
CCD () Logitech
QuickCam Pro 9000, a LC650-55-F, .
- 140 -
. 5.

3D

.
:
I.
RGB - GREY SCALE
THRESHOLD (
).
- ,
,
.
,
.
:

,
.

,
(RGB =
0,0,0).

, ,
Pix 0,0,0
.
III. .

,
THRESHOLD (. 6.).

e,
.
- CV_THRESH_BINARY -
- CV_THRESH_BINARY_INV -
- CV_THRESH_TRUNC -
- CV_THRESH_TOZERO -
- CV_THRESH_TOZERO_INV II.
. 6.

(1), (2), (3),

..

.

,

- 141 -
.

QT Creator MinGW Windows
. C++ .

OpenCV (Open Source Computer Vision).
[4] :
- IplImage* cvCreateImage(CvSize size, int depth,
int channels); - ;
3D
(1), (2), (3).

,

.
.
- int cvGrabFrame(capture);
;
- IplImage* cvRetrieveFrame( CvCapture* capture
); - ;
- void cvThreshold( const CvArr* src, CvArr* dst,
double
threshold,
double
maxValue,
int
thresholdType ); -
;
- void cvShowImage( const char* name, const
CvArr* image ); -
;
- cvReleaseCapture(&capture); - ;
GUI
(. 7.)

:
/
;
(sliders)

;

;

, b
(mm), ()
f (mm);

.
.

( ).

,

. 7.
2.3
a

:
- THRESHOLD;
-
;
-

THRESHOLD;

(THRESHOLD)
.

, R
(
), G B
.
( )
THRESHOLD
- .
. 10. . 8

, . 9
.

240
250.

640480dpi.
- 142 -
. 8. .
.

,

(
).
, .
o
,

.
4.

,
.
. 9. .
5.

1101/2011 -.
. 10.
3.

,
h
( z )
,

.

-

Binary Threshold.
Binary Threshold 240
250
.

.
3D

1. Peiravi, A., Taabbodi B., A Reliable 3D Laser

Triangulation-based Scanner with a New Simple but
Accurate Procedure for Finding Scanner
Parameters, Journal of American Science 2010;6(5)
2. Schfter + Kirchhoff. Applications: Measuring
and Process Control in 3D using Laser Lines, Laser
Spots and Laser Patterns, 2011.
3. http://www.lmi3d.com/education/linetriangulation
4. http://docs.opencv.org/
5. https://sites.google.com/site/todddanko/home/we
bcam_laser_ranger

Technical UniversityGabrovo
4 H. Dimitar St.
5300 Gabrovo
BULGARIA
E-mail: krasimir.krustev@yahoo.com
14.02.2013 .
- 143 -
- 144 -
- 145 -
- 146 -
- 147 -
- 148 -
16.02.2013 .
- 149 -

TECHSYS 2013
BULGARIA

POWOT

:
,
.
Powot.
.
Powot scale frequency

. .
: , , ,
, Powot Analyzer.
EMBEDDED SYSTEM FOR CONSUMPTION INVESTIGATION

OF FUNCTION CALLS WITH THE OPTIMIZATION PROGRAM
POWOT
LUBOMIR BOGDANOV
Abstract: This paper covers the development of a device that measures the current
consumption and the time for execution of the function calls in a microcontroller executing
bare-metal firmware. This is a specialized device and is developed for a tool called Powot and
its optimization needs. The data from the measurements provide valuable information for the
execution of the firmware. Because of it the Powot tool can apply the scale frequency rule
and can optimize the firmware of the target microcontroller in terms of clock frequency. The
latter will lead to optimization of the power consumption.
Key words: measurements, function calls, frequency optimization, power optimization, Powot
Analyzer.
1. Introduction
Power consumption of the embedded
devices has always been an important parameter.
Reduction in the consumption of a microcontroller
can be achieved with advancements in the
technology or with hardware and software methods.
A program, called Powot (Power Optimization
Tool), is being developed [2] that tries to optimize
the power consumption by modifying the firmware
of a microcontroller. This tool uses various methods

(also called rules) to achieve this and many of them
depend on the hardware capabilities of the
microcontroller. Nevertheless software methods
also exist and they dont depend on hardware. The
most obvious one is the dynamic changing of the
microprocessors clock. Whenever possible, a low
frequency could be used which will result in lower
power consumption. This rule is referred to as
- 150 -
scale frequency in the Powot tool. Almost every

microcontroller nowadays has the ability to change
its frequency during program execution and
therefore this feature can be used to optimize the
power performance of a microcontroller. This has to
be done solely by the software developer that must
keep in mind the syntax for the frequency
adjustment as well as the right place in code where
it has to be done. The Powot tool tries to simplify
this process by automatically placing the proper
API calls in firmware to adjust frequency. With the
implementation of this feature an important
question arises where do I put the frequency
adjustments and whats the value of the needed
frequency? To solve this problem a certain amount
of information is needed thats the information
about the power consumption of each function call
in the firmware. There are two approaches to get
this data: a software approach where firmware is
decomposed into assembly instructions and a power
model of these instructions is used to predict the
overall consumption of the function call; a
hardware approach where power measurements are
carried out while the firmware is executing. This
paper discusses the second approach.
To measure the consumption of each
function call a special data acquisition (DAQ)
system has to be developed. It has to measure the
current consumption of the microcontroller
synchronously to an external signal asserted before
and after each function call. Every DAQ nowadays,
as well as most of the digital oscilloscopes, can
achieve these results without any effort but the new
idea is to develop a device that is specifically suited
for the power measurements of a single-core
microcontroller. This device requires interaction
with the Powot tool, therefore it will be called
Powot Analyzer.
2. Hardware design of the Powot Analyzer.
The needed data acquisition system is fairly
simple and the absolute accuracy of the device is
not strict since only the power consumption ratio
between the separate function calls is of big
importance. The Powot tool will try to optimize the
function calls power that is distinctively higher
than the others.
In Fig. 1 a block diagram of the device is
shown. The development process is described next.
uC selection. Currently the Powot tool is
being developed using microcontrollers mainly
suited to run bare-metal firmware or simple
embedded operating systems and RTOS. The
resource of these controllers is humble
microprocessor clock of 40 120 MHz, 8 128 kB
of SRAM and 128 512 kB of Flash. Most of the
tests are carried out on Texas Instruments Stellaris
LM3S microcontroller family with chips that have

no more than 50-megahertz Cortex-M3 cores. For
the Powot Analyzer a slightly better microcontroller
will be chosen the LM4F232 with an 80megahertz Cortex-M4 core.
ADC selection. Obviously the most important
part in the design is the selection of the analog-todigital converter. As stated above, the absolute
accuracy is modest but the differential nonlinearity
(DNL) is crucial. The easiest way of hardware point
Linear
Voltage
Regulator
3,3 V
Shunt Resistor
Linear
Voltage
Regulator
Rectifier and
Filter
Linear
Voltage
Regulator
Rectifier and
Filter
Differential
Amplifier
Vout1
Lin. Volt.
Regulator
3,3 V or 2,5
V
Shunt Resistor
Mains
Differential
Amplifier
Vout2
Linear
Voltage
Regulator
+Voa
-Voa
3,3 /
2,5 V
select
Mains
Transformer
Rectifier and
Filter
Linear
Voltage
Regulator
Vin2
Linear
Voltage
Regulator
Stellaris
LM4F232
Vin1
Rectifier and
Filter
RS232
Translator
To PC
Fig. 1. Powot Analyzer high-level block diagram.

of view would be to use the LM4F232 built-in ADC
[6]. It is a 12-bit, 1-MSPS ADC with an internal
reference of 3 V. The DNL is 1 LSB max. Using
this data we can calculate the following
parameters[1]:
VLSB
VADCref
n
2 1
3V
0,733mV
2 1
12
VDNLerr 1LSB 0,733mV
(1)
(2)
Tmin Tsample Tproc 1s 0,2s 1,2s (3)

Equation (3) shows that a function call lasting
less than 1 s cannot be measured by the Powot
Analyzer. This number can be increased to 1,1 1,2
s to reflect the time of the LM4F232 to process the
measured data (Tproc).
Differential amplifier construction. The
differential amplifier will be built using discrete
components and an operational amplifier. The most
important criteria that must be met is the wide
bandwidth and the small DC errors due to
temperature drifts in certain parameters. The
- 151 -
LM7171 is suited for this purpose with its 200 MHz

unity-gain bandwidth and input offset voltage drift
VOS = 35 V / C.
Shunt resistor selection. The maximum
current consumption of the tested microcontrollers
is typically in the range of 100 200 mA, therefore
the shunt resistor must be chosen to produce a 3volt drop at the maximum current. The current is
increased, with a reserve, to 300 mA. We can
calculate the resistors value and its power rating
using:
Rsh
VADCref
I max
3V
10
0,3 A
2
P I max
.Rsh 0,32.10 0,9W
(4)
(5)
If needed, the value of the resistor can be

changed for other current ranges which will result in
small modifications of the software. The current
range must not exceed the mains transformer
capabilities.
The Powot Analyzer in general is built as a
two channel DAQ to provide the ability to measure
separately the current consumption of the whole
microcontroller as well as the microcontrollers
core. Some microcontrollers provide an extra pin to
power the core which will allow separate
measurements. If such measurements are
performed, the core usually works with a lower
voltage than the microcontroller. That is why the
second voltage regulator has the ability to change its
output value to 2,5 V (which is the core voltage for
the LM3S family, Fig. 1).
The additional voltage regulators before the
shunt resistor are used to reduce the CMRR error as
much as possible. All regulators are linear to avoid
noise interference, regardless of the increased
power consumption. Two regulators provide the
supply voltages +Voa and Voa for the operational
amplifiers and one for the LM4F232. A link with
a PC, running Powot, is made through the RS232
interface.
3. Firmware design of the Powot Analyzer.
The firmware of the Powot Analyzer must
provide the following two measurements for each
function call:
- current consumption;
- time for execution.
These power measurements assume that the
supply voltage of the tested microcontroller remains
unchanged during the execution of the program.
Now a description of the main blocks of the
firmware follows.
Initialization. The LM4F232 controller

initializes its UART with 115200 bauds, 8-bit
words, no parity check, 1 stop bit and no hardware
or software flow control. The core frequency is set
to maximum (80 MHz) to provide more accurate
time measurements. The ADC module is turned on
and no hardware averaging is used to provide the
best possible time resolution of the Powot Analyzer
- approximately the time of the sampling (1 s). A
pin is configured as an input and is used to trigger
an interrupt on each start and end of a given
function call. The device under test must be
configured to set and clear a pin on each entry and
exit of a function call. This behavior is programmed
by the Powot tool.
Awaiting commands. The analyzer is put to
sleep until a command is received from the UART
and an interrupt is triggered. The command is sent
automatically from the PC running the Powot tool.
Process command. Once received, the
command is processed by the LM4F232. The most
important commands are:
- START_MEASUREMENT
- GET_DATA.
The START_MEASUREMENT format of
the command is shown in Fig. 2. This command has
Byte 0
Byte 1
Byte 2
Byte 3
Byte 4
Byte 5
N+3
Least Significant Byte
ADC Channel
Number
Number Of Functions
Operation code
N
Most Significant Byte
N+1
N+2
Fig. 2. Packet format for the start measurements

command.
two parameters an integer representing the
expected number of function calls and a byte
naming the ADC channel number. The first
parameter tells the analyzer to prepare itself for
Number-Of-Functions synchronizing impulses
and the second the ADC channel on which the
measurements must be performed. When all of the
synchro impulses are received the measurement
stops and a DONE command is sent over the
UART.
The GET_DATA command invokes an
automatic sending of all the available data of the
measurements. Currently this data is limited to 100
measurements because the Powot Analyzer
accumulates up to 4096 current samples (32-bit
each, making it 16 kB) and calculates their average
value on return from the function call. The SRAM
is only 32 kB and care should be taken not to
exceed it, if their number is increased in the future.
The data format sent from the Analyzer to
the PC is shown in Fig. 3. The command number is
- 152 -
the enum value SAMPLE and must be prepended to

each set of measurements. This is needed for
compliance with the Powot Analyzer PC driver. It is
an event driven driver and each packet transmission
8-bit
32-bit
32-bit
32-bit
Command
number
Function
Number N
Current
Consumption
Elapsed
Time
Command
number
Function
Number N+1
Current
Consumption
Elapsed
Time
Command
number
Function
Number N+2
Current
Consumption
Elapsed
Time
Fig. 3. Format of the packet series sent from the

Powot Analyzer to the host PC.
must be marked with the SAMPLE command in
order to work correctly. Afterward a 32-bit number
is sent that represents the function call number. The
Powot tool has a look-up table that helps it make the
link between the function call number and the
function call prototype. When all of the data has
been transmitted a DONE command is sent. This
ends the measurements process and the Powot tool
can continue its execution on the host PC.
4. Powot Analyzer PC driver.
The whole process of establishing
connection with the Powot Analyzer, sending a
request for the measurements start, waiting for the
measurements to be performed, uploading the
measurements data to the PC is controlled by the
Powot Analyzer PC driver.
It is a driver written in C++ using the Qt
framework and the third-party low-level library
QExtSerialPort[3]. It is compiled as a static library
for Linux OS (with *.a prefix) and thanks to the Qt
Creator it can be run on Windows as well (Qt
Creator is a cross-platform IDE [4]). The driver
does the needed steps automatically which makes it
convenient to use in other projects. The driver itself
is separated in two parts:
- powotAnalyzer is a high-level class that
implements a black box taking input
parameters and returning output data;
- UARTCon and UARTrecv re low-level
classes that implement the UART
connection using the third-party
QExtSerialPort library.
The above-described structure is depicted in Fig. 4.
The work of the driver starts with the initialization
settings of the UART that are passed as arguments
to the constructor. Then a callback slot (function)
must be registered that is called when the execution
of the driver finishes. To do this a separate event
loop must be started before the construction of the
powotAnalyzer object. This is done with the help of

Qt Creators class QEventLoop[5]. Next, the needed
data for the measurements is sent between the
Powot class and the driver through the method:
void powotAnalyzer::input(
QString *PathToBinary,
int
NumOfFunctions,
QString *flash_string
);
Powot Analyzer PC driver
class powot
class powotAnalzyer
class UARTcon
class UARTrecv
Device Firmware
Fig. 4. A PC driver for communication with the

Powot Analyzer data acquisition system.
The first argument provides the path to the modified
target firmware that implements the setting and
clearing of a pin before and after each function call.
For example the Powot tool will transform the
following program
void main(void)
{
function_call_one();
function_call_two();
}
into the modified program below
void main(void)
{
set_pin_high();
function_call_one();
set_pin_low();
set_pin_high();
function_call_two();
set_pin_low();
}
The syntax for the pin toggling is read from a
configuration file. Afterward Powot will compile
the program, count the number of functions and
hand over the control to the Powot Analyzer driver.
- 153 -
The flash string is provided by the user in a

configuration file. The measurements are started by
calling the method:
Host PC
running
Powot
void powotAnalyzer::startMeasurements();
and afterward the program waits the measured data
transferred to the Powot with the help of the
registered callback slot.
A file called calib_constants.cfg contains
calibration information that is read by the driver and
the measured data is corrected according to the
values in the file. The file has the following
structure:
CURRENT_ADD +96
TIME_MULT 0.0000000125
DIFFAMP_RATIO 10
ADC_RESOLUTION 4096
ADC_REF 3.0
The CURRENT_ADD variable reflects any DC
error produced by the resistor tolerance, operational
amplifier offset voltages, bias currents and ADC
input errors. It can be a negative or a positive
number. The exact value must be chosen
(calibrated) with the help of an external instrument.
TIME_MULT is a variable that transforms the time
measurement from clock cycles to seconds. It
depends on the LM4F232 clock frequency which is
80 MHz (therefore each period is 12,5 ns or
0.0000000125 seconds).
This file also includes the I/V transfer ratio of the
measurement circuitry. The differential amplifiers
transfer ratio is 1. Therefore the current-to-voltage
transfer ratio is dictated only by the shunt resistor. If
the resistance is changed, this has to be reflected in
the configuration file on line DIFFAMP_RATIO.
ADC_RESOLUTION and ADC_REF reflect the
ADC resolution in quants and the ADC reference in
volts.
5. Testing an example.
To test the correct implementation of the
firmware and the PC driver a test measurement will
be performed. The setup is shown in Fig. 5.
The device under test (DUT) is a Texas
Instruments Stellaris LM3S6965 microcontroller. A
specific firmware, developed specially for this test,
is flashed into the DUT. It allows us to set an
arbitrary number of synchronizing impulses as well
as their length.
UART
(For DUT flashing)
UART
Powot
Analyzer
Synchronization
Line
(LM4F232)
Power Supply
Diff Amp
Out
DUT
Oscilloscope
LM3S6965
Fig. 5. Test setup.

The main part of the firmware is shown below[7]:
while(1)
{
UARTprintf("Enter number of impulses: 0x");
ulNumImpulses = gethex(8);
for(int i = 0; i < ulNumImpulses; i++)
{
UARTprintf("\r\n");
UARTprintf("Enter value for delay No%d in ms:
0x", i);
ulDelay[i] = gethex(8);
UARTprintf("\r\n");
}
for(int i=0;i<ulNumImpulses; i++)
{
GPIOPinWrite(1);
delay_ms(ulDelay[i]);
GPIOPinWrite(0);
delay_ms(ulDelay[i]);
}
}
The Host PC runs the Powot tool. The number of
the functions is hard-coded to 4, for example. The
flashing of the modified firmware is skipped (by
commenting out some lines). When the
optimization process is started and Powot awaits the
measurements a random number of values for the
function calls length is entered, e.g. 3 ms, 5 ms, 1
ms and 12 ms. This way we simulate an execution
of a program on the DUT and the length of each
function call is known (3, 5, 1 and 12 milliseconds).
With this setup we can predict the results that would
be returned from the driver to the Powot tool.
In Fig. 6 an oscillogram of the test is
shown. Channel one is connected to the
synchronization pin and channel two is connected to
the output of the differential amplifier. Two
amplitude cursors are enabled to measure the
amplitude of the differential amplifiers output
- 154 -
voltage. It can be seen that even though the

LM3S6965 is executing empty instructions (in the
delay_ms( ) function) its consumption is still raised.
The amplitude is 680 mV which must be divided by
the I/V transfer ratio in our case it is 10. Therefore
the current consumptions during the execution of
the function calls are
I uC
V ADC
0,68
68mA
Rsh .k diffamp 10.1
(6)
The Powot tool displayed the following statistics:

Measurements
------------------------------------------------------------------------| n | Function name
| Idd, A
| t exec, s |
------------------------------------------------------------------------|1 |init();
|0.0679687
|0.00300314
|2 |function_call_one(); |0.068042
|0.00500256
|3 |function_call_two(); |0.068042
|0.00100376
|4 |function_call_three();|0.068042
|0.0120004
-------------------------------------------------------------------------
The error is less than a milliampere which is fair for

a first prototype of the device.
References
1. , ., , , -, 1998.
2. Bogdanov, L., Ivanov R., Approaches for
reducing the power consumption in embedded
systems, Annual Journal Of Electronics, ISSN
1314-0078, Sofia, 2012, p.181 - 184.
3. http://code.google.com/p/qextserialport/
4. http://qt.digia.com/
5. http://qt-project.org/doc/qt-4.8/classes.html
6. Stellaris LM4F232H5QD Microcontroller data
sheet, Texas Instruments, 2012.
7. Stellaris Peripheral Driver Library USERS
GUIDE SW-DRL-UG-8555, Texas Instruments,
2012.
Lubomir Valeriev Bogdanov, PhD student
Department of Electronics
Technical University Sofia
8 Kliment Ohridski blvd.
1000 Sofia
BULGARIA
E-mail: lbogdanov@tu-sofia.bg
14.02.2013 .
Fig. 6 Example execution of a firmware with 4

function calls with 3-, 5-, 1- and 12-millisecond
length.
6. Conclusions.
The Powot tool development can continue
because of the data produced by the Powot
Analyzer. If better accuracy is needed in the future,
the only thing that has to be changed is the PC
driver. That is why this module is compiled in a
different project (as a static library) which allows
clear separation between Powot and Powot
Analyzer making the whole development more
agile.

TECHSYS 2013
BULGARIA

A , ,
:
.
.

.
,
, .
-
,
.
: , - , ,
,
REAL-TIME SUPERVISORY TUNING OF

PREDICTIVE CONTROLLER
MARGARITA TERZIYSKA1, YANCHO TODOROV2, MICHAIL PETROV1
Abstract: One of the main problems in predictive controllers is their tuning. The values of the
prediction and control horizons and weighting factors in the optimization criterion are usually
determined heuristically. This paper presents an approach of supervisory tuning of predictive
controller. The supervisor carries out the adaptive adjustment of the weighting factor that
directly affects the output value of control in constant horizons of prediction. The proposed
algorithm for neural-fuzzy predictive control has been tested in real conditions on a laboratory
heating system, which is located in the Faculty of Mechanical Engineering of the Technical
University in Prague.
Key words: predictive control, neuro-fuzzy model, supervisor, heating system, heat exchanger
1. Introduction
Model Predictive Control (MPC) is an
optimal control technique that relies on dynamic
model of the process, used to predict the future
response of a plant. Afterwards, an optimization
procedure computes an optimal control policy by
minimizing a prescribed cost function. In general,
the industrial processes are inherently nonlinear and
this implies the use of nonlinear models and
respectively Nonlinear Model Predictive Control

(NMPC) algorithms. Predictive controllers (linear
or nonlinear) have several tuning parameters
control horizon, prediction horizon, and weighting
factors in the cost function. The good system
performance in predictive control scheme strongly
depends on its values. In practice, there is no
systematic way for tuning the predictive controller.
This paper presents an approach of supervisory
- 156 tuning of predictive controller. The supervisor

carries out an adaptive adjustment of the weighting
factor that directly affects the output value of
control, while prediction horizons are constant. The
proposed algorithm for neural-fuzzy predictive
control has been tested in real conditions on a
laboratory heating system, which is located in the
Faculty of Mechanical Engineering of the Technical
University in Prague.
2. Neuro-fuzzy predictive control strategy
Nonlinear Model Predictive Control, or
NMPC, is a variant of model predictive control
(MPC) that is characterized by the use of nonlinear
system models in the prediction. While NMPC
application in the past have been mostly used in the
processes with relatively slow dynamics, today it is
wide spread as control strategy e.g., in the
automotive industry, or even when the states are
distributed in space.
In this paper NMPC based on fuzzy-neural
predictive model is used to control a heating system
with heat exchanger. In [3] it is described a fuzzy
model based predictive control of a tubular heat
exchanger system. Neural network based predictive
control of a heat exchanger is proposed in [4].
Cascade Generalized Predictive Control algoritm
for heat exchanger is described in [5]. In [6] is
suggest simplified scheme for predictive control for
a shell and tube heat exchanger.
NMPC as it is applied here with TakagiSugeno fuzzy-neural process model can be
described in general with a block diagram, as it is
depicted in Fig. 1.
~ (i )
~ (i )
(i )
R ( i ) : if x1 is A1 and x p is Ap then f y ( k )
(i )
f y(i) (k) = a1(i) y(k 1) + a2(i) y(k 2) + L+ any
y(k ny )
(i )
+ b1(i)u(k) + b2(i)u(k 1) + L+ bnu
u(k nu ) + c0(i)
(2)
(3)
where (i)= 1,2,N, N is the number of the fuzzy

rules, Ai is an activated fuzzy set defined in the
universe of discourse of the input xi and the crisp
coefficients a1, a2,...any, b1, b2,...bnu are the
coefficients into the Sugeno function fy(k).
In the Takagi-Sugeno fuzzy-neural model it
is needed to be determined the unknown parameters
the number of membership functions, their shape
and the parameters of the function fy in the
consequent part of the rules. This is an
identification procedure for which have been
proposed numerous approaches. In this work it is
applied a simplified fuzzy neural approach.
2.1 Learning algorithm for the designed
fuzzy-neural process model
It is used two steps gradient learning
procedure as a learning algorithm of the internal
fuzzy-neural model. This procedure is based on the
minimization of the instant error between the
process output and the model output. It is needed to
be adjusted two groups of parameters in the fuzzyneural architecture and they are: premise and
consequent parameters. The consequent parameters
are the coefficients a1, a2...any, b1, b2...bnu in the
Sugeno function fy and they are calculated at first
step by the following equations:
ij ( k + 1 ) = ij ( k ) + ( y( ) y M ( )) y ( k )xi ( k ) (4)
( j)
Optimizer
Takagi-Sugeno
predictive model
Heating system
Takagi-Sugeno
predictive model
( j)
0 j ( k + 1 ) = 0 j ( k ) + ( y( ) y M ( )) y ( k )
Fig. 1. Block diagram of a typical model predictive

control system
The Takagi-Sugeno fuzzy-neural models
are suitable to model nonlinear systems, as well as
heating systems with heat exchangers. As it is well
known a wide class of nonlinear dynamic systems
can be described in discrete time by the NARX
(Nonlinear AutoregRessive model with eXogenous
inputs) input-output model. The used model in this
paper is also taken in the NARX type:
y (k ) = f y ( x(k ))
(1)
where the unknown nonlinear function fy can be

approximated by Takagi-Sugeno type fuzzy rules:
(5)
where is the learning rate and ij is an adjustable ith coefficient (ai or bi) in the Sugeno function fy of
the j-th activated rule.
The premise parameters are the centre cij
and the deviation ij of a Gaussian fuzzy set. They
can be calculated at second step using the following
equations:
) [xi (k) cij (k)]

cij (k +1) = cij (k) +( y() yM ()) y (k)[ f y(i) () y(k)]
cij2 (k)
( j)
( j)
ij (k +1) = ij (k) +( y() yM ()) y (k)[ f y(i) () y(k)]
[xi (k) ij (k)]2
(6)
(7)
ij3 (k)
Using the Takagi-Sugeno fuzzy-neural

model, the Optimization Algorithm computes the
future control actions at each sampling period, by
minimizing the following cost function:
- 157 -
N2
Nu
i = N1
i =1
J( k ,u( k )) = ( r( k + i ) y( k + i ))2 +u( k + i 1)2
(8)
where y is the predicted model output, r is the

system reference and u is the control action. The
tuning parameters of the predictive controller are:
N1, N2, Nu and . N1 is the minimum prediction
horizon, N2 is the maximum prediction horizon, Nu
is the control horizon and is the weighting factor
penalizing changes in the control actions.
When the criterion function is a quadratic
one and there are no constraints on the control
action, the cost function can be minimized
analytically. If the criterion J is minimized with
respect to the future control actions u, then their
optimal values can be calculated by applying the
condition:
J[k,U(k)] J[k,U(k)]
J[k,U(k)]
,
,...,
J[k,U(k)]=
=0
u(k +1)
u(k + Nu 1)
u(k)
(9)
Each element from this gradient vector can

be calculated using the following equation:
J [ k , U ( k )]
= 2 R ( k ) Y ( k )
U ( k )
U ( k )
Y ( k )
+ 2 U ( k ) T
U ( k )
U ( k )
(10)
From the above expression can be seen that

it is needed to obtain two groups of partial
derivatives. The first one is Y ( k ) , and second
U (k )
one is
(k )
U
U (k )
. Each element from the first group
of partial derivatives is calculated with the

following equations:
N
y(k )
= b1( i ) y( i ) ( k )
u (k ) i =1
(11)
y(k + 1) N (i ) y(k )
= a 1
+ b (2i ) y(i ) (k + 1) (12)
u (k )
u (k )
i =1
y(k + 2) N (i ) y(k + 1) (i ) y(k ) (i )

= a 1
+ a2
y (k + 2)
u (k )
u (k )
u (k )
i =1
u (k)
u (k)
....
u (k)
u (k + Nu 1) (15)
(k)
U
=
.....
.....
.....
U(k) u (k + Nu 1)
u (k + Nu 1)
....
u (k)
u (k + Nu 1)
Since u ( k ) = u ( k ) u ( k 1) , this is:

1
1
(k)
U
=0
U(k)
0
0
0
0
1 1 0 0
0 1 1 0
0 0 1 1
(16)
3. Tuning of predictive controllers

The tuning parameters of a model predictive
controller are: prediction horizon N2, control
horizon Nu and weighting factors in the cost
function (8). Recently they are selected
heuristically, by trial and error procedures based
on simulations results [1, 2]. In [7] it is proposed
heuristic on-line tuning algorithm for Nonlinear
Model Predictive Controllers, which automatically
adjusts the prediction horizon N2 and the diagonal
elements of the input weight matrix . The control
horizon Nu is left constant. The tuning algorithm is
designed using the fuzzy logic concepts. Simple
auto-tuning rules for MPC are described in [8]. A
method for optimal tuning parameters of the
dynamic matrix predictive controller with
constraints based on genetic algorithms is proposed
in [9]. How to use genetic algorithms for tuning the
parameters of generalized predictive control is
described in [10]. In [11] it is derived an analytic
expression to determine the move suppression
parameter j for mono and multivariable systems, in
order to obtain an automatic tuning method.
In this paper it is proposed an idea for
adaptive tuning of the weighting factor in
optimization criterion (8), maintaining the other
parameters as constants. The modified MPC control
scheme with the applied fuzzy-neural supervisor is
shown on Fig.2
(13)
Supervisor
( i ) y ( k + N 2 1)
a
+ K
N 1
u ( k )
y ( k + N 2 )
(i ) (k + N 2 )
=
( k + N 2 2) y
u ( k )
i =1
( i ) y
+ a2
u ( k )
(14)
The second group partial derivatives have

the following form:
Optimizer
Takagi-Sugeno
predictive model
Heating system
Takagi-Sugeno
predictive model
Fig.2. MPC control scheme with a fuzzy-neural

supervisor
- 158 The used Fuzzy-Neural Supervisor is easily

implemented as a simple fuzzy-neural TakagiSugeno type architecture. This fuzzy-neural
network has the same structure as that one used as a
predictive model. It can be expressed in the
following form:
(k) = f (x(k))
(17)
where x(k ) = [ (k 1), (k 2), e(k ), e(k 1)] and e is

the system error.
The unknown nonlinear function f can be
approximated by Takagi-Sugeno type fuzzy rules:
~ (i )
~ (i)
(i)
Q ( i ) : if x 1 is C 1 and x p is C p then f ( k ) (18)
f (i) (k) = c1(i) (k 1) + c2(i) (k 2) + c3(i) e(k) + c4(i) e(k 1) + c0(i) (19)
(i)=1,2,, where is the number of the fuzzy

rules, Ci is an activated fuzzy set defined in the
universe of discourse of the input xi and the crisp
coefficients c1, c2, c3, c4, c0 are the coefficients into
the Sugeno function f.
As a learning procedure of the presented
adaptive fuzzy supervisor it is considered the same
two steps gradient algorithm discussed above
(Equations (4) - (7)).
4. Description of the laboratory heating
system
The considered plant is situated in Czech
Technical University in Prague, Faculty of
Mechanical Engineering. The laboratory heating
system (Fig. 3) is designed as a multiple input
multiple output (MIMO) system. The system
consists of two closed independent heating circuits,
in which the water is the heat transfer medium. Both
the circuits are equipped with a heater, cooler, pump
and valves, by which the heat transfer within the
circuit can be controlled. The heat flow can also be
transferred between the circuits through a multiplate heat exchanger. The used heat exchanger is
Zilmet Z 1/8 with heat transfer up to 3 kW and it
consists from 5 plates. Monitoring of the heating
system state is done via twelve thermometers and
four flow-rate-meters. The whole system is
controlled either by a PC equipped with data
acquisition cards or by a modular industrial PLC
Tecomat NS950 with analogue input-output
modules. In the Fig. 3 it is presented the photograph
of the laboratory heating system. As it can be seen
the main subsystem units of the system are
connected by pipe lines. It is used the five-layer
aluminum plastic composite pipes Seacomppipe
16x2, with inner diameter 12 mm., because of its
very good isolation features.
Fig.3. A photograph of the laboratory heating

system
The heating system can be divided into two
parts that is call left and right circuit, respectively.
The main difference between these two circuits is
the heater. For the investigations here it is used only
one circuit the left side one.
In the left circuit it is used an accumulationtype heater SHU 5 S STIEBEL ELTRON 071754.
The performance of this heater is 2 kW and its
capacity is 5 l. For the right circuit, a flow-type
heater was designed. The performance of the heater
is 2 kW and its capacity is 1.5 l. The used left
circuit can be described by the following equations:
Td
dd ( t )
= d ( t ) + K d a ( t d )
dt
dc ( t )
= c ( t c ) + K c d ( t c )
dt
d h (t )
Th
= h (t h ) + K b a ( t b ) +
dt
+ K u h , set (t u )
Tc
d a (t )
1+ q
= K a [ h (t )
a (t )
dt
2
1+ q
c (t c )] [ a (t ) a (t c )]
2
Ta
(20)
(21)
(22)
(23)
where Ta, Td, Tc and Th are the time

constants, Ka is the heat transfer coefficient, Kd, Kc,
Kb and Ku are the static gain coefficients, e, d, c
and b are the time delays.
5. Real-time experiments
The results are obtained via Matlab Realtime Toolbox with the next initial conditions:
N1=1, N2=5, Nu=3,
System reference r=65oC;
Consequent changes of the system reference
r=65oC; 75oC; 60oC
- 159 generalized predictive controller. The results are

shown in Fig. 6. Comparison between different
types of control schemes was done under the same
conditions. The value of the weighting factor for
GPC control is assumed to be = 0.001,
respectively, and the initial values for both methods
with a supervisor.
Fig. 4. Transient process responses in case of set

system reference with different initial values of the
weighting factor
At Fig.4 are presented experiments with the
proposed supervisory GPC control algorithm in case
of different initial values of the weighting factor .
The results show that there is no system overshoot
into the transient responses and the settling time of
the process is increased when using smaller values
of and vice versa. It can be mentioned also that
the used in simulation initial values of are valid
only for the considered heating system.
Fig. 5. Transient process responses in case of

variable system reference and with different
initial values of the weighting factor
Another part of experiments are made in
case of variable system reference (Fig. 6). From the
results it can be observed a similar system dynamics
as in the above considered case of invariable system
reference.
The proposed method for supervisory
tuning of predictive controller is compared with
another method for supervisory tuning, which is
based on the Mamdani mechanism and is described
in detail in [12], and with a standard neural-fuzzy
Fig.6. Transient process responses in case of

set system reference, GPC, TS supervisory
GPD and Mamdani supervisory GPC
(=0.001) control schemes
6. Conclusion
An
adaptive
predictive
supervisory
algorithm to the temperature control of a heating
system with a heat exchanger it is presented in this
paper. The nonlinear predictive control strategy is
designed on the basis of a Takagi-Sugeno fuzzyneural model and a simple optimization procedure.
An additional supervisory level in the control
system is introduced for adaptive tuning of a
weighting factor in the predefined optimization
criterion. The obtained real time results show the
efficiency of the proposed approach. The proposed
algorithm performs well in a variety of conditions.
The results in Fig. 6 show that the control
algorithm using an additional level supervising
structures of fuzzy Takagi-Sugeno gives the best
results. Adaptive tuning of the weighting factor
provides performance and process quality. Despite
the fact that the presence of modifications of model
predictive control scheme associated with additional
calculations resulted in a significant influence on
the transition time, in this case does not affect the
performance to reach the desired value. In the
method using Mamdani approach for tuning of
that is not the case. It can be concluded that a
system with supervisory level using Takagi-Sugeno
model is a better option for control the heating
- 160 system that provides better quality and efficiency of

the process.
REFERENCES
1. E. F. Camacho and C. Bordons Model
Predictive Control, Springer, New York, 2004.
2. J. Maciejowski Predictive Control with
Constraints, Englewood Cliffs, NJ, Prentice Hall,
2002.
3. Mazinan, A. H. and N. Sadati Fuzzy Multiple
Modeling and Fuzzy Predictive Control of A
Tubular Heat Exchanger System, 7th WSEAS
International Conference on Application of
Electrical Engineering (AEE08), Trondheim,
Norway, July 2-4, 2008.
4. Jalili-Kharaajoo, M. and B. N. Araabi Neural
Network Predictive Control of a heat exchanger
nonlinear process, Istanbul University Journal of
Electrical and Electronics Engineering, Vol. 4, No
2, 2004.
5. Kokate, R. D. and L. M. Waghmare Cascade
Generalized Predictive Control for Heat Exchanger
process, International Journal of Signal System
Control and Engineering Applications 3(2): 13-27,
2010.
6. Rajasekaran, S. A simplified predictive control
for a shell and tube heat exchanger, International
Journal of Engineering Science and Technology
Vol. 2 (12), 2010.
7. Emad Ali Heuristic On-Line Tuning for
Nonlinear Model Predictive Controllers Using
Fuzzy Logic, Journal Process Control, 13(5), 383396, 2003.
8. Valencia-Palomo, G., Rossiter, J. A. Autotuned predictive control based on minimal plant
informationq 7th IFAC International Symposium on
Advanced Control of Chemical Processes, 2009.
9. De Almeida, G. M., J. L. F. Salles and J. D.
Filho Optimal tuning parameters of the dynamic
matrix predictive controller with constraints, Lat.
Am. appl. res. [online]. 2009, vol.39, n.1, pp. 33-40.
ISSN 0327-0793.
10. De Almeida, G. M., J. L. F. Salles and J. D.
Filho Using genetic algorithms for tuning the
parameters of generalized predictive control, VII
Conferencia Internacional de Aplicaes Industriais
INDUSCON, Recife, 2006.
11. Dougherty, D. and D. Cooper Tuning
Guidelines of a Dynamic Matrix Controller for
Integrating (Non-Self-Regulating) Processes, Ind.
Eng. Chem. Res. 42, 17391752 (2003).

12. Terziyska, M., Y. Todorov and M. Petrov
Supervisory tuning of nonlinear predictive
controller. Proceedings of the International
Conference on "Intelligent Control Systems", Brno,
Czech Republic, 29 August - 11 September, 2005,
pp. 128 - 133, ISBN 80-214-2976-3.
4000 Plovdiv
BULGARIA
Department of "Intelligent Systems"
Institute of Infromation and Communication
Technologies
Bulgarian Academy of Sciences
Akad. G. Bonchev St, bl.2
1113 Sofia
BULGARIA
E-mail: terziyska@dir.bg
E-mail: yancho.todorov@iit.bas.bg
E-mail: mpetrov@tu-plovdiv.bg
28.02.2013 .
- 161 -

TECHSYS 2013
BULGARIA

, ,
: MOSFET
.
.
(CNT),
(Si), (ZnO), (InAs), (In2O3),
(GaSb) (GaN).
.
: , ,
OVERVIEW OF NANOWIRE FIELD-EFFECT

TRANSISTORS
MARIYA SPASOVA, GEORGE ANGELOV, MARIN HRISTOV
Abstract: An overview of the different types of nanowire MOSFETs which have large potential
to semiconductor industry and microelectronic systems is presented. The present paper is
focused on the type of nanowire FETs with carbon nanotube (CNT), Silicon (Si), Zinc Oxide
(ZnO), Indium Arsenide (InAs), Indium Oxide (In2O3), Gallium Antimonide (GaSb), Gallium
Nitride (GaN) channel. The current paper presents the structure and technology of the various
types of Nano Wire MOSFET.
Key words: CNTFET, nanotube, nanowire
1. Introduction
With their unique electrical and optical
properties, semiconducting nanowires offer
interesting perspectives for basic research as well as
for technology. Device scaling is critical for
continuing trend of more functionality in a chip. As
the scaling of Si MOSFET approaches towards its
limiting value, new alternatives are coming up to
overcome these limitations.
Many types of
transistors with nanomaterials in the channel are
studied so far, e.g. Carbon NanoTube Field-Effect
Transistor, Silicon (Si) nanowire transistors,
Gallium Antimonite (GaSb) nanowire transistors,
Gallium Nitride (GN) nanowire transistors, Zinc
Oxide (ZnO) nanowire transistor, Indium Oxide

(In2O3), Indium Arsenide (InAs), etc.
2. Carbon
nanotube
Field
Effect
Transistors - CNTFETs
The carbon nanotubes (CNT) are one of the
most important new materials with excellent
mechanical and electrical properties. The main
advantage of a MOSFET with graphene channel is
its high carrier mobility. CNTFET so far can be
classified into: Back-gated CNTFETs [1], [5], Topgate CNTFETs [11], [10], Wrap-around gate
CNTFETs [2], Multi-Wall CNTFET [13], Vertical
CNTFET [6], Local gated CNT FET devices [14],
- 162 -
[15], multichannel Carbon nanotube FET [3]. One

device fabricated with CNTs that has been highly
examined is the carbon nanotube field-effect
transistor, comprised of single-wall CNTs
(SWCNTs) as the active element between two metal
source and drain contacts. There are many
advantages to the CNTFET, such as size, high subthreshold slope, and low power consumption. There
are two main methods for CNT creation: CVD
Growth CNT Solution Deposition.
3. Silicon NanoWire Transistors
Type 1. Silicon nanowires can be prepared
with single-crystal structures, diameters as small as
several nanometers and controllable hole and
electron doping, and thus represent powerful
building blocks for nanoelectronics devices such as
field effect transistors. Thermal annealing and
passivation of oxide defects using chemical
modification were found to increase the average
transconductance from 45 to 800 ns and average
mobility from 30 to 560 cm2/V-s with peak values
of 2000 ns and 1350 cm2/V-s, respectively. Figure 1
shows schematic of a SiNW FET showing metal
source and drain electrodes with the NW and
contacts on the surface of SiO2/Si substrate. Highresolution transmission electron micrograph of a 5
nm diameter SiNW; the scale bar is 5 nm. Figure 1
(B) shows scanning electron micrograph of a SiNW
FET device; the scale bar is 500 nm. Thermal
annealing and passivation of oxide defects by
chemical modification were found to increase the
average transconductance from 45 to 800 ns and
average mobility from 30 to 560 cm2 /V-s with peak
values from 2000 ns and 1350 cm2/V-s,
respectively. Comparison of these results and other
key parameters with state-of-the-art planar silicon
devices shows substantial advantages for the SiNWs
as building blocks [4].
Fig. 1. Schematic of a SiNW FET [4].

Type 2. Here is demonstrated an atomic
force microscopy lithography that enables the
reproducible fabrication of complex singlecrystalline silicon nanowire field-effect transistors
with a high electrical performance. The nanowires
have been carved from a silicon-on-insulator wafer
by a combination of local oxidation processes with
a force microscope and etching steps. The silicon

nanowire transistor is fabricated with a channel
width of 4 nm. Fig. 2 shows images of a multiplelevel SiNWs transistor. The device is formed by
two perpendicular SiNW. Fig. 2 (a) shows
amplitude modulation AFM image of the local
oxide mask fabricated with the AFM. Fig. 2 (b)
shows amplitude modulation AFM image of two
perpendicular SiNWs (A-B and C-D) obtained after
etching with KOH. Fig. 2 (c) shows optical image
of the final device [12].
Fig 2. Images of a multiple-level SiNWs

transistor[12].
Type 3. Here is demonstrated vertical
surround-gate field-effect transistor VS-FET.
Exemplarily, Si nanowires are used and present a
first electrical characterization proving the
feasibility of the process developed and the basic
functionality of this device [17].
Fig. 3. a) TEM of a silicon nanowire VS-FET)

. b) Colored TEM image (green: silicon, blue:
aluminum)[17]
Type 4. This type presents the first attempt
to fabricate two transistors on a single vertical
silicon nanowire (Fig.4) realizing true 3-D
integration. The fabricated device architecture,
having two independently controlled MOSFETs
connected in series, does not occupy any additional
planar area except an extra gate contact.
Consequently, compared with the single vertical
nanowire transistor, the stacked MOSFETs
provided the two input AND gate functionality with
50% area savings. The drain current ID could be
modulated by either of the gate biases without
impacting the VT of the other transistor in the stack.
Furthermore, the tunability of the separation
between the two gates, which is 7 nm in the current
demonstration, makes the presented device
- 163 -
architecture promising for implementing electrically

doped tunneling FET (TFET) and stacked SONOS
memory cells [14].
grown InAs NWs were over 10 m long with a

radius range of 7-20 nm. The NWs are single
crystalline with a native oxide thickness of 2-2.5
nm. Energy dispersion spectrometry (EDS)
indicates that the chemical composition of In:As is
close to 1:1. Field-effect transistors in a back-gated
configuration were fabricated, for the electrical
transport measurements (Figure 6, panels a-b) [7].
Fig. 4. Schematic showing the 3-D view of the

stacked vertical twin-gate nanowire MOSFET [14].
4. ZnO Nano Wire Transistors
Type 1. This type demonstrates all solution
processed, low-temperature ZnO NW transistor
fabrication processes on polymer substrates by
combining two different NP (Au and ZnO) based
solution processes - Au NP direct nano imprinting
and the ZnO NW synthesis on ZnO NP seeds. The
ZnO NW network transistor (NWNT) fabrication
process consists of two main steps: (1) source-drain
electrode fabrication by the direct nanoimprinting of
Au NPs and (2) hydrothermal ZnO NW growth
from ZnO NP seeds in aqueous ZnO precursors.
Both steps are low-temperature (Tmax < 140C),
NP based solution processes. These characteristics
make this technique directly applicable to low-cost,
solution processed electronics on inexpensive
polymer substrates (Fig. 5) [16].
Fig. 5. Schematic illustration of the ZnO NW

network FET fabrication process [16].
Fig. 6. (a) A top-view schematic of a global backgated NW FET, used for the I-V characterization.
(b) SEM image of a representative back-gated NW
FET [7].
6. In2O3 Nano Wire Transistors
Type 1. In this study, the channel length was
altered by placing a conducting atomic force
microscopy (CAFM) tip at various positions along
an In2O3 nanowire. The In2O3 nanowires were
prepared by carbothermal reduction followed by a
catalyst-mediated heteroepitaxial growth technique.
A detailed description of the synthesis of In2O3
nanowires has been reported elsewhere. Fig. 7(a)
shows a typical field-emission scanning electron
microscopy (FE-SEM) image of In2O3 nanowires
grown on the Si/SiO2 substrate. Note that the
highresolution TEM (HRTEM) image shown in fig.
7(b) indicates that the In2O3 nanowire is a single
crystalline material with an interplanar spacing of
0.5 nm in the (100) direction. The electron
diffraction pattern was recorded along the (001)
zone axis, as shown in the inset of fig. 7(b). Figure
7 (c) shows schematic of the CAFM setup for
characterizing an In2O3 nanowire FET. The channel
length of the In2O3 nanowire FET was varied from 1
m to 20 nm. Figure 7(d) shows the AFM image of
the In2O3 nanowire FET device with diameter of 25
nm. The CAFM tip was positioned on the nanowire
at specific locations (a=1 m, b=500 nm, c=200
nm, d=50 nm, and e=20 nm). Note that the AFM
image (Fig. 7(d)) was obtained after the completion
of the CAFM experiments in order to avoid
potential damage to the nanowire by the CAFM tip
[8].
5. InAs Nano Wire Transistors

Type 1. InAs NWs are synthesized on
Si/SiO2 substrates by a physical vapor transport
method using Ni nanoparticles as the catalyst. The
- 164 -
8. GaSb Nano Wire Transistors

Type 1. Single GaSb Nanowire Field Effect
Transistors (NWFETs) were fabricated (fig. 9).
Among IIIV compound semiconductors, GaSb has
the potential to be a promising candidate for high
speed electronic and photonic applications in the
mid-infrared region. This is due to its high hole
mobility of GaSb (850cm2/Vs) and wide range of
band gaps available for GaSb-based alloys with
spectra range from ~0.3 to 1.58eV (0.8-4.3m). The
synthesis of GaSb nanowires has been studied and
the one-dimensional (1D) geometry of nanowires is
ideal building block for transistor devices [20].
Fig. 7. (a) FE-SEM image of In2O3 nanowires. (b)

HRTEM image of an In2O3 nanowire. (c)Schematic
of the CAFM setup for characterizing an In2O3
nanowire FET. (d) AFM image of the In2O3
nanowire FET device characterized [8].
7. ZnO Nano Wire Transistors
Type 1. ZnO nanowire field-effect
transistors (FETs) were fabricated and studied in
vacuum. In air, these n-type nanowire transistors
have some of the highest mobilities yet reported for
ZnO FETs (e = 13 5 cm2 V-1 s-1), with carrier
concentrations averaging 5.2 2.5 x 1017 cm-3 and
on-off current ratios ranging from 105 to 107. Four
probe measurements show that the resistivity of the
Ti/Au-ZnO contacts is 0.002-0.02 cm. Figure 8
is a scanning electron microscope (SEM) image of a
typical ZnO nanowire FET [9].
Fig. 8. SEM micrograph of a 101 nm diameter ZnO

nanowire device [9].
Fig. 9. Typical SEM image of one nanowire

device. The gap between the electrodes is 1m [20].
The
crystalline
GaSb
nanowires
characterized in this work are synthesized using the
self-catalyzed vapor-liquidsolid approach. Pools of
gallium supported on amorphous quartz substrates
are employed for this purpose. Antimony is
supplied through the vapor phase from a solid
antimony source by using 10% hydrogen in argon
as the carrier gas. The dissolution and subsequent
supersaturation of the molten gallium droplets
formed on the substrate with antimony lead to the
multiple nucleation and growth of GaSb nanowires.
Following growth, the nanowires are removed from
the growth substrates and suspended in isoproponal
solution. The so grown GaSb nanowries were then
drop cast from isoproponal dispersion onto an
oxidized silicon substrate. 200nm SiO2 is thermally
oxidized on top of a heavily p-doped silicon
substrate, which serves as a global back gate. Cr/Au
electrodes were patterned on top of each individual
nanowire using JEOL 6000 FS Electron Beam
Lithography to define the features in a 600nm thick
polymethylmethacrylate (PMMA) layer spun on the
surface of the substrate. The patterned PMMA films
were metalized by thermally evaporating 15nm Cr
and 200nm Au. After lift-off procedure, the asfabricated devices were annealed under forming gas
environment (Nitrogen 95%, Hydrogen 5%) at
300oC for 30 minutes [20].
- 165 -
9. GaN Nano Wire Transistors

Type 1. The focus of this research was to
demonstrate the use of gallium nitride (GaN)
nanowires to fabricate a CMOSFET using the
nanowire as the conducting channel (fig. 10.).
Fabrication of the nanoFET was initiated with a low
resistant, highly doped silicon (Si) wafer, which
was cut into two one-inch-pieces. These Si wafers
were then thermally oxidized in a quartz furnace at
1100C to form a layer of silicon dioxide (SiO2),
which will be used as the gate dielectric. SiO2 layers
of thickness between 275-300 nm were used. To
form the gate contact, a 150 nm thick aluminium
(Al) ohmic contact layer was deposited onto the
backside of the Si wafer by electron beam
evaporation. To protect the layer during subsequent
fabrication processes, a layer of photoresist was
spun onto the surface and baked for 30 min.
Employing photolithography, the source and drain
electrodes for the nanoFET were defined atop the
SiO2 layer. We then evaporated 150 nm of Al, or
gold (Au), over this surface. A liftoff process was
then used to form the source drain contacts. The
mask used in this work was specifically designed to
maximize the probablity of getting nanowires across
the source and drain. The GaN nanowires used were
grown in house and measured roughly 50 nm in
diameter and 10-100 m in length [18].
confirm nanowire placement on the source-drain

contacts (Fig. 11) [18].
10. Conclusion
This paper presented an overview of the
Nanowire Field Effect Transistors structures
depending on the type of nanowire and device
architecture. The present stady was focused on the
type of nanowire FETs with Carbon Nanotube
(CNT), Silicon (Si), Zinc Oxide (ZnO), Indium
Arsenide (InAs), Indium Oxide (In2O3), Gallium
Antimonide (GaSb), Gallium Nitride (GaN)
channel. Nanowire MOSFETs are viewed as the
most promising material for the upcoming
nanoelectronic transistors. However, the Nanowire
MOSFETs field is still in an early stage of
development.
Acknowledgement
This paper was financially supported by
contract No. 1220045-03, R&D Department,
Technical University of Sofia.
REFERENCE
1. Aikawa, S., Einarsson, E., Inoue, T., Xiang, R.,
Chiashi, S., Shiomi, J., Nishikawa, E.,
Maruyama , S. "Simple Fabrication Technique for
Field-Effect Transistor Array Using As-Grown
Single-Walled Carbon Nanotubes", Japanese
Journal of Applied Physics 50 (2011),
pp.04DN08(1-4)
2. Chen, Z., Farmer, D., Xu, S., Gordon, R.,
Avouris, P., Appenzeller, J. "Externally
Assembled Gate-All-Around Carbon Nanotube
Field-Effect Transistor",IEEE electron device
letters, VOL. 29, NO. 2, February 2008 pp. 183-185
Fig. 10. GaN Nanowire Transistors [18].
3. Chen, C., Xu, D., Kong, E., Zhang, Y.,

"Multichannel Carbon-Nanotube FETs and
Complementary Logic Gates With Nanowelded
Contacts", IEEE Electron Device Letters, Vol. 27,
No. 10, October 2006, pp. 852-855
4. Cui, Y. Z.Zhong, D. Wang, W. Wang, Ch.
Lieber High Performance silicon nanowire Field
Effect TransistorsNano lett.,Vol.3, No.2, pp. 149153, 2003.
Fig. 11. Scanning electron microscope [18]

The solvent was allowed to evaporate, and
the sample was then examined under an optical
microscope and a scanning electron microscope to
5. Derycke, V., Martel, R., Appenzeller, J.,

Avouris Ph.Carbon Nanotube Inter- and
Intramolecular Logic Gates American Chemical
Society, Vol.1, Num. 9, pp 452-456, September
2001.
6. Franklin, A., Sayer, R., Sands, T., Fisher, T.,
Janes, D. Toward surround gates on vertical
single-walled carbon nanotube devices Mar/Apr
2009,
American
Vacuum
Society,
DOI:
10.1116/1.3054266, pp 821-826
- 166 -
7. Ford A., Ho J., Y. Chuen, Y. Tseng, Z. Fan, J.

Guo, J. Bokor, A. Javey, Diameter-dependent
electron mobility of InAs NanowiresAmerican
Chemical Society, Nano Letters, Vol. 9, No. 1, pp.
360-365, 2009
17. V. Schmidt, R. Heike, S. Senz, S. Karg, W.

Riess, U. Gsele, Realization of a Silicon
Nanowire Vertical Surround-Gate Field-Effecs
Transistor Nano Small Micro, Vol.2, No.1, pp. 8588, 7 November, 2005.
8. Gunho, J., Jongsun M., Tae-Wook K., H.

Woong-Ki, J. Minseok, H. Hyunsang, L. Takhee
Effects of channel-length scaling on In2O3
nanowire field effect transistors studied by
conducting atomic force microscopy American
Institute of Physics, Applied physics letters 90,
pp.173106-1 173106-3, 2007
18. Wingfield, A. Fabrication of a Gallium Nitride

Nano-Field Effect Transistor NNIN REU Research
Accomplishments, pp.80-82, 2010
9. Goldberger, J., Sirbuly D., Law M., Yang P.,

ZnO Nanowire Transistors The journal of
physical chemistry, B. Letters, pp. 9-14, 13
december 2004
10. Hu, Y., Yao, K., Wang, S., Zhang, Z., Liang,
X., Chen, Q., Peng, L."Fabrication of high
performance top-gate complementary inverter using
a single carbon nanotube and via a simple process",
American Institute of Physics, Applied Physics
letters 90, 223116, 2007, pp. 223116 (1-3)
11. Lin, H., Park, Y., Tiwari, S. "A Compact
Single-Walled Carbon Nanotube Transistor
Integrated with a Silicon MOSFET Using a Single
Common Gate", Materials Research Society 2007,
Vol. 963
12. Martinez, J., R.V. Martinez, R. Garcia
Silicon Nanowire Transistors with a Channel
Width of 4 nm Fabricated by Atomic Force
Microscope Nanolitography American Chemical
Society, pp. 3636-3639, Vol. 8, No.11, 2008
13. Shun, M., Ganhua L., Kehan, Y., Junhong,
Ch.,Specific biosensing using carbon nanotubes
functionalized with gold nanoparticleantibody
conjugates Elsevier Ltd. All rights reserved,
September 26, 2009, pp. 479486
14. Stokes, P., Khondaker, S. Local-gated singlewalled carbon nanotube field effect transistors
assembled by AC dielectrophoresis University of
Central Florida, Department of Physics, Orlando ,
USA
15. Svensson, P., Tarakanov, T., Lee, D.,
Kinaret, K., Park, Y. Campbell. B.,A carbon
nanotube gated carbon nanotube transistor with 5 ps
gate delay IOP publishing, Online at
stacks.iop.org/Nano/19/325201
16. Seung K., Inkyu P., P.Heng, M. Nipun, M.
Rogers, C. Grigoropoulos, A. Pisano, ZnO
nanowire network transistor fabrication on a
polymer substrate by low-temperature, all-inorganic
nanoparticle solution process, Applied Physics
Letters, Vol. 92, pp. 154102-1 - 154102-3, 2008
19. Xiang L., Ch. Chixian, Sh. Nansheng, D.

Sarkar, N. Singh, K. Banerjee, G. Lo, D. Kwong
Vertically Stacked and Independently Controlled
Twin-Gate MOSFETs on a Single Si Nanowire,
IEEE Electron Device Letters, pp. 1492-1494,
Vol.32, No.11, November 2011.
20. Xu, W., Chin A.,Y. Laura, N. Cun-Cheng, Y.
Hongbin,Electrical and optical characterization of
individual GaSb nanowires, Proc. Of SPIE Vol.
7224, pp. 72240G-1 72240G-7, 2009
Department of Microelectronics,
Faculty of Electronic Engineering and
Technologies
ECAD Laboratory.
Technical UniversitySofia,
1000 Sofia.
BULGARIA
E-mail: mls@ecad.tu-sofia.bg
E-mail: gva@ecad.tu-sofia.bg
E-mail: mhristov@ecad.tu-sofia.bg
15.03.2013 .
- 167 -

TECHSYS 2013
BULGARIA
,
:
1MW, , .
- . , ,
.
.
. ,
.
.
: , ,
PHOTOVOLTAIC GENERATION PLANT

DESIGN
MIHAIL KALBANOV, RUMEN POPOV
Abstract Conceptual project for grid connected 1 MW photovoltaic generation plant, located
near village Resen, Veliko Tarnovo region, was developed. The project stages, methods,
instruments and results are described. The solar radiation resources and the main technical
properties of the plant were determined. he complete equipment set variant was chosen.
Financial analyses were accomplished and the photovoltaic generation plant technical
requirements were defined. The plant ecological impact was analyzed and estimated.
Key words: photovoltaic plant, solar resources, design methods
1.

()
,

,
.

.
,
. ,

.
2.

.
- 168 -
,
,
,
,
.
)

.
PVGIS Europe ,
[7].

,
e
[5].
10 , 1981 1990.
CM SAF PVGIS
.
, CM SAF [1, 2, 3]
.
12+
.
Meteosat (Meteosat 5 7),
MFG 1998 2005,
( MSG)
2006. .
1,5 ( 3 km
0 N, 0 W).
35 S 58 N 18 W
55 E.
,

3.8%.
:
: 4311'35"N, 2532'30" E,
: 68 m (.1);

: Climate SAF PVGIS;
: 1.0 W;
:
;
: 9.4% (
);
,
: 3.3%;
(, .): 7.0%;
: 18.7%.
)
(
, ,
, ,
.)
,
,
. 1.
)

, ,

.
)
PVplanet
(PV Plant Engineering Toolbox) Siemens [6].
)

.
)

,

.

), ) ), -.
3.
CM SAF
PVGIS

.
.1.
:
Hh:
(Wh/m2/day);
Hopt:
(Wh/m2/day);
H(90): : 90. (Wh/m2/day);
Iopt: ();
- 169 -
T24h: O 24h (C).

-

,
.
1.
; : [4]
H(90) Iopt T24h
Hh
Hopt
1400
2100
2070
61 0.4
2340
3330
3030
55 3.1
3450
4280
3260
43 7.0
4740
5150
3070
28 12.3
6120
6060
2870
15 18.1
6600
6220
2580
9 22.1
6670
6450
2790
13 24.3
6010
6380
3400
24 23.9
4230
5110
3580
39 18.5
2800
3940
3450
53 13.4
1730
2750
2750
62 7.5
1250
1860
1850
61 1.9
3950
4480
2890
33 12.7

( ): Iopt = 33.
( ,
) (.2).
9 62 ,

.
. 2.
: 10 - , 30 - 45 - .

.
. . 2...4.
:
Ed:
(kWh);
Em:
(kWh);
Hd:
,
(kWh/m2);
Hm:
,
(kWh/m2).
2.
10 ; : [4]
-: 10, 0
Ed
Em
Hd
Hm
1470.00 45700
1.65
51.2
2370.00 66400
2.70
75.7
3230.00 100000 3.79

118
4070.00 122000 4.97

149
4970.00 154000 6.25

194
5160.00 155000 6.63

199
5240.00 162000 6.76

210
4880.00 151000 6.28

195
3700.00 111000 4.62
139
2680.00 83200
3.23
100
1810.00 54300
2.10
62.9
1290.00 40100
1.47
45.5
128
3410.00 104000 4.21

1250000
1540
3.
30 ; : [4]
-: 30, 0
Ed
Em
Hd
Hm
1850.00 57200
2.06
63.7
2850.00 79900
3.27
91.4
3580.00 111000 4.25

132
4170.00 125000 5.15

155
4830.00 150000 6.13

190
4870.00 146000 6.31

189
5030.00 156000 6.54

203
4940.00 153000 6.41

199
4030.00 121000 5.08
152
3200.00 99100
3.88
120
2330.00 69900
2.68
80.4
1620.00 50300
1.82
56.5
3610.00
110000
4.47
136

1320000
1630
- 170 -
4.
45 ; : [4]
-: 45, 0
Ed
Em
Hd
Hm
2000.00 61900
2.24
69.5
3030.00 85000
3.49
97.8
3640.00 113000 4.34

134
4040.00 121000 4.99

150
4490.00 139000 5.69

176
4430.00 133000 5.73

172
4610.00 143000 5.99

186
4710.00 146000 6.12

190
4050.00 122000 5.12
154
3380.00 105000 4.12

128
2560.00 76700
2.96
88.7
1760.00 54400
1.99
61.6
134
3560.00 108000 4.40

1300000
1610
-

1 320 000 kWh.
, 10 -,
30 45 -,
1 367 000 kWh,
3,56 % - .
10
- 45 -
1 361 000 kWh,

0,44% -
. ,
,
.

:
: 280W;
: 3570; : 13;
: 1,000160 W;
23 ;

( )

10 45 (. 3).
. 4.
. 3.
. 4. : 20kV . 20kV III,

. 1, , . , . .
- 171 -
20kV
20kV ,
20/400 3 . .
:
LDK-280P-24 3570 .
42 . (600V);
SMA Sunny Central 500U
500 kW 2 .;
Schneider Electric (
);
TM 1000kVA 20/0.4kV.

. , ,

, TUV
.
,
,
.

.

25., 10. 2011, 0,33
/kWh, 70% 2550000
. (
) . 5.

9,5 .
5.
25 .; : [6]

1kWh [] [kWh]
[]
1
0,330
1 178 080
388 766,40
2
0,330
1 172 190
386 822,57
3
0,330
1 166 329
384 888,46
4
0,330
1 160 497
382 964,01
5
0,330
1 154 695
381 049,19
6
0,330
1 148 921
379 143,95
7
0,330
1 143 176
377 248,23

1kWh [] [kWh]
8
0,330
1 137 461
9
0,330
1 131 773
10
0,330
1 126 114
11
0,330
1 120 484
12
0,330
1 114 881
13
0,330
1 109 307
14
0,330
1 103 760
15
0,330
1 098 242
16
0,330
1 092 750
17
0,330
1 087 287
18
0,330
1 081 850
19
0,330
1 076 441
20
0,330
1 071 059
21
0,330
1 065 704
22
0,330
1 060 375
23
0,330
1 055 073
24
0,330
1 049 798
25
0,330
1 044 549
[]
375 361,99
373 485,18
371 617,75
369 759,66
367 910,86
366 071,31
364 240,95
362 419,75
360 607,65
358 804,61
357 010,59
355 225,53
353 449,41
351 682,16
349 923,75
348 174,13
346 433,26
344 701,09
, (IEC)
(TC82),

.
(
, )

EC .
PV :
IEC 60904: Measurement Principles
for Terrestrial Photovoltaic (PV) Solar Devices
with Reference Spectral Irradiance Data;
IEC 61215: Crystalline silicon
terrestrial photovoltaic (PV) modules Design
qualification and type approval;
IEC 61646: Thin film terrestrial
photovoltaic (PV) modules Design
qualification and type approval;
IEC 61853: Performance testing and
energy rating of terrestrial photovoltaic (PV)
modules;
IEC 60038.
:
E-
EN 50081,50082;
EN 50160, 61000.

N 60555, N 61000-3-2.
:
IEC 61277: Terrestrial photovoltaic
(PV) power generating systems - General and
guide;
- 172 -
IEC 61727: Photovoltaic (PV)

Systems - Characteristics of The Utility
Interface;
IEC 62116 : Testing Procedure Of
Islanding Prevention Measures For Gridconnected Photovoltaic Power Generation
Systems;
IEC 62446: Grid connected
photovoltaic systems - Minimum system
documentation, commissioning tests and
inspection requirements.
:
IEC 61173: Overvoltage protection
for photovoltaic (PV) power generating
Systems;
IEC 61508: Safety Integrity Level SIL ;
IEC 61730: Photovoltaic module
safety qualification.
:
IEC 60364-7-712.
:
ISO 20282-1.
:
IEC 1024-1, IEC 81/64/CD.

:

.

.

.
(6 ),

,
.
,
,

, .
4.

,

, ,
.

.
,

.
1. Huld T., Mller R., Gambardella A., A new

solar radiation database for estimating PV
performance in Europe and Africa. Solar Energy,
86, 6, 1803-1815 (2012)
2. Ineichen P., Barroso C.S., Geiger B.,
Hollmann, R., Marsouin A., Mller R., Satellite
Application Facilities irradiance products: hourly
time step comparison and validation over Europe',
International Journal of Remote Sensing, 30, 55495571 (2009)
3. Mller R., Matsoukas C., Gratzki A., Behr
H.D., Hollmann R., The CM-SAF operational
scheme for the satellite based retrieval ofsolar
surface irradiance - A LUT badsed eigenvector
hybrid approach. Remote Sensing of Environment,
113, 1012-1024 (2009)
4. PVGIS (c) European Communities, 2001-2012:
http://re.jrc.ec.europa.eu/pvgis/apps4/pvest.php
5. Scharmer K. Greif J. The european solar
radiation atlas. Vol. 1: Fundamentals and maps, Les
Presses de lcole des Mines, Paris, 2000.
6. Siemens PVplanet Ref. No: IN 2012.06.5e
http://www.siemens.com/innovation/en/news/2012/
e_inno_1219_1.htm
7. ri M., Huld T.A., Dunlop E.D., PVGIS: a
web-based solar radiation database for the
calculation of PV potential in Europe. International
Journal of Sustainable Energy, 24, 2, 55-67 (2005)
4000 Plovdiv
BULGARIA
E-mail: kalbanov@abv.bg;
E-mail: rum_pop@yahoo.com
26.02.2013 .
- 173 -

TECHSYS 2013
BULGARIA
A STUDY OF SOME BLIND IMAGE

DECONVOLUTION TECHNIQUES
MIRCEA WEINGART, OREST VASCAN
Abstract: Image deconvolution techniques evolved to complex algorithms that try to

reconstitute the image degraded by motion blur or from optical camera focus reasons .
Non-blind, semi-blind and blind deconvolution algorithms have been implemented to restore
images with a known , partially known or unknown PSF (Point Spread Function).
In this paper it is assumed that the PSF is unknown and that the image is both blurred and also
has salt-and- pepper noise . Techniques are proposed for image restoration and experimental
results are presented .
Key words: blind image deconvolution , Mumford-Shah algorithm , denoising and deblurring ,
Karhunen-Loeve transform , radial-basis neural networks
1. Introduction
Images are often subject to a process of degradation
that is due to one or more different factors.
The sources that generate images can be quite
different: medical imaging equipment ,photo
cameras ,astronomical observations ,microscopy ,
radar and so on.
All these image generating sources and the
communication channel can produce a degradation
of the image .
Examples of image degradation are :
- Optical blur , due to camera out of
focus;
- Motion blur; Additive or multiplicative
noise ;
- Quantization used for image processing
The image degradation model may be described as
follows :
g = h*f + n
(1)
Copyright
where g is the degraded image , f the original

true image , h is the operator that degrades the
image , and n is the noise.
For a two dimensional image , and if h is a
linear shift-invariant operator , the relation (1)
may be written as a convolution in the spatial
domain (here in plane).
g(x,y) = h(x,y)*f(x,y) + n(x,y)
(2)
g is called a blur image and h acts as a blurring
filter named also Point Spread Function (PSF)..
Apply Fourier transform to (2) and get
G() = H()F() + N()
(3)
It results that
F( )
G() N()
H()
(4)
and the reconstructed image f is the inverse

Fourier transform of F .
The solution of this inverse problem is difficult
to find if H() has zeroes or is too small or if
the noise dominates the solution .
2013 by Technical University Sofia, branch Plovdiv, Plovdiv, BULGARIA. ISSN 1310 - 8271
- 174 -
A regularization needs to be applied to

solve an optimization convex problem .
1
f(x,y) arg min || g
2
f * h ||2
J( f )
(5)
where J(f) is a regularization term and a

Lagrange multiplier.
The simplest regularization avoiding the unwanted
noise effect in the inversion is the l 2 norm:
J(f) =
1
|| f ||2 , ensuring that the restored image
2
has a bounded energy .

If PSF is known the problem of restoring the image
f is called a non-blind deconvolution.
If the PSF is unknown there is a blind
deconvolution problem .
In this case both the original true image must be
restored and also recover the PSF.
Techniques were proposed to recover the image
from blur and noise through deconvolution using a
unified variational approach as in [8] or a two
stage denoise by median filtering-followed by
deconvolution as in [14].
Neural networks were also used for blind image
deconvolution by neural recursive function
approximation as described in [4] .
2.
Description of the solution and
techniques used for a comparative study of
blind image deconvolution
2.1
The proposed solution for blind
image deconvolution
It is proposed a four phase approach for blind
image deconvolution and denoising.
It is assumed that there is no knowledge of the
PSF for the processed images.
The First phase :
it is performed a Principal Components Analysis
( Karhunen-Loeve ) transform of the blurred and
noisy image in order to decorrelate the image
pixels-by orthogonalizing the new basis vectors
.It also orders -according to ordered decaying
eigenvalues -the resulting orthogonal principal
components , allowing us to consider only
components with the largest variations and so to
get a compression of the image data set.Noise is
present in the higher frequency domain , so by
selecting the first principal components it is
possible to remove some noise .
The Second phase :
As recommended in [3] a median filtering gives
good results before the actual deconvolution
takes place. Median filtering is applied here

using the median non-linear spatial filter that is a
low-pass filter .
The Third phase :
it is performed a deconvolution based on the
maximum likelihood algorithm (Matlabs
deconvblind [5] applied to the result image of
the second phase).
The Fourth phase :
It is performed a minimization of the MumfordShah functional that has good edge preserving
properties and isolates noise through a process of
image segmentation .This minimization is
applied to the imageresulted from phase three .
It is used the complete combinatorial solution
that uses graph based tools for minimization of
the functional as presented in [1] and [2] .
This phase may be followed by image data
compression as proposed in [15] for an optimal
storage and data transmission channel usage and
by classification with neural networks .
2.2 Description of the
techniques
used for blind image deconvolution and
denoising
Principal
Components
Analysis
(Karhunen-Loeve transform)
Let X represent n vectors with K components
each .
T
X= [ x1, x2 ,......., xn ]
(6)
A transformation M : X Y = M(X- mX ) is
sought such that the covariance matrix of Y is
diagonal.
For each vector xj :
1 K
mX = E{X} =
xi = (m1 ,m2 ,mn )T (7)
K i1
is the vector of mean values of the n input
vectors .
T
= E{(X- mX ) ( X m X ) }
(8)
is the covariance matrix of X.
If the rows of M are the orthogonal eigenvectors
(they are the principal components)of
X
ordered
according
to
decaying
eigenvalues ,then
is a diagonal covariance
Y
matrix of Y with the diagonal comprised of the
eigenvalues of
, so that the decorrelation
X
of pixels is achieved in the new coordinate basis
system formed by the principal components
- 175 -
with the preservation of variance .The i-th

component of
is the variance of X along
Y
the i-th principal component coordinate.
References are in [9] , [10] , [11] .
Median filtering
It is used the implicit 3x3 mask moving over the
image and taking the median value from the
ordered set of values under the mask - it is used
the medfilt2 function of Matlab referred here in
[5] and [6].
All Matlab processing was performed with
Matlab R2011b.
Deconvolution
Is performed on a weighted blurred and noisy
image.
This image has a salt-and-pepper noise and a
blur resulted from a Gaussian low-pass filter
(image is represented as a normalized matrice
with values in the [0 1] interval) .It is used
deconvolution (Matlab weighted deconvblind) as
in [12].
Minimization of the Mumford-Shah
functional
This implemented algorithm is applied to the
result from the previous phase three .
Given R a domain decomposed in regions Ri (
i=1,.,n) that are connected components of R .
R R1 R2 .......Rn
, with Ri R
and
is the boundary separating the regions.f
approximates piecewise smooth the true image g
The
Mumford-Shah
functional
is
2
2
(9)
||grad ( f ) || dxdy ( )
2 ( f g ) dxdy
R
and the solution is a pair (f, ) that minimizes

the functional. The boundary should have a
minimum length. and are parameters. The
first term of the functional is called the data
fidelity term and represents how close is f to g
The second term represents a smoothing of the
edges and has the role of edge preservation.The
third term represents the boundary with a length
that has to be minimal.
The image g is
supposed to vary smoothly in each Ri , and varies
discontinuously or rapidly across
-the
R
.
i
boundary that separates the regions
Copyright
The block representation of the proposed

solution
BLURRED
NO ISY
IMAGE
PRINCIPAL
COMPONENTS
ANALYSIS
WEIGHT ED
DECO NVO LUTIO N
MEDIAN
FILTER
MUMFORD SHAH
FUNCTIO NAL
MINIMIZATIO N
UNSHARP
FILTER
Fig.1 The block components representing the

proposed solution
3.Experimental Results
For the computer software simulated experiments
it was used Matlab code ,and also was used code
from [2] and [7] processed on a laptop with
Pentium 4 Mobile processor at 2GHZ , and 1 GB
RAM .
Lena , Barbara and Peppers original grayscale
images are in bitmap format , with dimension
512x512 pixels with 8 bits per pixel .First of all
blur was simulated with a rotationally symmetric
Gaussian lowpass filter that was applied to the
original image .The Gaussian filter size ( hsize) is
15 , for standard deviations (sigma) : =3 and
=5. Then 5.4 % salt-and-pepper noise was added
.Next phase was to perform Karhunen-Loeve for
the noisy and blurred image.and were used the
principal components of the image for further
processing . The image was denoised by using a
median filter ,then
was deblurred by
deconvolution (with a weighted deconvblind).The
informational energy in the first 474 , and
respectively 342 principal components is 99.99 %
- from the total energy corresponding to all 512
principal components of the image.MUMFORDSHAH segmentation was then applied for the case
where Gaussian filtering and salt-and-pepper was
added .Unsharp filtering was tested as an
alternative solution for edge sharpening and was
applied to the denoised and deblurred image , and
the result was below the Mumford-Shah
algorithm performance.
- 176 -
BLUR
AND
NOISY
IMAGE
DENOISED
blur
BARBARA IMAGE
filter
PSNR (dB)
sigma
=3
16.39
=5
Without
Mumford
-Shah
AND
DEBLURRED
IMAGE
16.14
=3
512
PC
=5
512
PC
=3
472
PC
=5
472
PC
=3
342
PC
=5
342
PC
22.05
With
MumfordShah
22.05
BLUR
AND
NOISY
IMAGE
DENOISED
21.48
19.11
19.13
18.73
18.76
18.35
18.46
18.04
18.20
Table 1 Barbara image results

PC: represents the number of selected principal
components in all tables and images attached in
this article.
A supervised Neural network , variant of Radial
basis neural network called Probabilistic Neural
network was trained with the denoised and
deblurred Lena, Barbara and Peppers images as one
set , for each of these images were selected all 512
principal components . In [16] is mentioned that a
pre-processing - such as Principal Components
Analysis of the training input vectors may improve
efficiency of the neural neworks . Each of the three
training images was assigned to one class that has a
unique numeric code .This type of neural network
has two layers. The first layer is the Radial Basis
Layer and computes the distance between the input
vector and the input training vectors and it has an
exponential transfer function the exponent is the
weighted distance mentioned before.The input layer
receives the training vectors and the target vector
that represents the classes assigned to the training
vectors.The second layer-called the Competitive
Layer- classifies the input vector into a class based
on a compete transfer function that selects the class
assigned to the input vector ,such that this class to
=5
DEBLURRED
17.24
Without
With
Mumford- MumfordShah
Shah
AND
IMAGE
IMAGE
21.48
blur
LENA IMAGE
filter PSNR (dB)
sigma
=3
17.47
=3
512
PC
=5
512
PC
=3
472
PC
=5
472
PC
=3
342
PC
=5
342
PC
28.05
28.06
26.54
26.55
20.70
20.76
20.26
20.30
19.75
19.94
19.59
19.77
Table 2 Lena image results

be correct with the maximum probability .
In the experiments was simulated a classification
with this neural network for each of the above
mentioned three images separately , by selection of
32 , 72 , 142 , 342 and 472 principal components
for each image , for the deblurred and denoised
images . The classification was 100% correct for
all cases mentioned above .
In the noisy and blurred images simulation case for
the same selection of principal components (32 ,
72 , 142 , 342 and 472) the classification was also
correct with the exception of the peppers image for
the selection of 32 principal components with a
50% rate of success .
PSNR (peak signal-to-noise ratio) measures the
difference between two images and it is defined as
PSNR = 20 * log10 (255/rms) , where rms is the
root mean square difference between the two
images.
- 177 -
blur
filter
sigma
BLUR
AND
NOISY
IMAGE
=3
16.39
=5
16.09
DENOISED
AND
DEBLURRED
IMAGE
PEPPERS IMAGE
PSNR (dB)
Without
Mumford
-Shah
With
MumfordShah
25.83
25.83
=3
512
PC
=5
512
23.86
PC
=3
472
20.15
PC
=5
472
19.74
PC
=3
342
19.20
PC
=5
342
18.56
PC
Table 3 Peppers image results
23.86
20.16
19.77
19.33
used here for image deblur and denoise .From the

images presented above results that phase is much
degraded with blurring and in the presence of noise
and that the images FFT (Fast Fourier Transform)
magnitude is located in a lower frequency domain
when is compared with the true original image FFT
magnitude spectrum. A correct phase restoration
would help the correct image restoration.By using a
multi phase approach as described above we get an
improvement of the PSNR and image quality for
the case of the selection of the relevant principal
components of the deblurred and denoised image.
The Mumford-Shah algorithm performs a better
edge preservation of the image than unsharp
filtering as can be seen in the images from this
article.
Neural networks can be used in this article is
presented the case of the probabilistic neural
network , a variant of the radial basis neural
network for automatic image classification of the
deblurred and denoised images. The code of the
class can be used for data compression if it
associated with an image belonging to the specified
image class .
Thanks to Professor Victor-Emil Neagoe for advice
and to Professor Gabriel Peyre and Dr. Leo Grady
for the permission to use their specific application
code and also for suggestions.
18.69
Interpretation of the results:

-Mumford-Shah filter improved the PSNR for the
cases where 342 and 474 principal components
were selected .This filter was applied to the
deblurred and denoised image .
- Mumford-Shah filter applied for 512 principal
components that is the case when all principal
components were used may improve also the
PSNR or converge to the same PSNR result as for
the deblurred and denoised image that was not
processed with the Mumford-Shah algorithm .
4. CONCLUSIONS
AND ACKNOWLEDGEMENTS
When blur and noise are acting on an image , a
non-linear process of degradation takes place.
The blur can be regarded for the linear shift
invariant case - as a result of low-pass filtering ,
while the noise adds non-linearity to the image.This
non-linearity justifies the non-linear techniques
Copyright
- 178 -
Barbara image deblur

and denoise
No Mumford-Shah
512PCs
Barbara
image
deblur and denoise
No Mumford-Shah
472PCs
Lena image
FFT2 PHASE
blurred and noisy
Lena image FFT2

MAGNITUDE
blurred and noisy
Barbara image
blur and noise
Lena- blur
and noise
Lena image deblur

denoise
UNSHARP
filtering 512 PCS
Barbara
deblur
denoise
UNSHARP
filtering 512 PCS
Lena image deblur

and
denoise
472 PCS
with Mumford-Shah
Lena deblur
denoise 512 PCS
with
MumfordShah
Lena original image

FFT2 PHASE
Lena
original
image
FFT2
MAGNITUDE
Barbara
FFT2
PHASE
blurred and noisy
image
FFT2
MAGNITUDE
Barbara
blurred
and noisy image
Peppers
image
Peppers
deblur
and denoise - 342
PCS
with
Mumford-Shah
original
- 179 -
Barbara
image
deblur and denoise
with Mumford-Shah
512PCs
Barbara
image
deblur and denoise
with
MumfordShah 472PCs
Barbara
original image
Lena
original image
Lena image
deblur denoise
472 PCS
no Mumford-Shah
Lena
image
deblur
and
denoise 512 PCS
no Mumford-Shah
FFT2
PHASE
original
image Barbara
FFT2
MAGNITUDE
original
image Barbara
References
1. Leo Grady, Christopher Alvino
IEEE Transactions on image processing, VOL. 18,
NO. 11, NOVEMBER 2009 The Piecewise Smooth
Mumford-Shah Functional on an Arbitrary Graph
6. Alasdair McAndrew : Introduction to

Digital Image Processing with Matlab Notes
for SCM2511 Image Processing 1 ,Semester 1,
2004
2. Leo Grady , Christopher Alvino Combinatorial

Mumford-Shah functional code
7. Gabriel Peyr A Numerical Tour of Signal

Processing :Advanced Signal, Image and
Surface Processing : Ceremade, Universit
Paris-Dauphine
3 Jian-Feng Cai, Raymond H. Chan , , Mila

Nikolova : Two-phase approach for deblurring
images corrupted by impulse plus gaussian noise
Inverse Problems and Imaging , Volume 2, No. 2,
2008, 187204
4. Jiann-Ming Wu, Hsiao-Chang Chen, ChunChang Wu, and Pei-Hsun Hsu ,World Academy
of Science, Engineering and Technology 46 2010
Blind image deconvolution by neural recursive
function approximation
5. Matlab : deconvblind algorithm.
Copyright
8. Leah Bar, Nir Sochen, Nahum Kiryati :

Scale Space and PDE Methods in: Computer
Vision
9. Victor Neagoe , Octavian Stanasila: The
theory of pattern recognition , Editura
Academiei Romane , Bucharest , 1992
10. Jonathon Shlens : A Tutorial on
Principal Component Analysis ; Center for
Neural Science, New York University, New
York City, April 22, 2009; Version 3.01
- 180 -
11. Mark Richardson ,Mathematical Institute

University of Oxford :Principal component analysis,
"Special topic, 2009
12. Mathworks web-site: Deblurring Images
Using the Blind Deconvolution Algorithm
http://www.mathworks.com/products/demos/image/
ipexblind/ipexblind.html
13. D. Mumford , J. Shah. Optimal
Approximations by Piecewise Smooth Functions
and
Associated
Variational
Problems.
Communications on Pure and Applied Mathematics,
Vol. 42, pp 577-686, 1989.
14. Jian-Feng Cai Raymond H. Chan Mila
Nikolova Fast Two-Phase Image Deblurring Under
Impulse Noise , J Math Imaging Vis (2010) 36: 46
53 DOI 10.1007/s10851-009-0169-7
15. Vascan Orest , Weingart Mircea
Cascading Transformations and Vector quantization
in Image Compression : Recent Researches in
Information Science and Applications WSEAS
Conference January 9-11 , 2013 , Milan, Italy ;
Lecture Notes in Computer Science Volume 3459,
2005, pp 107-118
16. Howard Demuth , Mark Beale , Martin
Hagan Neural Network Toolbox - for use with
MATLAB
Mircea Weingart and Orest Vascan are PhD

students , The Faculty of Electronics,
Telecommunications
and
Information
Technology , University Politehnica
Bucharest , B-dul Iuliu Maniu 1-3, cod postal
061071, sector 6, Bucuresti, Local Leu, corp
A, et.1,
Romania
E-mail: mirceaw201@yahoo.com
E-mail: mircea.weingart@radioromania.ro
E-mail: ovascan@crescendo.ro
04.03.2013 .
- 181 -

TECHSYS 2013
BULGARIA

, ,
: ,

.
.
. ,

.
.

/
.
: , ,
INFORMATION TECHNOLOGY ASSESSMENT

OF ENERGY SAVINGS IN INDUSTRY
NANKO BOZUKOV, DIMITAR SPIROV, NIKOLAY KOMITOV
Abstract: The report discussed the algorithm of the program to process data, which is
important for accurate and fair assessment to determine the energy efficiency developed and
validated methodologies. The data and methods are based on measurements and are based on
estimates. Before any assessment must provide accurate data possible. Presented algorithm is a
sub-program that can be used as a standalone program to determine the energy savings for the
production of a product in the industry. Amount of energy saved in one year is calculated by a
formula specified in methodology. The evaluation of the energy savings in industry is done by
measuring and / or estimating consumption before and after the implementation of the energy
efficiency measure.
Key words: energy savings in industry,program algorithm, energy efficiency
1.

. -

.

.

- 182 -
.

.

.
, ,
, .
2.

.

.

,

[7].

/

.

[6].
3.

,

,
,
.
[6]:
-
.

, . ,
,

.
-
.

.

.

,

,

.
-
.

.

,
.
. 1.
, -, -
- 183 -
4. -
.

(.1).

(1, 2, 3,...12)

.
,
.

.
3.

.

,

MatLab,
.
MatLab ,

.
,
, ,
, , ,
, ,
,
[2, 3, 4].

(. 2).
:
SECI -

kWh/../;
SECI -

kWh/../;
W
(, 2, ..);

FES [kWh/.]
FES=( SECI - SECI)* W , kWh/
(1)
,
,
, .

,
.
[6].

,

:
,
,
, ,
,
.
. 2.
,
-, -
4.

,
,

- 184 -
.

.

. ,
,

.
5.
,

.

,
.

.
.

.

.

,
,
,
.

- .

. ,

.
1. .., MATLAB 6: ,
, , 2001.
2. ., , . MATLAB
, , ,
2001.
3. http://www1.bpo.bg/images/stories/buletini/
binder-2012-09.pdf
4. www.mathworks.com
5. http://www.eulaw.egov.bg/Document
Display.aspx?ID=382929
6. http://www.mi.government.bg/files/useruploads/f
iles/microsoft_word-metodiki.pdf
7. http://www.climamarket.bg/
Department of Informatics and Statistics

University of Food Technologies - Plovdiv
26 Maritza Blvd.
4000 Plovdiv
BULGARIA
E-mail: bozukovnanko@abv.bg
E-mail: nikkomitov@abv.bg
13.02.2013 .
- 185 -

TECHSYS 2013
BULGARIA
APPLICATION OF NEW INFORMATION

TECHNOLOGY IN TRANSMISSION POWER
SYSTEM
NEVENKA ROGLEVA, VANGEL FUSTIK, SOFIJA NIKOLOVA-POCEVA,
ATANAS ILIEV, ANTON CAUSEVSKI
Abstract: The application of new Information Technology in transmission power systems will
be discussed in this paper. Web services and Service-Oriented Architecture (SOA) will be
analyzed. Implementing Web Services and SOA in Power system lead to higher system
interoperability and simplify planning, monitoring and controlling the system, but on the other
hand control system is exposed to vulnerabilities, hackers attack, undelivered or missed
information and system disturbances.
In this paper SOA model is used to store the data from Transmission System Operator, like
faults in substation, power lines and plants. Designed SOA services can be used either by end
users or programmer code. Using SOA data base the analyst can make risk assessment of
transmission system faults and use the results from the model to imply new investments in power
system - plants, substations, power lines.
Key words: Transmission system,Web services and Service Oriented Architecture, Faults,
Risks, Risk Management
1. Introduction
The transmission power systems in
Republic of Macedonia consist of a high voltage
transmission substations and power lines of 110,
220 and 400 kV. There is only one transmission
system operator witch has a single centralized
control center. Because of the liberalization of the
electricity market and connection to ENTSO it is
very important for the operator and for the
economic stability of the country to have Power
System that is reliable and available and safety
during the time.
The ENTSO requirements concerning
network components are implementation of
redundant equipment and access points that in case
of equipment failure or possible fault in part of the
system, the system to stay stable and reliable.
In order to decrease the consequences of the
this kind of failures and possible risk events in the
Power System, not only transmission operator but
distribution and plant operators have to identify the
potential failures or uncertain event, to identify the
vulnerabilities and attacks or at least to plan
periodically measures and equipment maintenance.

This is part of risk analysis procedure that has to be
implemented in investment and maintenance
department of national power system operator.
2. Introduction to Web Services and SOA
A Service-Oriented Architecture (SOA) is a
component model that supports interaction between
services. A service is a software module with
defined interface specifying a set of operations that
the service provides and a set of messages that the
service receives/sends, an implementation of the
interface, and if deployed, a binding to a
documented network address. SOA can be
implemented using several alternative technologies
including Web services [1].
A Web service is a service that defines its
interface using the Web Services Description
Language, such a service can be accessed using a
protocol that is compliant with the Web Services
Interoperability standards [2]. Web service
interfaces are platform and language independent,
- 186 -
thereby allowing Web services running on different

platforms to interoperate.
Usually SOA model consists of a three
components-service provider, service broker and
service consumer. Provider is a service that has the
ability to describe the capabilities and
characteristics and to publish them. Service
provider model contains service design, future
developments, testing and monitoring assets [1,2].
Service consumer, needs an information
about the basic system components in every
moment during its operating time. For example the
number of substation and their features, number of
transformers, buses and power lines. For high
voltage power lines and buses it is important which
one is sending bus/power line or receiving
bus/power line.
Un
(kV)
Table 1. Event list from TSO
400/110
Power
line in
defect
X
Time
of
switching
off
09:51
400/110
400/110
11:07
12:56
109
Failure in
power switch
400/110
09:48
11:15
93
Current
transformer
failure
08:07
Time
of
switching
on
12:48
Failure
time
(min)
177
Cause
11:04
117
Bad weather
condition
Hot spots
Service consumer searches for available

services through a registry and if service is not
already approved he sends a request to the center
followed with a specification for the format of the
data. This request message is send to the provider
and after validation phase provider sends a response
message to the consumer. The exchange massages
should be defined in an understandable way for both
services, usually SOAP massaging format used by
the service and its requestor [3].
necessary for service provider are the basic system

requirements or functions, load flow, voltage level,
failure type.
3. Why SOA application in Electrical
utilities?
SOA features are used to fulfill the
electrical functions, to exchange the data between
different operators, or to control the functions
between them, and to discover the possible
uncertainty and even to remove system/equipment
failure.
In the SOA based model, for the
transmission system, SQL and Derby data base is
created. This base contains the name, the voltage
level and the information gathered from Intelligent
Electronic Devices from the HV plants, about all
types of failures and uncertain events for the entire
high voltage substations, power lines and the
installed equipment [12,13].
If a failure appear in already existing
substation, power line or equipment, than the Date
Controller component (algorithm logic) search if the
failure is already defined or not and than sends a
massage to the Data Relayer component. Data
Relayer constis of amount of logic focused on
validating data reports received from the Data
Controller component and if it necessary it convert
them to the required format and data model. If the
answer is positive only the time of failure and the
time of repair is put into the model data base and for
negative answer (a failure appears for the first
time), a new data row is inserted in the base. New
data row contains the information about the type of
the failure for the certain equipment and the time of
failure and repair [5,6] (fig 2.)
Fig. 1. SOA components [3]

The last component in SOA model is Service
broker, which stores service description as building
templates, updates registry and decides if the
service provider should publish its service
description. It is central controlled and helps service
consumer/power client to communicate and to
exchange massages with other services. Information
Fig. 2. Data Controller and Data Relayer

Architecture
- 187 -
Fig. 3. Data flow

3. Results
In order to simplify the operation planning,
maintenance and repair phase and also to
accomplish risk analysis for the transmission power
system, Service Oriented Architecture application is
used. Using NetBeans Ide software Java application
with access to Derby data base is created.
Working with NetBeans IDE can be easily
developed Java desktop, mobile, and web
applications, because it provides great tools for PHP
and C/C++ developers. It is free and open source
and has a large community of users and developers
around the world.
NetBeans
IDE
provides
first-class
comprehensive support for the newest Java
technologies and latest Java enhancements before
other IDEs. It is the first IDE providing support for
JDK 7, Java EE 6, and JavaFX 2.
With its constantly improving Java Editor, many
rich features and an extensive range of tools,
templates and samples. NetBeans IDE provides
different views of imported data, from multiple
project windows to helpful tools for setting up the
applications and managing them efficiently, letting
them to be drilled down into data quickly and
easily, while giving versioning tools via Subversion,
Mercurial, and Git integration out of the box.
From Web Service that was created for the
model, Service Consumer, using user name and
password for validation, can access the data from
Derby data base and use the information about the
existing failure and faults in the transmission
system. As Service Consumer, domestic customer,
distribution operator, department for planning,
maintenance, investment and so on can be declared.
They can use the available information for
their purpose. For example the risk analyst can use
the data for particular failure, perform an risk
assessment and give a risk response about the
failure time, working periods and suggest repair
periods or investment in new equipment [8,9,10].
Also depend the results from the data base
undesired events can be predicted and consequences
from the failure will be decreased.
Fig. 4. Project view in NetBeans

4. Conclusions
Web Services and SOA application for transmission
system was presented. Using Java based software
new type of control center was suggested. This kind
of control center according to IBM helps customers
to increase the flexibility of their business
processes, strengthen their underlying IT
infrastructure and reuse their existing IT

investments by creating connections among
disparate applications and information sources. Data
base can be use for planning future investments in
high voltage substation and power lines equipment,
for defining period of maintenance and repair, for
identifying the weakest chain in power transmission
- 188 -
system and to perform qualitative and quantitative

risk analysis.
Another advantage from SOA based control
center is availability of the failure data to the
consumer. If there is a problem in the power grid,
by using web service, the consumer will be
informed about the failure and prognoses about the
time needed to remove the failure in the system.
References
1. Ferguson, D., Stockton, M.: Service-Oriented
Architecture: Programming Model and Product
Architecture. IBM Systems Journal 44(4) (2005)
753780.
2. Alonso, G., Casati, F., Kuna, H., Machiraju,
V.: Web Services: Concepts, Architectures and
Applications. Springer-Verlag (2004).
3. Biju
Naduvathuparambil
et
al,
Communication
Delays
in
Wide
Area
Measurement Systems, Proceedings of the ThirtyFourth Southeastern Symposium on System Theory,
18-19 March 2002, Pages: 118 122.
4. Bin Qiu, Next Generation Information
Communication Infrastructure and Case Studies for
Future Power Systems Ph.D dissertation Virginia
Tech, April 2002.
5. Mashup Patterns: Designs and Examples for the
Modern Enterprise, Mar 18, 2009 by AddisonWesley Professiona.
6. Qizhi Chen, Ghenniwa, H. and Weiming Shen.
"Web-services infrastructure for information
integration in power systems." IEEE Power
Engineering Society General Meeting, 2006.
7. V.Gomathi,
V.Ramachandran:
Enhanced
Distributed Service Model For Power System State
Estimation, International Journal on Computer
Science and Engineering (IJCSE).
8. A. Petrovski*, V. Fustik*, N. Kiteva Rogleva*,
Goran Leci: Risk Management And Application Of
Hardfibber Process Bus System, Proceedings of 25th International Scientific Conference Systems for
Automation of Engineering and Research, 15-16
September 2011, St. Konstantin and Elena resort Varna, (Bulgaria), (Full paper on CD, Proceedings
pp.). (ISNN 1314-1023).
9. V. Fustik*, A. Petrovski*, N. Kiteva Rogleva*,
G. Leci: Functional Requirements For Electronic
Highway And Risk Analysis For Data Management,
Proceedings of 25-th International Scientific
Conference Systems for Automation of Engineering
and Research, 15-16 September 2011, St. St.
Konstantin and Elena resort -Varna, (Bulgaria),
(Full paper on CD, Proceedings pp.). (ISNN 13141023).

10. Nevenka Kiteva Rogleva, Vangel Fustik, and
Vladimir Trajkovic: Risk Management Methods
For Service Oriented Architecture Implementation
In Electric Power System, 10th IASTED European
Conference on Power and Energy Systems
(EuroPES 2011), June 22 - 24, 2011, Crete, Greece.
11. Qizhi Chen, Ghenniwa, H. and Weiming
Shen. "Web-services infrastructure for information
integration in power systems." IEEE Power
Engineering Society General Meeting, 2006.
Department of Power Plants and Substation,
Faculty of Electrical Engineering and
Information Technologies Skopje,
Macedonia
Ruger Boskovik bb.
1000 Skopje
Macedonia
E-mail: nkiteva@feit.ukim.edu.mk
E-mail: vfustic@feit.ukim.edu.mk
E-mail: nsofija@feit.ukim.edu.mk
E-mail: ailiev@feit.ukim.edu.mk
E-mail: caus@feit.ukim.edu.mk
02.03.2013 .
- 189 -

TECHSYS 2013
BULGARIA
,
:
, .
.
, .
, ,
.
: , , ,
STRENGTH MEASURING SENSORS OF

ELECTRICALLY CONDUCTIVE ELASTOMERS
OPERATING AT VARIABLE ELECTRICAL
MODES
NIKOLA GEORGIEV, VASILINA ZLATANOVA
Abstract: An opportunity to explore strength measuring sensors of electrically conductive
elastomers, operating at variable electrical modes, is viewed. Different schemes for conducting
the experiment have been implemented. Taken was data for the output voltage and electricity of
the sensor at different operating conditions. Its static characteristics have been analyzed while
taken experimentally depending on the frequency, the magnitude of the applied static force and
the dimensions.
Key words: strength measuring sensors, elastomers, sensitivity, resonance
1.

,
, [1].
- ,

/ (,
) (, )
[2].
()
.
,

- 190 -
,
[3].

,
[2].

, : ,
, ,
.
,
[4],
:
,
.

,
,
.
RL
Rg
R
e(t)
. 2. I
,
(. 3.)
2.
-
.
(
)
.

.
.1.
.
Rg
Rg
e(t)
RL
L
. 3. II
1 mm ( I)
2 mm ( II).
3.
. 1.
3.1. I
(.2.)

U=f(F), ,
I, .4.

R C .

, C ,

.

,
,

.
.2

, RL
. 4.
U=f(F) I
- 191 -
, ,
.. - .
F > 3 N ,
.
f
.

- U.
U=f(F)
II (. 5.)
,
-.
.6.
S.
, S -
I.

I = f(F). ,

.7. .8.
. 7.
I=f(F) I
. 5.
U=f(F) II
f=5 kHz 10 kHz

.
, II , - .
.
. 8.
I=f(F) II
, f
= 5 kHz 10 kHz,
. ,
F 3 N, F
,
.
II
-.
3.2 II
(.2.)
. 6. S
- 192 -

I=f(F) I II.
F
,
(.9. .10.).
I , -.
,
,
;
. 11. S
. 9.
I=f(F) I
II
-.
- - , -
;
-
,
.
1.
.,
,
, -, 1999.
2. .,
,
2603, , 1983.
3. ., . ,
, , , 1982.
4. Hillis W., Active Touch Sensing. M.I.T.,
artificial intelligence laboratory, 1981.
. 10.
I=f(F) II
.11.
, -
I .
4.

:
-

5. Weismantee H., A. Gairola, Die verwendberkeit

von leitenden kunsstoffen fur tarfile Sensoren,
Feinwerktechnik, Messtechnik 89, 1981.
4000 Plovdiv
BULGARIA
E-mail: geotek@abv.bg
E-mail: v_zlatanowa@abv.bg
05.03.2013 .

TECHSYS 2013
BULGARIA
,
MEMS XPLORER SKILL
, ,
: . ,
. 80-
. ( - )
,
,
. ;
; . 90-

, ,
; () .
: SKILL, ,
MEMS DESIGN USING MEMS XPLORER AND

SKILL
NIKOLAY DELIBOZOV, ROSSEN RADONOV, MARIN HRISTOV
Abstract: Harvey C. Nathanson is an American electrical engineer who invented the first
MEMS device. Since 1980s in the research labs have been developed MEMS devices. MEMS
(Micro Electro Mechanical Systems) are miniature devices formed by combining mechanical
parts and electronic circuits, typically on a semiconductor chip, with dimensions from tens to a
few hundred micrometres. MEMS are used to make pressure, temperature, chemical and
vibration sensors. At the end of 1990s, most of MEMS devices with various sensing or actuating
mechanisms were fabricated using silicon bulk micromachining, surface micromachining, and
lithography, galvanoforming, moulding (LIGA) processes.
Key words: SKILL, technology file, MEMS
1. Introduction
MEMS are not for any one application or
device, nor are defined by a single fabrication
process or limited to a few materials [1]. Their
fabrication encompasses the advantages of
miniaturization,
multiple
components,
and
microelectronics to design and construct integrated
electromechanical systems. The three characteristic
features of MEMS fabrication technologies are

miniaturization, multiplicity, and microelectronics.
Miniaturization enables the fabrication of compact,
quick-response devices. Multiplicity refers to the
batch fabrication inherent in semiconductor
processing, which allows thousands or millions of
components to be easily and concurrently
fabricated.
Microelectronics
provides
the
intelligence to MEMS and allows the monolithic
- 194 merger of sensors, actuators, and logic to build

closed-loop feedback components and systems.
There are several methods for MEMS to be
designed. In this paper is presented method using
Cadence design environment, SKILL program
language as well as MEMS Xplorer of the
SoftMEMS software.
Microfluidic MEMS
Microfluidic MEMS are
interact with fluid-based systems.
2. MEMS design technology

MEMS components can be classified into
six individual categories [2]. These categories of
MEMS components are based on their application.
These categories include:
MEMS Xplorer provides a set of libraries

containing 34 components from which complete
MEMS devices can be built (fig.3).
designed
to
Bio MEMS
Bio MEMS are designed to interact with
proteins, biological cells, medical reagents, etc.
Sensors
Sensors can be chemical, motion, inertial,
thermal, and optical.
-
Actuators
MEMS actuators can provide power using
either an electrostatic or thermal stimulus.
-
RF MEMS
RF MEMS are devices used to switch,
transmit, filter, and manipulate radio frequency
(RF) signals (fig.1).
Fig. 3. Library Components Window
Fig. 1. RF MEMS switch

-
Optical MEMS
Optical MEMS include optical switches and
reflectors (fig.2).
You can create your own technology as

well. Libraries include basic and advanced device
generators for BULK and the SURFACE
micromachining
technologies.
When
new
technology is made all its features are described by
SKILL language. SKILL provides a safe, high-level
programming environment that automatically
handles many traditional system programming
operations. When create own user-defined
technology is needed binary (ASCII) file
techfile.cds so-called technology file and parameter
files. Technology file consists of Layer Definitions,
Layer Rules and Physical Rules. Each layer in
Layer Definition is defined by a layer-purpose pair,
which consists of a unique layer name and purpose
combination (figs.4 a and b).
Fig. 4a. Layer Definition

Fig. 2. 3x3 micromirror array with the center
mirror actuated to 15 tilt position
- 195 -
Physical rules define the following:

-
Spacing information for individual objects,

for example, width and notch spacing rules.
Spacing information for two objects, for
example, the minimum distance allowed
between objects on the same layer or
different layers.
The amount of space required when one
object encloses another.
The manufacturing grid resolution.
Spacing rules specify the distance required

between layers and the width of objects and paths.
The parameter file of a device generator is a
text file containing all its needed variables
declaration (fig.7).
Fig. 4b. Layer Definition
The layer name usually indicates a type of
manufacturing material. The purpose indicates the
use of layer or material. Multiple layers with the
same name but different purposes can be created.
Layer Rules must be specified to establish the
relationships and interactions between layers (fig.5).
Fig. 7. Parameter File

Fig. 5. Layer Rules
Layer rules define the following:
-
Via layers that connect two conducting

layers.
Layers that are physically and electrically
equivalent.
Stream translation data for a layer.
3. Compiling Technology File

When ASCII technology file is ready it is
necessary to be compiled in order to create
technology library (fig.8).
Physical rules must be specified to establish

spacing within and between objects in the design
and to specify the grid snapping (fig.6).
Fig. 8. Import Technology file Window
Fig. 6. Physical Rules
After compiling of ASCII technology file it

is necessary to check the file (fig.9).
- 196 -
ACKNOWLEDGEMENT
This paper was financially supported by
contract No. 1220038-03, R&D Department,
Technical University of Sofia.
Fig. 9. Check Window
Technology library must be attached to a
design library to use it in design process (fig.10).
REFERENCE
1. Mohamed Gad-el-Hak MEMS Applications,
CRC Pres Taylor & Francis Group, 2006, ISBN: 08493-9139-3.
2. Walraven J. Future Challenges for MEMS
Failure Analysis, Sandia National Laboratories.
Albuquerque, NM USA, 2003.
Fig. 10. Attach Window

When a technology library is attached the
system updates properties of the design library and
updates the technology devices in each cellview to
reference the new technology file.
4. Conclusion
As a conclusion it can be said that MEMS
can be classified into several categories. Each
category consists of several components which help
us to build desired MEMS device. When we want to
make new MEMS device using different type of
technology an own user-defined technology is
necessary to be created. When new technology is
made all its features are described by SKILL
language. When we use own user-defined
technology we need a technology file and parameter
files. The major steps for technology file are
described in the following chart flow:
Create an ASCII technology file

Compile the ASCII technology file
Check the technology file for
conformance to application requirements
Attach the technology library to design
library, cell, or cellview
Run a design session
Every design uses a technology library.
Department of Microelectronics
Faculty of Electronic Engineering and
Technologies
ECAD Laboratory
Technical University Sofia
1000 Sofia
BULGARIA
E-mail: n.delibozov@ecad.tu-sofia.bg
E-mail: mhristov@ecad.tu-sofia.bg
E-mail: Rossen.Radonov@ecad.tu-sofia.bg
17.02.2013 .
- 197 -

TECHSYS 2013
BULGARIA

(SDN).

: ,
NetFlow
OpenFlow.
.
. sFlow
.
: , ,
ANALYSIS OF SOFTWARE DEFINED

NETWORKS (SDN). MONITORING OF
NETWORKS WITH CONTEMPORARY
STANDARDS AND PROTOCOLS
PAVEL DZHUNEV
Abstract: This report analyzes contemporary software networks specifying options for control
and monitoring using NetFlow protocol and standard OpenFlow. There are also elements of a
software-defined networks and their interaction. Research and investigate are basic models and
patterns of modern communication networks. Also is explored technology sFlow - technology to
monitor and decode traffic.
Key words: SDN, Netflow, OpenFlow
1.

, -
,
.
SDN
.

,

,
.

,
.
,

NetFlow OpenFlow.
,
.

.
- 198 -
2.

.1
,
:

,

. [1]

, :
,
,
,
.
3.
- r
(API)
:
- NetFlow
- IPFIX
- sFlow
. 2. - r (API)
. 2 -

. 1.

,

.
API

.
OpenFlow
.
OpenFlow

,
- .

,

(ACL) ,

NetFlow/IPFIX -

.
Cisco Systems.
NetFlow, 5 9.
9
IPFIX (Internet Protocol Flow
Information Export)
. -
NetFlow 2055, 9555
9995 [2].
NetFlow IPFIX

: , IP
, IP , IP ,
TCP / UDP , TCP /
UDP , IP ToS,
, ,
.
. 2 . 3

.
NetFlow ,

.
,

.

- 199 -
sFlow sFlow
, -

.

:
,
,
.

.
. 3
sFlow
. [3].
. 4. SDN
, ,
,
SDN .
4. Layer 4-7
,
L2 L3 , OpenStack.
. 3. sFlow
,

,
.
,
firmware, firmware
,

. ,
. [4]
:
-
;
-
, ,
.
3.

. SDN e

Facebook,
Google, Microsoft Yahoo,

(Open Network Foundation).

[5]

, OpenFlow,

, .
OpenFlow ,

,
LAN WAN .
,
- 200 -

.

,

.
,
Cloud Computing Technologies,
.
1. Kim, H. Improving network management with

software defined networking, Communications
Magazine, IEEE vol. 51, no2 February 2013 , pp
114-119.
2. http://en.wikipedia.org/wiki/NetFlow
3. http://blogs.cisco.com/news/is-it-just-sdn/
4. http://networkworld.bg
5. http://computerworld.bg

Technologies
South-West University "Neofit Rilski" Blagoevgrad
66 Ivan Michailov st.
2700, Blagoevgrad
E-mail: djunev@gmail.com
01.03.2013 .
- 201 -

TECHSYS 2013
BULGARIA

,
:

.

,
,
.
: , 3D
TECHNIQUE FOR 3D VISUALIZATION OF

OBJECTS OF REAL IMAGES
PETYA PAVLOVA, NINA STOILOVA
Abstract: The present work describes a technique developed for 3D presentation of object in
dependence on distribution of the lightness in real image. The technique is demonstrated by
results of application of two special programs: the first one for extraction and visualization of
depth of the visible parts of the object and the second for processing of consequences of
images, extracting the time-spatial gradient..
Key words: Three dimensional visualization, 3D rendering
1.

,

[5].

. ,

,

.

[1].

,
[4].

[2].

-
[3].

.
,
- 202 -

.
-
-
.

, ,

,

.

,
.

,

.

.
2.

:
,

;

.

.

[8].
:
- ,
-x,y
z
;
- .

,
,

,
,

.
;
-
;
-
;
-
.

.
1

.
)
)
. 1. 2D ),
).

.

.

,
.

.
,
.
:
-
,
;
-
;
-
;
-
;
-
. -
,
(1)
arctg (
B / y
)
B / x
(1)
B / x x ,
B / y - y .
- 203 -

,
. ,
,
.
2.
[H1-H2] (o)
339 0;0 - 22
23 - 67
68 - 112
113 - 157
158 - 202
203 - 248
249- 294
295 - 338
EE
NE
NN
NW
WW
SW
SS
SE
. 2.

.

.

,
. 3

.
3.
4 5

)
.4. ),
),
).
)
. 3 :
), ).

[6] [7].

Visual Studio
GdiPlus
OpenGl.
jpg .
)
.5. ),
),
).
- 204 -
) )
4 5,

.
,

.

, ,
.
4.
,

.

1. David E., J. Baek, M. Levoy, Focal Stack
Compositing for Depth of Field Control, Stanford
Computer Graphics Laboratory Technical Report
2012-1
2. Edirisinghe E., J. Jiang, Stereo Imaging, an
Emerging Technology - http://www.loreti.it/
Download/PDF/3D/067.pdf.
3.http://en.wikipedia.org/wiki/Autostereogram
4. http://www.cs.cf.ac.uk/Dave/Vision_lecture/
node11.html
5. Levoy M., P. Hanrahan, Light Field
Rendering, Proc. ACM Siggraph, ACM Press,
1996, pp. 31-42
6. Pavlova P., K. Koleva. Technique for tracking
and visualization of motion in sequence of images
of the solar corona. VI Serbian Bulgarian
Astronomical Conference (VI SBIC), Belgrade,
Serbia, May 2008 pp:247-256.
7. Pavlova P., E. Duncheva, K. Koleva. Method
for Tracking and mapping a motion based on
images of the Solar Corona, VII Bulgarian -Serbian
Astronomical
Conference
(VII
SBIC),
Chepelare,Bulgaria, May 2010.
8. Pharr M., G. Humphreys, Physically Based
Rendering: From Theory To Implementation,
Second Edition, Morgan Kaufmann, Elsevier Inc.

2010.
Department of Computer systems and
technologies
4000 Plovdiv
BULGARIA
E-mail: p_pavlova@tu-plovdiv.bg
14.02.2013 .
- 205 -

TECHSYS 2013
BULGARIA
,
:

.
.
: ,
ELECTRONIC SYSTEM FOR

DEMAGNETIZATION IN MAGNETIC
SEPARATION
ROMEO ALEXANDROV, ANGEL ZABTCHEV
Abstract: It is proposed an electronic system of electromagnetic separator with
demagnetization function in the working area which has the possibility for frequency control of
demagnetization magnetic field. The operation principle of the system is described and some
important simulation solutions are given.
Key words: electromagnetic, finite element method
1.

,

.

.

.
,
-

.

. [2]
- 206 -
2.

.

.
. 1

. 1 [3]

(3). (2)
,

(1),
.

(7)
,
.
,
.
(4)
.
. 1

[1]

.
,

.

[2].

,
,

-
.

,
.
3. .
,

.

.
ERROR! OBJECTS CANNOT BE CREATED
FROM EDITING FIELD CODES.
. 2
. 2
,

.
,

,

.
[1],
- 207 -
,
, ,

.

,
,
.

.
.
,
15
20 Hz.

,

VT1 VT2.

VT1
YA1 VT2
.
VT1 ,

. VT2,
,

VD1 VT1.

VT1 VT2.
,
.

VS1.
R2
C,
. ,

, ,
, .

,
.
,
- .
VS2
C
.

R2.
,

.
. 3

. 4

,
,

.

.

-
.

R2 .
- 208 -
. 3, . 4,
-
. 2

.

. 4 ,
.

,

.
1. , .,

.
. 54 , ,
2011..
2. ., ,
, 1989.
3.
BG
64303
B1
, 2004.
Faculty of Mining Electromechanics
University of Mining and Geology
St. Ivan Rilski 1700 Sofia
BULGARIA
E-mail: romeo.alexandrov@abv.bg
. 5
. 5
C. ,
,
.

,

,
C
.

.
4.

,

.

.
28.02.2013 .
104941
- 209 -

TECHSYS 2013
BULGARIA

.
, ,
:
- .
LabView
.
24 ADC - DAQ Emant 300.
: ,LabView.
POWER MONITORING OF THE

PHOTOVOLTAIC PANEL
ROSEN SPIRIDONOV, VANIA GARBEVA, NIKOLAI PAUNKOV
Abstract: The theme for the use of renewable sources of electricity is very relevant in modern
energetics . By using LabView programming environment is implemented virtual instrument for
monitoring the active power of low power photovoltaic panel. The connection with personal
computer is established through measurement scheme and 24 bit ADC - DAC Emant300.
Key words: Monitoring of photovoltaic cell,LabView.
1.

-
.

1300 - 1500 kWh/m2
(
,
).

[1,3]:
-

,
.
-, -.
12%.
,

- 20 .
-
,

1 m2,

120 Wp/m2
-
-
,
.
,

. ,

0,2 mm 200
mm,
- 210 -

.

.

.
() -

,
.
, -
, 8%,
10
.
-

,
,
.

.
10 , 6-9%,
- ,
- 62 Wp/m2.
,
,

.
,

.

,
.

.
.

: ,
,
, ,
.

,
.

-
(Data Acquisition -DAQ) ,
-,
[2].

(Virtual Instrumentation)

, , ,
.

/. LabVIEW (Laboratory Virtual
Instrument Engineering Workbench)

C BASIC . LabVIEW

, LabVIEW
G.
LabVIEW, C BASIC
,
.
LabVIEW ,
, .

, GPIB, VXI,
PXI, RS-232, RS-485.
,
LabVIEW
(virtual instruments),
.
(VI)
,
(
) ,
VI
VI.

.
,

, ,

.
,
.
2.

.
, .

,
-
.
/
/ e 6
10W. 22V,
0,6.

1.
,
AIN3 DAQ 10
- 211 -
.
:
V R2
V AIN 3
R1 R2
VAIN3 VAIN 2
I CELL RSENESE R2
R1 R2

85,5*35,5*3,5mm.
100,
0,22.
R0603 18K 1% R0603 2K 1%. .2
.
. 1.
.
AIN3 AINCOM

10,

. ICELL
RSENS.
0,2,
RSENS 0,6 120mV.
0,25W.
VAIN2
R1=R3 R2=R4
. 2.
EMANT300, , .
.

// , .3
. 3. .
- 212 -
. 4 ,
. 4. .
.5.
. 5. .
,
,
.

.
,

,

- 213 -

.
3.
. 3 ,
178 mW 21
.

120 mW
.
39,54 mA 4,492V. ,
,

.
.

. 1.
60
180mW.

Write To Measurement
File
,
.
1.
4.

.

.
,
.
24 ADC 2,5V
5/16777216 V 0,3
V.
3. http://solar.bghot.com
4. http://see.ni.com
Nikolai Paunkov
4000 Plovdiv
BULGARIA
E-mail: nick123@abv.bg
14.02.2013 .
1. ., . .
, ,
2002
2. . - 3, 2008
- 214 -
- 215 -

TECHSYS 2013
BULGARIA

,
.. .
.

.
, Matlab/Simulink.
..

.
: , ,
, Matlab/Simulink
AN EFFECTIVE ALGORITHM FOR

LOCALIZATION OF SINGLE LINE TO GROUND
FAULTS IN POWER TRANSMISSION LINES
THROUGH BILATERAL MEASUREMENTS
ROSEN STAMATOV
Abstract: This paper presents an effective algorithm for determining the place of single phase
to ground short circuit in the power transmission line. Records of voltage and currents from
both substations are used as input parameters of the algorithm. To extract the values of the
three-phase voltages and currents and phase angles between them is used discrete Fourier
transformation. The modeling of the transmission line, along with the algorithm is performed
using a package Matlab/Simulink. The results of determining the place of single phase to
ground short circuit in the lines show the validity of the algorithm, even at high levels of
resistance to ground.
Key words: fault location, single line to ground, bilateral measurements, Matlab/Simulink
1.

-
-
,
- 216 -
. ,
:
, ,
, ,
.

.
,

,

.

... 90 % ..
..,

.

.

[8].

. ,

(
).
(1978)
M
[2]. , ,
, .
,
.

.

,
r = 0,
.
,

.

[8].

. ,
,
.

[6,7]
()

[1,3,4,5].

.

,
,
.
,
,

.

,
. . . (1999),

[6]. ,

. ,

.
,

.

. - 1%,
,
,
.
,

.

, 10%
.
. ,

,
[7].

.

,

- 217 -
.

,
.

,

.

[3]
[1,4,5].
,
-

,

.

. . (1982)
,

[1].
,

. ,

,
-
. ,

. . .
(1988) [4]

.
.
,
. ,
,
.

,
,

.
,
. ,
,

.

, .
,

.

,

. ,

.
2.
IA
mZ1
(1-m)Z1 IB
Vk
Rk
Ik
.1. ..
.1
..
.
m
(1 - m) , m
(..).
- Rk, - Vk
- m .

,

. , ,
.
.1

,
. . 2 () . 2 ()
,
,
.
mZ1
mZ2
mZ0
VA1
VA2
VA3
VA
3ZF
Vk
IA0
.2..

- 218 -
(1-m)Z1
(1-m)Z2
VB1
VB2
VA, VB,
IA IB, ,

.
(1-m)Z0
VB0
VB
3Zk
IB0
3.
.2..

: m ; Z1
; Z2 ; Z0 ; VA
; VB ; VA1

; VA2 -
; VA0
; Vk
; IA0
; IB0
; Zk
;
, Vk

;
(1)
Vk V A mZ1 Z 2 Z 0 ;
Vk VB 1 m
. Z1 Z 2 Z 0 ;
(2)
I A, 0 I A
I B ,0 I B
(3)
;
(4)
(1) (2) :
V A mZ1 Z 2 Z 0
VB 1 m
. Z1 Z 2 Z 0
(5)
(5) m
:
V A VB I B .Z1 Z 2 Z 0
3
;
m
I A I B .Z Z Z
1
2
0
3
(6)
(6) .
m,

VA, VB, IA, IB, Z1, Z2
Z0. Z1, Z2 Z0

,

Matlab/Simulink
Matlab ,
Simulink,
. ,
SimPowerSystems,

,
.
Out4
4
2
Out2
1
C onnection
Port
Out3
3
Out1
Vabc
Iabc
Three-Phase
Source
Vabc
Iabc
Three-Phase
V-I Measurement
Distributed
Parameters Line
Distributed
Parameters Line1
Three-Phase
V-I Measurement1
Three-Phase
Source1
.3.
Matlab/Simulink
,

- .

,
. . 3
,
Matlab / Simulink.
,
1.
, ,

. ,
,
.
,
,

.

. Simulink,

.
628,3185 / .
- 219 -
In1
Out1
In2
Out2
In3
Out3
In4
Out4
In5
Out5
In6
Out6
Out1
In1
Out2
Out3
In2
Out4
Radian 1
(pi/180).u
Out5
In3
Va
Out6
Out7
Theta Va
In4
Out8
Out9
In5
Out10
Out2
Ia
Theta Ia
j
In6
Vb
Out12
Out3
Connection Port
Out4
LPF 1
Out13
Display
Theta Vb
In7
Out14
Out5
EP
Radian 2
(pi/180).u
Out11
In1
Out1
In2
Out2
In3
Out3
Out15
In8
Out16
Radian 3
(pi/180).u
Ib
Theta Ib
Out17
In9
Calculator
Out18
Out19
In10
Out20
In4
c
1
Out4
Out21
In11
Out22
Breaker
Rk
In5
Out5
In6
Out6
Radian 4
(pi/180).u
Out23
In12
Out24
FFT
LPF 2
.4.

.4 ,
, ,
(FFT)
.
FFT

.
1

47
km
50
Hz
110
kV
()

0.045531917
/km

0.151489359
/km

0.000617657
H/km

0.001533983
H/km
4.
.

,
(7):
l. l .
.100
l .
l.. ;
l. ;
2 3

Rk = 5 50.
, ,
,

.
5
.
2

Rk = 5

km
km
%
5
5,748
1,592
10
10,550
1,170
15
15,340
0,723
20
20,142
0,303
25
24,937
-0,133
30
29,747
-0,538
35
34,530
-1,000
40
39,347
-1,389
45
44,154
-1,799
3

Rk = 50

km
km
%
5
5,903
1,922
10
10,663
1,410
15
15,400
0,851
20
20,162
0,345
25
24,929
-0,151
30
29,716
-0,605
35
34,469
-1,129
40
39,250
-1,595
45
44,008
-2,111
(7)
:
% - , %;
l. ;
- 220 -
3. Joe-Air Jiang, Jun-Zhe Yang, Ying-Hong Lin,

Chih-Wen Liu and Jih-Chen Ma. An adaptive
PMU based fault detection/location technique for
transmission lines part I: theory and algorithms,
IEEE Transactions on Power Delivery, vol. 15, no.
2, pp. 486-493, Apr. 2000.
.5.

Rk
5.

..
.

.

,

.
:
1.
,

.
2.
,
,
.
3. m,
,
(1) (2).
1. E. Schweilzer, Evaluation and Development of

Transmission Line Fault-Locating Techniques
Which Use Sinusoidal Steady-State Information,
Ninth
Annual
Western
Protective
Relay
Conference, Spokane, Washington, October 1982.
2. H. W. Dommel and J. M. Michels, High Speed
Relaying Using Travelling Wave Transient
Analysis, IEEE PES Winter Power Meeting, New
York, Jan. 29-Feb. 3, 1978, pp.1-7.
4. M. Sachdev and R. Agarwal, A Technique for

Estimating Transmission Line Fault Locations From
Digital Impedance Relay Measurements, IEEE
Transaction on Bower Delivery, Vol. 3, No. 1,
January 1988, pp.121-129.
5. P. Balcerek and J. Izykowski, Improved
unsynchronized two-end algorithm for locating
faults in power transmission lines, Proc. IEEE
Bologna PowerTech Conference, Bologna, June
2003.
6. Q. Zhang, Y. Zhang, W. Song and Y. Yu,.
Transmission Line Fault Location for Phase-toEarth Using One Terminal Data, IEE Proc.Gener. Transm. Distrib., Vol. 146, No. 2. March
1999.
7. T. Adu. A New Transmission Line Fault
Locating System, IEEE Transactions on Power
Delivery, Vol. 16, No. 4, October 2001.
8. Y. Cheng, J. Suonan, G. Song and X. Kang.
One-Terminal Impedance Fault Location Algorithm
for Single Phase to Earth Fault of Transmission
Line, IEEE 2010.
Pedagogical Faculty of Engineering - Sliven
59 Burgas road St.
8800 Sliven
BULGARIA
E-mail: rstamatov@yandex.ru
26.02.2013 .
- 221 -

TECHSYS 2013
BULGARIA
STRATEGY FOR OPERATION OF HYBRID

ENERGY SYSTEM WITH INTEGRATED WIND
AND SOLAR POWER PLANTS
SOFIJA NIKOLOVA-POCEVA, ANTON CAUSEVSKI, NEVENKA KITEVA ROGLEVA,
VANGEL FUSTIK, ATANAS ILIEV
Abstract: This paper presents an approach for solving the generation scheduling problem of a
complex system consisted of conventional and renewable energy sources (RES). Wind and solar
power plants are integrated into the system in order to reduce the total thermal power plant fuel
costs. The optimization problem consists of minimizing the total production costs, respecting
power balance equations for each time interval and all operational system constraints. The
proposed approach is applied on a specific system consisted of thermal power plants (TPP),
storage hydro power plants (HPP), pumped-storage hydro power plant (PSHPP), wind power
plant (WPP) and solar power plant (SPP). In this paper the benefits of energy production from
WPP and SPP, in terms of reducing the production costs of conventional thermal power plants
are investigated.
Key words: renewable energy resources, conventional power plants, optimization, production
costs.
1. Introduction
Thermal power generation requires burning
fossil fuels which creates significant emission of
gases such as CO2, SO2, NOx and has negative
environmental
effects.
Therefore,
using
environmentally friendly fuels, such as renewable
energy sources, can eliminate emissions associated
with electric generation. However due to their
unpredictable and stochastic nature the renewable
energy sources (RES) are not yet in that position to
fully replace the conventional generation. Power
system composed of classical sources of electrical
energy with RES are hybrid systems whose
operation is heavily dependent on the conditions of
occurrence of the source of energy from renewable
nature, and the technological characteristics of the
renewable energy power plant. Wind and solar
power are widely researched and various methods
have been developed for operation and generation
scheduling of such hybrid systems [1-4].
Time-series data representing parameters of
wind and solar have stochastic nature with great
unpredictability, as opposed to the classical energy
sources (hydro, thermal and nuclear) that can with
great precision to fit the needs and plan according to
the load demand. Because of the uncertain nature of

wind and solar power it is necessary to cope in
situations of unexpected declines in power
production or in situations of excess electrical
energy. Integration of pumped storage hydro power
plants in a system with wind and solar power plants
is an alternative for compensating the intermittent
nature of renewable generation.
This paper presents an approach for solving
the generation scheduling problem of a complex
system comprised of conventional power plants,
wind power plants and solar power plants. The
proposed model take into account the following
operational constraints: power balance equations,
constraints resulting from power output limits of
thermal and hydro generating units, power output
limits of wind energy system, power output limits
of solar energy system, balance equations for
available volume of water of HPP and PSHPP and
balance equations for contracted energy for TPPs
over the study period. The proposed model does not
take into account spinning reserve level, the system
configuration and network losses. Lagrange method
is used for solving the optimization problem. For
that purpose MATLAB source code is developed.
- 222 -
2. Methodology
The complex system that is analyzed is
consisted of thermal power plants (TPP), storage
hydro power plants (HPP), pumped-storage hydro
power plant (PSHPP), wind power plant (WPP),
solar power plant (SPP) and additional systems to
deal with pumping water of PSHPP. The TPPs
included in the model are TPPs with and without
contracted energy production over the study period.
For purposes of pumping water at pump
storage hydro power plants WPP and SPP will be
engaged, if they are active in the intervals of the
pump mode, and additional units engaged in energy
for pumping.
The objective of the generation scheduling
problem in such a hybrid system is to determine the
optimal amounts of generation power of
conventional power plants over the study period so
that the total production costs of thermal power
plants and additional units engaged for pumping
water of PSHPP be minimal and be satisfied the
relevant real operational constraints of the system.
Here also power generation of PSHPP is valorized.
The objective of generation scheduling
problem is to minimize the total production cost (F)
while satisfying the operational constraints of the
system.
T NT
MinF Di ,t ( Pi ,t ) Ci Pi ,t Pi , sr t
t 1 i 1
ST
Ds ,t ( Ps ,t ) t
t 1 s 1
K d ,t ( Pphpum,t ) t
D
(1)
tTpum d 1
K g ,t ( Pphgen,t ) t
NPH
tTgen g 1
where:
F - total fuel cost of thermal power plants and
additional units engaged for pumping water of
PSHPP over the study period;
C i - penalty factor;
Di ,t ( Pi ,t ) and D s ,t ( Ps ,t ) -fuel/generation
cost
characteristic of TPP ( i ) with contracted energy
production and TPP ( s ) without contracted energy
production;
Pi ,t and Ps,t - power output of TPP ( i ) and TPP (s)
in time interval t ;
Pi , sr - average power of the TPP (i) in the
optimization period;
pum
K d ( Pph
,t ) - energetic characteristic of additional
unit (d) engaged for pumping water of PSHPP;
K g ,t ( Pphgen,t ) - benefit characteristic of power
production of PSHPP;
Pphpum
,t - engaged power in pump operation mode of
PSHPP (ph) in time interval t ;
t - duration of the time interval;
NT - number of TPPs with contracted energy

production over the study period;
NPH - number of pumped-storage hydro power
plants;
ST - number of TPPs without contracted energy
production over the study period;
D - number of additional units engaged for pumping
water of PSHPP;
T - number of time intervals.
2
Through the part C i Pi ,t Pi , sr we
control the power generated from the TPP ( i ) to be
near the average power in the analyzed period.
The total number of time intervals T is
divided into number of time intervals when PSHPP
operate in generator mode ( T gen ) and number of
time intervals when PSHPP operate in pump mode

( T pum ), where: T Tgen T pum .
The fuel/generation cost characteristic of
each thermal power plant with contracted energy
production will be expressed as a parabolic function
of the power generation Pi :
Di ,t ( Pi ,t ) a i bi Pi ,t c i Pi ,2t
(2)
where ai , bi and ci are constants which are
given.
Analogously energy characteristics of

additional units engaged for pumping water of
PSHPP is defined:
2
K ( Pphpum
) bd Pphpum
c d Pphpum
(3)
,
t
,
t
,
t
d
where bd and cd are constants which are given.
The fuel/generation cost characteristic of
each thermal power plant without contracted energy
production will be expressed as a parabolic function
of the power generation Ps :
D s ,t ( Ps ,t ) bs Ps ,t c s Ps2,t
(4)
where bs and cs are constants which are given.

The benefit characteristic of each PSHPP
will be expressed as a:
K g ,t ( Pphgen,t ) bg Pphgen,t c g Pphgen,t
(5)
where bg and c g are constants which are given.

In each time interval over the study period
the following constraints must be satisfied:
A) Power balance equations
NT
NH
NPH
NPH
h 1
ph 1
ph 1
v 1
gen
pum
Pi ,t Ph,t Pph,t Pph,t Pv ,t
i 1
PV
(6)
Ppv,t Pl ,t , t 1,2, ..., T

pv 1
where:
Ph,t - power output of HPP (h) in time interval t ;
Pphgen,t -power output of PSHPP (ph) when it operates

in generator mode in time interval t ;
- 223 Pv ,t - predicted power output of wind power plant

( v ) in time interval t ;
Ppv ,t - predicted power output of solar power plant
( pv ) in time interval t ;
Pl ,t - system load demand in time interval t ;
NH - number of storage hydro power plants;
V - number of wind power plants;
PV - number of solar power plants.
B) Balance equations for available volume of
water for HPPs
T
Vhdoz Qh,t ( Ph,t ) t , h 1,..., NH
(7)
t 1
where Vhdoz
is total available volume of water for

HPP h, over the study period.
The stream flow characteristic of hydro
generating units will be expressed as a parabolic
function of the power generation Ph :
Qh,t ( Ph,t ) a h bh Ph,t c h Ph2,t
(8)
where ah , bh and ch are constants which are given.
The stream flow characteristic of hydro
generating units is taken to be function only from
the power generation Ph . The stream flow
dependence from available head is not taken into
account.
The operation of hydro generating units
should be in such a way that overall available
amount of water over the optimization period
should be spent.
C) Balance equations for available volume of
water for PSHPPs
doz
V ph
gen
gen
Q ph,t ( Pph,t ) t
tTgen
tTpum
pum
pum
Q ph
,t ( Pph,t ) t
(9)
ph 1,..., NPH
where
doz
V ph
is total available volume of water for
PSHPP ph, over the study period.

Energetic characteristic of PSHPP in
gen
gen
generator operation mode Q ph,t ( Pph,t ) and pump
pum
pum
operation mode Q ph,t ( Pph,t ) are modeled with
gen
parabolic function of the power output Pph
and
,t
pum
Pph
,t respectively:
2
gen
gen
gen
gen gen
gen gen
Q ph
,t ( Pph,t ) a ph b ph Pph,t c ph Pph,t
and
pum
pum
Q ph
,t ( Pph,t )
(10)
pum
pum pum
pum pum2
a ph
b ph
Pph,t c ph
Pph,t
gen
gen
gen
pum
pum
pum
are
a ph
, b ph
, c ph
, a ph
, b ph
and c ph
given constants.
D) Balance equations for contracted energy for
TPPs over the study period
where
Pi ,t t Wi , i 1,..., NT
t 1
(11)
where Wi is contracted energy for TPP (i) over the

study period.
E) Inequality constraints resulting from minimum
and maximum power output limits of TPP, HPP and
PSHPP (when it operate in generator mode and
pump mode).
F) Power output limits of wind turbines [3]:
PW ,t
Vt VI Vt VO
0
t 1,2,...,T
(Vt )
VI Vt VR
P
VR Vt VO
Wn
(12)
where:
Vt - forecast wind speed at time interval t ;
V I - cut-in wind turbine speed;
V R - rated wind turbine speed;
VO - cut-out wind turbine speed;
(Vt ) - power output at wind speed Vt ;
PWn - rated power output of wind turbine.
G) Power output limits of solar energy system
[9]:
Ppv,t
Gt 2
PPVn
Gstd Rc
Gt
P
PVn Gstd
0 Gt Rc
t 1,2,...,T
(13)
Gt Rc
It is assumed that PV cell temperature is

neglected.
where:
PPVn - rated power output of PV generator,
Gt - forecast solar radiation at time interval t ,
Gstd - solar radiation in the standard environment
set as 1000 W/m2 ,
Rc - a certain radiation point set as 150 W/m2 .
3. System description
The system that is analyzed is composed of
two thermal power plants with contracted energy
production (TPP1 and TPP2), one thermal power
plant without contracted energy production (TPP3),
three storage hydro power plants, one pump-storage
hydro power plant, one wind power plant, one solar
power plant and one additional unit engaged for
pumping water of PSHPP.
The study period is three days and it is
divided into 72 time intervals. PSHPP running in
pump operating mode from 3 h to 8 h each day.
In order to solve the generation scheduling
problem in the considered hybrid power system the
following data we must be known: system load
demand, wind speed, solar radiation and available
amount of water for HPP and PSHPP over the study
period, given in equations (6), (12), (13), (7) and (9)
respectively. Active hourly load over study period is
given in Figure 1. The total energy demand is 74840
MWh.
- 224 -
gen
gen
gen
K g ,t ( Pph
,t ) 3Pph,t 0.0003Pph,t
Fig. 1. Load demand over study period

The data for TPPs, HPPs and PSHPP are
given in Table 1, Table 2 and Table 3 respectively.
(15)
Wind power plant consists of 15 wind

turbines type Nordex N80/2.5 MW, arranged in 5
rows and each row has 3 turbines at a distance of
5D. The installed capacity of the WPP is 37.5 MW.
The hub height of the wind turbines is 80 m. For the
prediction of power produced by wind power plant
a program is developed in MATLAB which take
into account the wake effect [7]. The determination
of WPP production is given in [8]. The generated
results for the predicted power produced by the
wind power plant are presented in Figure 2. The
total generated electrical energy from the WPP is
1602.94 MWh.
Table 1
Data for thermal power plants

TPP1
TPP2
TPP3
a (p.e./h)
65
35
b (p.e./MWh)
1.9
2.041
2.352
c (p.e./MW2h)
0.0026
0.00129 0.00098
Pi min (MW)
120
160
Pi max (MW)
300
370
W i ( MWh)
15 542
25 537
Table 2
Data for HPPs
HPP1
HPP2
HPP3
3
3
a h 10 (m /h)
1.58
1.98
0.936
3
3
bh 10 (m /MWh)
0.512
0.306
0.612
3
3
2
c h 10 (m /MW h) 0.000216 0.000196 0.000296
Ph min (MW)
5
5
5
Ph max (MW)
100
180
130
3
3
doz
Vh *10 (m )
2400
3600
3000
Data for PSHPP
Pmgen
in (MW)
gen
a ph
gen
b ph
gen
c ph
0.306
0.000216
pum
a ph
pum
b ph
pum
c ph
0.2142
0.00015
Solar power plant is with nominal capacity

of 20 MW. Based on the predicted solar radiation
over 72 hours and the relation (13) the predicted
power produced by the solar power plant is
obtained, as it is shown in Figure 3. The total
generated electrical energy from the SPP is 176.95
MWh.
Table 3
Pmgen
ax (MW)
20
200
Pmpum
in (MW)
Pmpum
ax (MW)
20
Fig. 2. Predicted power generated by the WPP
200
The total available volume of water for

PSHPP over the study period is 1000 10 3 m 3 .
The energetic characteristic of the
additional unit engaged for pumping water of
PSHPP is presented with:
pum
pum
K d , t ( Pphpum
,t ) 0.9 Pph,t 0.003Pph,t
(14)
The characteristic which present the benefit
of power generation of PSHPP is presented with:
Fig. 3. Predicted power generated by the SPP

4. Results
The following two cases are analyzed:
Study case 1: TPPs, HPPs and PSHPP, are used to
supply the system load demand.
Study case 2: TPPs, HPPs, PSHPP, WPP, SPP are
connected to supply the system load demand.
Newton-Raphson iterative process is used
in order to obtain the optimal amount of power
produced by the conventional power plants.
- 225 -
As a result from solving the generation

scheduling problem the following results are
obtained: the optimal hourly amounts of generating
power of all conventional power plants, the hourly
stream flows of hydro power plants and the total
fuel cost of the analyzed hybrid system.
First we analyze the first configuration of
the system in which WPP and SPP are not included.
The generating results for optimal operation of a
system composed of TPPs, HPPs and PSHPP (when
it operates in generator mode) are presented in
Figure 4. TPP 3 in this case generate 11 850 MWh
and covers 15.83% of the total load demand. TPP1
and TPP2 cover 54.889%, storage hydro power
plants cover 25.699%, while pumped-storage hydro
power plant (when operate in generator mode)
covers 5.49% of the total load demand. When
PSHPP operate in pump mode the power is 1.908%
of the total load demand.
only when PSHPP operate in pump mode, and the

total fuel cost is 1 669.5p.e.. In this study case the
total fuel cost of the system is 140756.5... The
benefit of production of PSHPP is 12423p.e.. Figure
7 presents the hourly stream flows of HPPs and
PSHPP. The negative stream flows indicate the
pumping operation for the PSHPP.
Fig. 4. Optimal amounts of generating power of

TPPs, HPPs and PSHPP, study case 1
Fig. 7. Stream flows of HPPs and

PSHPP, study case1
The operation of PSHPP in the analyzed

period is given on Figure 5. The negative power
output indicate the pumping operation for the
PSHPP. The volume of water in pump operation
mode of PSHPP in this case is 325.779 10 3 m3.
The second case takes into consideration the

WPP and SPP integrated into the system consisted
of thermal energy system and hydro energy systems
(storage hydro power plants and pump-storage
hydro power plant). The WPP production is given
in the previous section (Figure 2) and it covers
2.14% of the total load demand. The SPP
production is given in the previous section (Figure
3) and it covers 0.236% of the total load demand.
The generating results of the generation
scheduling problem of conventional power plants
together with WPP and SPP production are
presented in Figure 8. From the gained results the
TPP3 produces 10103MWh electrical energy, it is
lower in comparison with previous case because of
the integration of WPP and SPP into the system.
HPPs cover 25.7%, while pumped-storage hydro
power plant (when operate in generator mode)
covers 5.6% of the total load demand. When PSHPP
operate in pump mode the power is 2.06% of the
total load demand.
In comparison with previous case the
percentage of saving the fuel cost at TPP 3 is
15.5%. The operation of PSHPP in the analyzed
period is given on Figure 9. The volume of water in
Fig. 5. Operation of PSHPP

Figure 6 presents the fuel costs of each
thermal generating unit and fuel costs (Kd) of
additional unit engaged for pumping water of
PSHPP. From the results in Fig. 6 it is observed that
the TPP 2 has the highest fuel cost (66325p.e), then
follows the TPP 1 (42933p.e.) and TPP 3 (29
829p.e.). For the additional unit there are fuel costs
Fig. 6. Production cost of TPPs and the additional

unit, study case 1
- 226 -
pump operation mode of PSHPP in this case is

351.6295 10 3 m3, which is larger then in the
previous case.
units and the fuel costs of the analyzed hybrid

system.
References
1. Pappala V. S., Erlich I., Rohrig K.,
Dobschinski J. A Stochastic Model for the Optimal
Operation of a Wind-Thermal Power System, IEEE
Transactions on Power Systems, Vol. 24, No. 2,
May 2009, 940-949.
Fig. 8. Optimal amounts of generating power of

TPPs, HPPs and PSHPP, study case 2
2. Li F., Kuri B. Generation Scheduling in a system

with Wind Power, In: IEEE/PES Transmission and
Distribution Conference and Exhibition, Asia and
Pacific, 1-6, 2005.
3. Liang R.H., Liao J.H. A Fuzzy-Optimization
Approach for Generation Scheduling With Wind
and Solar Energy Systems, IEEE Transactions on
Power Systems, Vol. 22, No. 4, November 2007,
1665-1674.
4. Castronuov E. D., Peas Lopes J. A. On the
optimization of the daily operation of a wind- hydro
power plant, IEEE Transactions on Power
Systems, Vol. 19, No. 3, August 2004, 1599- 1606.
Fig. 9. Operation of PSHPP

5. Conclusions
In this paper an approach for solving
generation scheduling problem of conventional
power plants in a system consisting of TPPs, HPPs,
PSHPPs, WPPs, SPPs and additional systems to
deal with pumping water of PSHPP has been
presented. Integrated WPPs and SPPs in the system
contribute to minimizing the total production costs
of thermal power plants. The problem model does
not take into account spinning reserve level, the
system configuration and network losses. MATLAB
program is developed for solving the optimization
problem.
The power output from wind power plant is
obtained by using predicted wind speeds, modeled
power curves of the wind turbine integrated in the
wind power plant and the wake effect model. For
this purpose developed MATLAB source code is
used.
In order to perform verification, the
proposed approach is applied to a concrete system
composed of three TPPs, three HPPs, one PSHPP,
one WPP, one SPP and one additional aggregate
engaged for pumping water of PSHPP.
Two study cases have been analyzed. In
each study case the following results are obtained:
the optimal hourly amounts of conventional power
plants, the hourly stream flows of hydro generating
5. Barton J. P., Infield D. G. Energy Storage and

Its Use With Intermittent Renewable Energy, IEEE
Transactions on Energy Conversion, Vol. 19, No. 2,
June 2004, 441-448.
6. Barton J. P., Infield D. G. Energy Storage and
its Use with Wind Power, IEEE Power Eng.Soc.
General Meeting, Vol. 2, June 2005, 1934-1938.
7. Jensen N.O. A Note on Wind Generator
Interaction. Riso National Laboratory, DK-4000
Roskilde, Denmark, 1983.
8. Nikolova S., Iliev A. Short term prediction of
wind farm power production, International
Scientific Conference ICEST 2009, Vol. 2, June
2009, p.p. 351-354.
9. Marwali M. K. C, Haili M., Shahidehpour S.
M., Abdul-Rahman K. H. Short-term generation
scheduling in photovoltaic-utility grid with battery
storage, IEEE Transactions on Power Systems, Vol.
13, 1998, No. 3, 10571062.
Department of Power Plants and Systems
Information Technologies Skopje
Rugjer Boshkovik bb
1000 Skopje
REPUBLIC OF MACEDONIA
E-mail: nsofija@feit.ukim.edu.mk
E-mail: caus@feit.ukim.edu.mk
E-mail: nkiteva@ feit.ukim.edu.mk
E-mail: vfustic@ feit.ukim.edu.mk
E-mail: ailiev@ feit.ukim.edu.mk
01.03.2013 .
- 227 -

TECHSYS 2013
BULGARIA
,
: ,
.
.
: ,
DEVICE FOR MEASURING POTENTIAL,

INTENSITY AND CHARGE OF THE FIELD OF
ELECTROSTATICALLY CHARGED
MATERIALS
STEFAN STEFANOV, IVAN PRODANOV
Abstract: A device for measuring potential, intensity and charge of the field of electrostatically
charged materials is presented. Conducted research is being analyzed.
Key words: electrostatic, measurement of the fields charge
1.
-
,

,

.

, ,
.. .
: ,

:
;
;
. -

[1,
2]. ,

.

.

.
- 228 -

.

.
2.

. 1.
. 1.
:
1, 2,

7,
( ) 3,

5
(
),
,
6 8.

3
1.
4,

.
9,
10
Re = 1,5.1010 ,

() 11 12.
6 ,
12.
3.

[3].
(),
1 2
.
,
1 2
, 3

1
2.
,
Le
1 2, S,

, R
. S, R,
F,
3
1 2
Le .
, Le

U,
.
11
12

U = f (Le), . 2,

= f (Le), . 3.

7.
1
F,
, .. :
F = E . q = m . g
(1)
:
q , ;
m , kg;
g , m/s2;
E
, V/m.
m = 0,83 gr,
d = 10 mm. (1) .
,

,

. (1) , :
- 229 -
= m.g / q
(2)
:
= U / Le
(3)
(3) (2)
q,

:
q = Le . m.g / U
(4)

. 2, 3 4. ,
5
10 nC. U = f (Le)
= f (Le) . q = f (Le),
. 4, , ,
Le U,
.
,
,
()
1.

7. ,

1 .
,

.
10.
. 3. = f (Le)
. 4. q = f (Le)
4.

,

.
. 2. U = f (Le)
- 230 -
1. . ., . . .
. , 1982.
2. . , .
- , 1981.
3. Stefanov S., I. Milev, I. Prodanov. Explosion
proof inductive neutralizer with discharge element a
mobile conductive sphere. University of Targu Jiu,
9 edition-wih international participation. November
5-6, 2004.
University of Mining and Geology

St. Ivan Rilski
1700 Sofia
BULGARIA
E-mail: enemko@mail.bg
12.02.2013 .
- 231 -

TECHSYS 2013
BULGARIA
GAUGE BLOCK CALIBRATION BY MEANS OF

THE INTERFEROMETRY
TPNKA DVOKOV, FRANTIEK DVOEK
JI SOBOTKA, PAVEL BRDLK, ZUZANA ANDROV, LUK VOLESK
LENKA DDKOV
Abstract: There are using two basic methods for gauge blocks calibration both are described
by international standard EN ISO 3650. First of them is for calibration using laser, eventually
multi-waves interferometry (interferential method). In this case gauge block (GB) is wrung to
the reference plane from the free face of gauge block (gauge length). The second method is
based on the comparison of length of calibrated GB with length of reference GB (comparative
method). In this case there is used double-sided contact measurement by means of length
incremental sensors. This paper deals with interferometry calibration method. Benefit of this
method rests in its high accuracy. On the other hand it is time consuming method and there is
also wear of gauge blocks. During measurement there was investigated GB surface from the
microstructure point of view (microstrutural analysis) by interferometric method. It was proved
that interferometric method has a great influence on the GB wear.
Key words: gauge blocks, interferometry, wear
1. Introduction
Absolute length calibration of gauge blocks
traceable to the definition of meter is an important
task of the national metrology institutes responsible
for providing reliable length artifacts for the
industrial use. The calibration task has been
accomplished by way of optical interferometry with
various working techniques that have been
progressively evolving in response to the
ever-increasing demand on calibration accuracy. [1]
The calibration of gauge blocks against an
optical wave length standard is usually performed in
a specially designed interferometer. Michelson and
Benoit first reported on a comparison between a
standard metre bar and the wavelength of red light
emitted by cadmium in 1893. In effect, they
performed the reverse experiment to what is now
regarded as gauge block interferometry - they
calibrated the wavelength using the material
standard as a reference. This idea led to the eventual
re-definition of the metre in terms of a
spectroscopic wavelength in krypton and, finally, to
the most recent re-definition of the metre in 1983, in
terms of the distance travelled by light in a small
fraction of a second. Historically, several designs of
interferometer have been used for the measurement

of gauge blocks. These generally come from two
classes: static interferometers using fringe pattern
evaluation and, later, moving mirror interferometers
using fringe counting and fiducialization based on
short coherence white light fringes. All the
interferometer designs measure the length of the
gauge block in relation to the wavelength of the
light used, providing a clear traceability route to the
metre. [2]
2. Gauge Blocks
The first and the second industrial
revolution rapidly increased demands on products
accuracy. Consequently it has been necessary to
develop suitable instruments which ensured fast and
precise measuring. This task was met by exploring
wringing of metal blocks in 19th century. The
wringing effect is ensured by slid of two hard blast
surface blocks together. Booth surfaces wring each
together and it evoked large force holding. Easy
connection of gauges makes possible to substitute
large number of individual length standards with
small set gauges. The reduction in the number of
gauges with high accuracy, relatively low cost
- 232 -
evoked that gauge blocks became the primary

method dissemination length traceability for over a
century. 2
Although the new long ranges sensors
technology partially replaced gauge blocks. They
are still important part of length precise control.
Currently, there are a very large number of different
sizes, shapes (comparing rings, cylinders and
spheres), materials (steel, chrome carbide,
aluminum, copper alloys, ceramics), and thermal
expanses properties of gauge blocks. Effort of
researchers is focused to continuously increase
length, thermal stability and degrease of wearing.
Very interesting approach is also connection gauge
blocks as solid thermometers.
3. Gauge block measurement by
interferometry
To measure the length of a gauge block in terms of
an optical wavelength, several steps are necessary.
First, the gauge has to be wrung to a flat plate
(platen) of similar material and left to reach thermal
equilibrium inside the interferometer. The
interferometer measurement beam is then aligned
with the gauge block and platen surface normals,
either by individual adjustment or by precision
tolerance of the platen mount as part of the
interferometer design. After thermal equilibrium has
been achieved, usually in a laboratory that is
temperature controlled to the standard reference
temperature of 20C, the measurement process is
undertaken. The model equations used for length
determination are as follows. Usually, the reported
result is the deviation, d, of the actual gauge block
length, l, from the nominal length, L. [2]
(1)
d lL
d l fit L lt lw L l A ln lG l (2)
The parameters of equation (2) are
identified and discussed in detail below.
The primary influence parameter, which is
then corrected for the effect of non-ideal conditions,
is lfit. This is the best-fit solution for gauge block
length, based on an analysis technique known as the
method of exact fractions. This technique uses
measurements at several wavelengths that are then
compared against each other over a range of
possible solutions, to choose the correct solution
corresponding to the gauge block length. If the
number of wavelengths used in the interferometer is
q, then lfit can be expressed as
l fit
1
q
q
i 1
( i Fi )
i
2
(3)
where i is the wavelength of light source i, Fi is the

measured fringe fraction and i is an integer
representing a fringe order. For given values of Fi
and i there are an infinite number of solutions for
lfit, based on (3). However, the method of exact

fractions only allows solutions where there is good
agreement between the results obtained for different
wavelengths. With careful selection of the
wavelengths, (3) leads to only one solution, where
the results across all wavelengths are sufficiently
close to one another and sufficiently close to the
nominal value for l. The nominal value for the
gauge block length should be checked, prior to
measurement by interferometry, by less accurate
mechanical means or from knowledge of previous
measurements.
The
more
accurately
the
wavelengths are known and the more accurately the
parameters Fi are measured, the fewer wavelengths
are required to guarantee a unique solution close to
the nominal length of the gauge block. Note that in
the National Physical Laboratory (NPL)
interferometer, after equation (3) has been used to
determine the correct solution, the length
calculation relies solely on the red laser wavelength,
as this has a smaller uncertainty than that of the
other laser(s).
The term lt is the gauge temperature
correction, based on T, the departure of the gauge
temperature, tg, from 20 C and the gauges
coefficient of thermal expansion (CTE), ,
(4)
lt TL ,
T 20C - t g .
(5)
The term lw represents the thickness of the

wringing film between the gauge and platen
surfaces. The gauge block length is defined to
include the thickness of one wringing film and so
the expectation value of lw is actually zero.
However, variability of wringing technique between
operators and the different wringing qualities of
platens lead to a variation in the achieved wringing
thickness and hence an uncertainty in lw when
comparing different measurements.
A correction, l, is often necessary to
account for the finite aperture size of the
interferometer optics causing an obliquity
correction, and any residual wavefront errors of the
optics are collected into the term lA.
Due to the measurement taking place in air,
there is a correction, ln, for the refractive index, n,
of the air to account for the difference between the
ambient optical wavelength of the laser light and the
calibrated (vacuum) wavelength
(6)
ln (n 1)L
It is usual to calculate the refractive index, n,
from measurement of parameters including the air
pressure, p, temperature, ta, and humidity, h,
through the use of empirical equations based on the
work of Edn and subsequent re-validations and remeasurements.
- 233 -
Geometrical errors of the gauge block which may

influence the central length determination, such as
imperfect flatness, are accounted for in the term lG.
The differences in surface roughness and complex
refractive indices of the platen and gauge block
surfaces (which account for the difference in the
apparent optical length compared with the
mechanical length) are handled by the so-called
phase correction, l.
Thus, in order to measure the length of the gauge
block, several parameters need to be measured at
the same time, for each wavelength: p, ta, h, tg and
Fi . Prior calibration of each vacuum wavelength, i,
is needed, as well as a sufficiently accurate estimate
of l to allow i to be estimated to within a few
integer orders. Also, the operator (or result
calculation system) needs to have detailed
knowledge of lw, lA, l, lG and l to apply these
corrections accurately, or allow for their influence
in the uncertainty calculation. In early
interferometers, the determination of many
parameters was performed visually by the operatorfringe fraction estimation by viewing a static fringe
pattern through an eyepiece, reading of analogue
instruments for air temperature, pressure and
humidity values. Multiple measurements of these
parameters were required for each gauge block,
requiring constant re-focusing and accommodation
of the operators eyes. Not surprisingly, the work
was tiring and operator-dependent offsets in the
results were not uncommon.
4. Gauge Blocks Calibration by the

Interferometry in the Czech Metrology
Institute workplace, the Regional
Inspectorate of Liberec
Calibration procedure that is mentioned
below is applied on the interferometer NPL TESA
AGI 300. The interferometer runs by means of laser
interference principle with using two He-Ne
stabilized lasers, the red with wavelength 633 nm
and the green one with wavelength 543 nm.
Gauge blocks calibration procedure is given by
the technical standard EN ISO 3650. As a basic
measured parameter there is central length of the
end block, or more precisely deviation of central
length from nominal value:
- central length of gauge block lc (Fig.2)
length of end block from the center of free
allowable measuring area
-
length deviation e from the nominal

length in any point algebraical
difference l ln, where ln is nominal length
of the gauge block, l is length in any point.
Fig. 2. Central length lc and example of length l in

any point of the gauge block measuring area wrung
to the auxiliary plane plate (1)
Fig. 1. Schema of the NPL gauge block

interferometer
Calibration principle of gauge blocks is

given by the interferometry measurement of gauge
blocks which are wrung to the basic plane plate.
Gauge block is placed on the edge of the auxiliary
plane plate by its lapped surface and under the
permanent pressure perpendicular to the second
lapped surface is such gauge block moved to the
required position to properly wring to the plate so
there is no possibility to separate them by lifting but
only by their displacement.
In schematic form in figure 1 is shown one

of the most expanded gauge block interferometer
developed in NPL.
Fig. 3. Gauge block wrung to the auxiliary plate
- 234 -
After wringing of gauge block to the plate

there is measurement of the length of the gauge
block in so-called measurement chamber (Fig. 4)
where interferometer NPL TESA AGI 300 (Fig. 1)
is one of its components.
Fig. 4. Gauge blocks in the measurement chamber

The interferometer is scanning interference
fringe of the gauge block surface and auxiliary
plate. By this plate is continuously rotating so that
there is always just one gauge block in the scanning
area. Surface of the gauge block and auxiliary plate
(inc. interference fringe) is scanned by the camera.
Very high accuracy can be achieved at the
measurement by such method. However there is
very important requirement about wringing of the
gauge block to the auxiliary plate. Great problem of
such calibration procedure deals with the great wear
of the gauge block lapped surfaces due to the
sliding of the gauge block to the auxiliary plate and
their subsequent removal. Due to all of that there is
not only wear of the gauge block and the auxiliary
plate but also rapid lowering of their working life.
In this paper was explored surface from the gauge
block microstructure (microstructural analysis)
point of view at measurement by interferometry.
5. Analysis of gauge block surfaces
For detailed study of gauge blocks surfaces
from the microstructure (microstructural analysis)
point of view during calibration by interferometry
method for measurement there were used gauge
block with the highest tolerance grades grade K
acc. to the standard EN ISO 3650 with nominal
lengths 5; 7 and 10 mm. There was used
microscopy of the surface method which is making
use of the thin stylus which copies surface. AFM
method (Atomic Force Microscopy) is based on the
weak attractive forces (e.g. Van der Walls and so
on). Regarding the setting, such method is able to
detect differences in the chemical composition or

surface relief. More over there was used the optical
microscopy method which enables to observe and
analyze non-transparent microscopic objects and
structures in the reflected light up to the 1000x
magnification without any special modifications of
the microscope and also under common preparation
of samples by the grinding and polishing. [3]
Detailed research was carried out on the
AFM microscope JPK NanoWizard 3 at mode
called as the Noncontact mode. There was also used
microscope Carl Zeiss AXIO IMAGER A2 at
modes DF for the optical part of research.
At given gauge blocks (with nominal
dimensions mentioned above) were for surface
research compared surface roughness. To describe
surface roughness, parameters of Ra , Rq (RMS)
and Rt were chosen. Definitions and parameters
commonly used for surface structure determination
are provided by the international standards.
To objectively evaluate quality of the
investigated gauge blocks surfaces at interferometry
calibration method was firstly carried out
measurement of the roughness (already before
calibration itself) and subsequently after calibration
by AFM method inc. records from the optical
microscopy. The gauge blocks surface research at
interferometry calibration method was carried out
twice under the same calibration conditions and for
the same nominal lengths of gauge blocks.
Table 1
Parameters of the surface
Surface parameters for select lapped area before
calibration on interferometer NPL TESA AGI 300
5 mm
7mm
10 mm *
2,53
2,82
3,82
Ra, nm
3,72
4,93
Rq (RMS), 3,99
nm
20,37
25,81
29,2
Rt, nm
First calibration of gauge block on the
interferometer NPL TESA AGI 300
Surface parameters for select lapped area after
5 mm *
7mm
10 mm *
Ra, nm
Rq (RMS),
nm
Rt, nm
3,72
4,56
2,98
3,85
4,23
5,37
38,60
28,37
43,00
* repeated sliding of the GB on the auxiliary plate
- 235 -
Table 2
Parameters of the surface
Surface parameters for select lapped area before
5 mm
7mm
10 mm
3,79
3,09
4,46
Ra, nm
Rq (RMS),
4,69
3,96
5,51
nm
43,23
31,68
44,10
Rt, nm
Second calibration of gauge block on the
interferometer NPL TESA AGI 300
Surface parameters for select lapped area after
5 mm
7mm *
10 mm *
4,56
6,10
6,98
Ra, nm
Rq (RMS),
5,98
6,99
6,70
nm
41,23
46,32
46,0
Rt, nm
* repeated sliding of the GB on the auxiliary plate
The table 1 and 2 shows overview of the
surface parameters statistical values which were
detected by the AFM method. Its possible to state
that there is increase of surface roughness values
with increasing number of GB sliding on the
auxiliary plate and not only within sliding at
calibration but also with amount of calibrations.
If it is compared for example quality of
surface through surface roughness values after the
first calibration by interference method for nominal
length of GB 5mm, it is obviously that value of Ra
increased from 2,53 nm to 3,72 nm. As well the
value of Rt which increased from value 20,37 nm to
38,60 nm. This is the proof that repeated sliding
(fixing) of GB on the plate resulted in such increase.
In the fig. 5A, 5B, 6A and 6B are shown figures of
GB lapped surfaces of nominal length 5 mm. There
are always figures before and after calibration.
Fig. 5A. The AFM scan of GB lapped area with

nominal length 5mm before calibration
Fig. 5B. The optical scan of GB lapped area with


nominal length 5mm after calibration
Fig. 6B. The optical scan of GB lapped area with

- 236 -
In the next chosen AFM scans (fig. 7A and

7B) is presented BG surface quality with nominal
length 10 mm before and after second calibration.
The measured Ra value from this scan is 6,98 nm.
By means of the AFM application and

optical microscopy was verified that there is at the
interference calibration method heavy wear of the
GB lapped areas.
Increasing of the BG wear was also
observed within extension and increasing of origin
local artifacts amount. Artifacts are objects which
are shown in the figures but they dont exist in real.
5. Acknowledgement
This paper is related to the investigation on

the Research Projects TA: Advanced systems for
length calibration and surface inspection of end
standards, which are supported by the Technological
Agency of the Czech Republic.

References
1. Jonghan J., Young-Jin K., Yunseok K. and
Seung-Woo K., Absolute length calibration of
gauge blocks using optical comb of a femtosecond
pulse laser, OPTICS EXPRESS, 2006, 5968 5974.
2. Lewis A. J., Hughes B., Aldred P. J. E., Longterm study of gauge block interferometer
performance and gauge block stability, Metrologia
47, 2010, 473486
3. Andrsova, Z., Kejzlar, P., Volesky, L.:
Obrazov analza jako podklad pro vyuit UZ
strukturoskopie. In JMO vol. 58,2013, no. 1, p.3741, ISSN 0447-6441.
Fig. 7B. The AFM scan of GB lapped area with

4. Conclusions
From the measured results is obvious that
there is heavy wear of GB lapped areas due to the
sliding and subsequent removal of GB on the
auxiliary plate. This can be stated from the analysis
of the observed surface roughness statistical values
as it is shown in table 1. From the comparison of
surface quality by means of surface roughness
values already at first interference calibration
method is obvious that value of Ra is increasing
with the increasing number of GB sliding on the
auxiliary plane plate.
tepnka Dvokov / Ji Sobotka / Pavel Brdlk

Department of Engineering Technology
Faculty of Mechanical Engineering
Technical University of Liberec
Studentska 2
Liberec 461 17
Czech Republic
Frantiek Dvoek / Lenka Ddkov
Czech Metrology Institute - Regional Inspectorate
of Liberec
Slunecna 23
Liberec 460 01
Czech Republic
E-mail: stepanka.dvorackova@centrum.cz
E-mail: fdvoracek@cmi.cz
08.03.2013 .
- 237 -

TECHSYS 2013
BULGARIA

: (2PL),

.
: , time out
. -
().
2PL
.

.
GPSS World Personal
Version : -
.
: , ,
, , ,

AN ALGORITHM MODELLING CENTRALIZED

TWO-PHASE LOCKING WITH INTEGRATED
MECHANISM OF TIMESTAMPS
SVETLANA VASILEVA
Abstract: In this paper we present an algorithm for two-phase locking (2PL) in which
deadlocks of distributed transactions in distributed database management systems are avoided.
Three methods for solving deadlocks are known so far: Wait-for Graph, time out and timestamp
ordering. The paper considers the method of timestamps as the most appropriate to be realized
in distributed database management systems (DDBMS). From the algorithms for 2PL in
DDBMS, we view the centralized 2PL as a basis for developing the rest of the modeling
algorithms. From the two strategies for timestamps mechanism for deadlocks avoiding, we
consider the wound wait strategy. There is presented a comparative variant of the results
from the modeling in GPSS World Personal Version environment of the two algorithms: waitdie method and wound wait.
Key words: distributed databases, distributed transactions, concurrency control, centralized
two-phase locking, deadlock, timestamp ordering
- 238 -
1.

()
()
.

(
) (
),
. [1],
[2], [3], [5] ,

(Two-phase locking 2PL)
- .
[1], [2],
[3], [5] [7].

(deadlock - DL),

[5] .
2PL
:

DL.

(Wait-For Graph
(WFG) based).
time-out (
,
, )
(Timestamp (TS) based)

, -
.

[1]
[6], rollback
,

. timeout
( , ,

), -

.
2.

(..
) timestamp ordering
:
([1] [5] ) -
. ,
,

-
[1].

[1] [3]:
- wait die,
i e - j,

(TS(Ti)<TS(Tj)), i
, i e - j (TS(Ti) > TS(Tj)), i
e . 2PL
wait
die [9]. . 1
,
Ti,
.
. 1. wait die

i
- wound wait, e
:
,
Ti -
Tj, Ti Tj. [1]
- Tj
- 239 -
,
. Tj,
(rollback). Ti Tj, Ti
. . 2

wound wait,

Ti.
. 2.

i wound wait
3.
2PL
(wound wait)

GPSS World
2PL
(wound wait) 6
[8] [9]

2PL .
2
1 ( -
, 1 ).
.

,
:
P1 - ;
P2 - , ;
Pel1 - ,
(El1);
Pbl1 - El1: 1
(rl) if (El1); 2 (wl) if (El1);
Pel2 - ,
(El2);
Pbl2 - El2: 1
(rl) if (El2); 2 (wl) if (El2);
P5 - :
0

read/write 1;

5=3;
P6 - ,
El1;
P7 - ,
El1;
P8 - ,
El2;
P9 - ,
El2;
P11 - ,
;
Pvr1 - ,
Pvr1=1;
Pvr2 - ,
Pvr2=1.

2PL
(wound wait)
. 3.
TP2P1
P2
(1 2
) - 1 . 3

( 8 . 3).
9

DMP6, (DMP7), (DMP8
DMP9), TP2,P6P1, (TP2,P7P1),
(TP2,P8P1 TP2,P9P1)
/
El1 El2. P2

LM0 ( 2 5 . 3).

LT0

( 3 . 3),

LT0.
( 4)
,
P2,
P2
LM0 El2 (
5 . 3).
- 240 -
El2 ,

El2
( 6 . 3) TMP2
El2
( 7).

,
-
, :

TP1P2 - (-),
,
LTA (
)
,
;

TP1P2 - (-)
(
LTA ,
),
,
.

,

/ (
10 11 . 3).
12
/
( 12 . 3),

2 ( 13 14 . 3).
,

( 15),
LM0. 2
l1/El2
( 16 .
3). (
, ..
El1) ( 17),
2 12
( 18 . 3). 2
- ( 19)
write, 12
LDBP6 LDBP7 (
LDBP6 LDBP7).
rollback ( 20)
TP2,P6P1, (TP2,P7P1), (TP2,P8P1 TP2,P9P1)
, GBDA1
GBDA2,

2 ( 21 22 . 3).
,

2

12 (. 3) 23, 25
27 ( El1) 24, 26 28 ( El2).
,
/,

( 25 26 . 3).

27 28.

,
,
.
, .
. 3.
2PL wound
wait

El1 El2
TEl1P1 PEl2P1
-
( 27 28 . 3).

,
( 29).

2PL
- 241 -

: 2PL, 2PL
2PL .
4.
,
,

2PL
wound-wait
. (Tn)
GPSS World
.

.
. 4

wound wait

( ): ,
6 ,
4,17
tr/s
(
25 tr/s);
, 6 ,
8,33 tr/s (

50 tr/s); ,
- 6
tr/s
(
16,67

100 tr/s).
120
100
80
60
25 tr/s
40
50 tr/s
20
100 tr/s
0
0
14
29
300
Nc
Tn
(1),
N
,
(Tn).

Ps
2 [3]:
Ps
Nc
Ng
(2),
NC
, Ng
.
. 5

2PL
wound-wait
(
. 4).
2PL "wound - wait"

Ps
1
0,9
0,8
0,7
0,6
0,5
0,4
0,3
25 tr/s
0,2
100 tr/s
50 tr/s
0,1
0
0
14
29
300
600
1200
Tn (s)
. 5.
a 2PL
wound-wait
2PL "wound - wait"

a
tr/s
600 1200
Tn (s )
. 4.
a 2PL
wound wait

,
[3, .71].
1:
. 6

: wound wait
wait die.

:

;
;

.
- 242 -
110
Tr/s100
90
80
2PL_WD
70
2PL_WD
60
2PL_WD
50
2PL_WW
40
2PL_WW
30
2PL_WW
1.
.
Deadlocks.
, RSDN
Magazine #5, 2003, http://sasynok.narod.ru
/index.htm?omvs.htm.
2. -, ., . , .
, , , 2003.
20
10
0
3,6
4,5
7,2
14,4
28,8
300,0
600,0
1200,0
Tn
. 6.
wound-wait wait-die
2PL ,

. 6
2PL

,

.
5.

:
-

2PL
wound-wait,
.
( ,
.)

;
-
2PL ,

;
- 2PL
wound-wait
,
2PL
wait-die;
-

2PL
(,
)
(,
wound-wait wait-die)
,
.
3. , ., . , ,
- -, 2003.
4. , . GPSS World
, , , 2004.
5. , . . ,
, CIT FORUM, 2008, http://www.
citforum.ru/database/advanced_intro/43.shtml
6. , ., .
, ,
, 2003.
7. Krivokapic, N., A. Kemper, E. Gudes Deadlock
detection in distributed database systems: A new
algorithm and a comparative performance analysis,
http://masters.donntu.edu.ua/2005/fvti/kovalyova/li
brary/d1.pdf.
8. Vasileva, S., A. Milev Simulation Models of
Two-Phase Locking of Distributed transactions,
ACM, New York, Vol. 374, Article No. 74, 2008.
9. Vasileva, S. An algorithm for deadlock avoiding
in distributed database management systems. //
Journal of the Technical University Sofia, branch
Plovdiv. Fundamental Sciences and Applications,
Vol. 16(1), International Conference Engineering,
Technologies and Systems TechSys 2011, 2011,
pp.377-382, (In Bulgaria)
College - Dobrich
Shumen University Konstantin Preslavski
12 Dobrotica
9302 Dobrich
BULGARIA
E-mail: svetlanaeli@shu.bg
06.03.2013 .
- 243 -

TECHSYS 2013
BULGARIA

,
:
, dI/dt
MOSFET .
.

.
.
.
: , MOSFET , ,

CONTROL OF DC MOTORS WITH ACTIVE DRIVER

CIRCUIT
SVETOSLAV IVANOV, IVAN TANEV
Abstract: This article presents research on impulse control of a DC motor with active driver
with feedback dI / dt in the management of powerful MOSFET transistor. Composed is the
equivalent circuit of the controlled field-effect transistors and active inductive load. Analytically
explore the composition of equation is the transfer function of the equivalent circuit. Presents
oscillograms of the transitional processes in experimental research. With the use of this driver
in the management of power electronic devices will improve their electromagnetic compatibility.
Using this driver in the management of power electronic devices will improve their EMC.
Key words: power semiconductor devices, MOSFET driver, Active control, electromagnetic
Compatibility (EMC)
1.

,

.
MOSFET dI/dt

[3, 4, 6].

[1].

MOSFET

- ,
- 244 -
. ,
,
" " ,

[2, 5]. . 1
,

. ,
(6,5 kHz)
-20 dB/decade,

(10 kHz).
,
- 40 dB/decade,
60 dB/decade.
. 1.

.

,
MOSFET .

.
2.

. 2.
TTL

,
.

Rs.

- PIC 8 6/2.5.
U = 25V,
250 Hz,
D = 70%.
. 2.
,
.

- dI dt .

[7].
E = 15V.

R = 3,7 , L = 4,7
mH.

-
,
, . 3.
Iin(t)

,
.
(G = 1).
,

,
K.
Iout(t)
.
. 3.
-
.
- 245 -

:
F
G
,
dI (t )
1 K . out
dt
(1)
G . I in t
I out (t )
(2)
dI (t )
1 K . in
dt

:
I out t
R
.t
U E
.1 e L
R
U E RL .t
1 K .
.e
L
(3)
:
U
;
E

;
R
;
L
.
. 4.

.
3.

TC4421
,
. 5.

K = 0,0001.

.4.
2,77 mS.

. 4.
1

.

0,0001 1,
.

Maple.
. 5.
TC4421 1,
- 2.
- 246 -
. 6

1 2.

dI/dt -

.
- K
,

- 1A.

3A.
1. Andreycak W., Practical Considerations In

High Performance MOSFET, IGBT and MCT Gate
Drive Circuits, Untried Corporation, Application
Note U-137.
2. Boora A., Arash and Zare, Firuz and Ghosh,
Arindam (2009) Efficient Voltage/Current Spike
Reduction by Active Gate Signaling. In:
Proceedings of the Electromagnetic Compatibility
Symposium, Adelaide 2009, 16-18 September 2009,
Lakes Hotel Resort, Adelaide.
3. Ivanov Svetoslav, Neichev Ivan, Tanev Ivan,
Active control of a driving circuit by a derivative
with respect to DI/DT of the output current,

, RES-OP-DEV MISETC code:222, , 14-15 2012 . . 200205.
4. Musumeci, S.; Raciti, A.; Testa, A.; Galluzo,
A.; Melito, M.: Switching Behaviour Improvement
of Insulated Gate Controlled Devices; IEEE
Transactions on Power Electronics, Vol. 12, No. 4,
1997, pp. 645-653.
5. Neacsu D., Active Gate Drivers for motor
control applications, IEEE PESC 2001,
Vancouver, CANADA, June, 17-21, 2001.
6. Shihong Park, Thomas M. Jahns, Flexible
dv dt
di dt
. 6.
1,
2.
and
Control Method for Insulated
Gate Power Switches, IEEE TRANSACTIONS ON
INDUSTRY APPLICATIONS, VOL. 39, NO. 3,
MAY/JUNE 2003, pp.
4.

-
.

,

,

.

,

.
7.Van den Bossche, A.; Valtchev, V.; Clotea, L.;

Melkebeek, J.: Fast Isolated IGBT Driver with
Desaturation Protection and Fault Feedback; EPE
European
Power
Elcectronics
Conference,
Lausanne, 1999; Poeceedings on CD.

4000 Plovdiv
BULGARIA
E-mail: etehsv@gmail.com
E-mail: mastervanio@yahoo.de
22.03.2013 .
- 247 -

TECHSYS 2013
BULGARIA

,
: -
,
(~ /)
F- .
,
,
. , .
,
.
, .
: , ,

ANALYSIS AND SIMULATION INVESTIGATION

OF DOUBLE PULSE FLASH-LAMP PUMP
LASER ELECTRICAL SYSTEM
SVETOSLAV IVANOV, YANCA KISSOVA
Abstract: The important type of lasers for application in LIDAR technique for atmosphere
pollutant monitoring, for spectroscopy and medicine are the flash-lamp pumped tunable and
high output energy (~Joule/s) dye lasers and F-color center lasers. The very convenient
technique to obtain an efficient their pumping is so called double pulse method, combining
the preliminary low power and slowly rise discharge through the flash-lamp on which is
superimposed the main, high power discharge. This technique is successfully applied using a
simple auto-synchronizing electric circuit, however without the theoretical and experimental
investigation of this circuit. In this work we present the theoretical modeling of such electric
system and its computer simulation analysis. The results are important to better understanding
the action of the system and its optimization.
Key words: xenon flash-lamp, optical pumping, electrical system double pulse type,
theoretical study.
- 248 -
1. Introduction.
The lasers are useful tools for many
practical applications in modern technologies, in
local and distance atmosphere pollutants
monitoring, in spectroscopy for detection of
different composant of the sample under
investigation, in metrology, in medicine some
applications especially spectroscopy. For spectral
selective action on the matter and in LIDAR
atmosphere monitoring, it is very useful the tunable
laser light and also with high energy (~ Joule/s) and
power (~ MW). The important type of lasers that
are convenient to assure such type of emission is the
flash-lamp pumped dye lasers and F-color center
lasers [5,2]. The working scheme of such laser is
the type of without metastable upper laser level
with a ~ 10-8s lifetime , that leads to necessity of
excitation with very short rise time (~ 1 s and less)
front and pulses [3,2,4]. For the flash-lamp
pumping, this needs of special, usually home-made
flash-lamp with a very low lifetime and low
efficiency. The use for this aim of the widespread
and commercially available flash-lamp needs its
connection in a special electric system that permits
to obtain the noted up lamp-emission parameters.
The very convenient system is so called double
pulse that is based of intensive and longtime rise
preliminary electric pulse, providing intensive
ionization of the working gas in the lamp. With
convenient delay, on the preliminary pulse is
superimposed the main, high energy electric
discharge. The intensive preliminary gas ionization
leads to two favorable effects: 1) before the main
pulses the gas is intensively ionized and does not
need of time for ionization by the main pulse; 2) the
relatively low power and long-time developed preionization prevent the flash-lamp destruction by the
intensive shock wave if the main, high energy and
quick rise time main pulse is applied. The
application of such type of exploitation of the
standard flash-lamps lead to increasing the dye laser
output energy few times in comparison with direct
application of the main pulse and to prevent the
lamp destruction. In the literature [7,5] and in the
some flash-lamp dye laser construction is proposed
and successfully used an auto-synchronized electric
system for a flash-lamp double pulse. However,
the system is realized only practically without the
theoretical treatment. The last, except the theoretical
understanding of the scheme action, is needed,
especially, to evaluate the optimal system
parameters and values of the elements for maximum
efficiency of the lasers. Such theoretical modeling,
analysis and computer simulations are the goal of
this work.
2. Description of the scheme for electrical

excitation at double pulse
Fig.1 illustrates the scheme under
investigation, in which the discharge flash-lamp
used is filled with the gas xenon under
comparatively low pressure of a few hundreds of
Pascals.
Fig. 1. Double pulse scheme

The cylindrical tube of the lamp is made of
quartz with usual thickness of about 1mm. The
flash-lamp works in the following way: the
accumulating capacitor is connected to the
electrodes of the flash-lamp. In order to cause
discharge between the electrodes by means of an
additional ignition device, connected to the flashlamp, an impulse is fed, which pre-ionizes the gas
[6]. The ignition device is presented in the scheme
both by the pulse generator and by the switch S2.
The resistance of the lamp decreases and the
accumulating capacitor discharges through it. The
discharge forms pre-ionizing preliminary low power
pulse, which provides for the intensive ionization of
the lamp. Its duration is determined by the capacity
of the accumulating capacitor and the inductance of
the discharge circuit L1. The breakdown voltage of
the discharger, presented in the scheme by S1 and
R4, is selected so that the voltage between the
electrodes of the C1 before its discharge is smaller
than the breakdown voltage.
During the process of discharging the
voltage between the electrodes of the discharger
continuously increases, reaching at some point in
time the required value for ensuring the breakdown.
Thus the second major high-power pulse begins
with a steep front ( 5s). The intensive preliminary
discharge protects the flash-lamp from destruction
and provides a very steep front of the main pulse
[4]. In Fig.1 the flash-lamp is presented as a coil
with inductance L2, resistance R2, and capacity C3,
connected to it in parallel.
In Fig.2 the shape of both the voltage and
the current through the discharge pulse flash-lamp
is shown. It can be seen that at t=40s the current in
- 249 -
the circuit reaches its established value. Since the

time constant is 4 times less than this time, at R2
=1 the following value of the inductance is
obtained: L2=10H. The internal resistances of the
voltage sources V1 and V2 are denoted by R1 and R3
(Fig.1).
3. Results
Fig. 3 shows the change of the current
through the pulse flash-lamp IL2 and the shape of
the voltages across the capacitors C1 and C2,
connected in parallel to the supplying sources: UC1
and UC2 . The voltages of the supplying sources and
are 5kV and 10kV, respectively. The simulations are
realized in Multisim.
Fig. 2. Voltage and current shape through the flashlamp

The duration of the pulse from the source,
needed for ignition of the lamp, is 26s. The delay
of the pulse from the source, which is 5s,
simulates starting of the discharger. This is the time,
during which the capacitor discharges through the
flash-lamp and the voltage of the discharger
increases until it reaches the required value for a
breakdown. This value is 7kV in our case.
The volt-ampere characteristic of the
discharge pulse flash-lamp can be expressed by the
formula [5]:
V = K 0 (i )
0, 5
(1),
where V is the voltage of the flash-lamp in volts, i

is the current of the discharge in amperes, and K0 is
the impedance of the flash-lamp. The equation is
valid for current density over 500A/cm2. The
parameter depends on the length of the flash-lamp,
on its internal diameter, on the gas, and on the
pressure in the flash-lamp.
p
K 0 = 1,28
450
0, 2
l
d
(2),
where P is the pressure in the lamp in torr, l and d

are correspondingly its length and its internal
diameter, measured in mm. Formula (2) is valid
when the gas in the flash-lamp is xenon.
Fig. 3. Graphs of IL2(t),UC1(t) and UC2(t)

At t = 0 the supplying source V1 is turned on
by means of the switch S2, controlled by the pulse
generator V3 with duration of the pulse 26s. After
a period of time, equal to 5s, potential difference is
established between the capacitors C1 and C2,
having the value of 7kV. This is the time of delay T delay, after which the switch S1 is turned on. By
the time of delay and turning on the switch S1, the
discharger is stimulated.
From the graph in Fig.3 the change of the
current through the pulse flash-lamp IL2 before and
after turning the second supplying source V2 on can
be seen. At the moment of turning on the
discharger, the current through the pulse flash-lump
IL2 has reached the value of 670 . The amplitude
value, reached by this current at the second current
pulse, is approximately 3,8. The recovery time of
the initial values of the voltages on the capacitors
UC1 and UC2 is 1,67ms. At the end of the current
pulse through the flash-lamp, the voltage UC1=7,4kV, and UC2 =180V.
Fig. 4 illustrates the change of the currents
through the pulse flash-lamp IL2, the coil L1 and the
discharger IR4.
- 250 -
interesting point that follows from the analysis is

that during two discharges, part of the energy from
the main capacitor is transferred to the first. It is of
interest to obtain the conditions to decrease this
effect, leading to loss of part of the storied energy in
the main capacitor and thus decreasing the
efficiency of the laser excitation. One point of the
optimization in the future development is to find the
conditions to decrease or completely eliminate this
non-desired transfer of energy this transfer. The
modeling we modeled the electric system and made
its computer simulation analysis. The results are
important to better understanding the action of the
system and its optimization.
Fig. 4. Graphs of IL1(t),IL2(t) and IR4(t)

Up to the moment of turning on the
discharger, the same current flows both through the
pulse flash-lamp and through the coil L1. After
turning on the discharger, the current through the
pulse flash-lamp IL2 increases incrementally,
following the change of the current through the
discharger IR4. Part of the current flows through the
coil L1 and chargers the capacitor with voltage up to
UC1=5,9kV. The amplitude value of the current
through the discharger is IR4=4,6kA, while the
amplitude value of the current, charging the
capacitor C1 is IL1=730A.
From the conducted investigations it can be
seen that the minimum period for turning on the
flash-lamp is calculated by the formula:
Tmin=tp+trecov.
(3)
and by the equality Tmin=1,7ms. In formula (3) tp
and trecov. are correspondingly the duration of the
pulse and the recovery time of the flash-lamp.
Consequently, the maximum frequency of operation
of the flash-lamp with a double pulse, calculated
by the formula
f max =
1
Tmin
(4)
And fmax=588Hz.
4. Conclusions
In this work we present the theoretical
modeling of electric system of type double pulse
for flash-lamp pumping of lasers. We considered
and modeled a simple auto-synchronized scheme
that consists of two connected in time sub-circuits,
providing related oscillations of the current in the
flash-lamp. The computer analysis is in accordance
with the preliminary given experimental data and
shows the essentially decreasing of the rise-time of
the start front of main pulse. This decreasing does
not depend essentially for some variation of the
moment of switching of the main pulse. The
1. Bassov Y.G. ,Systems for flash-lamp pumping

of a dye lasers, review article, Sov. J. Prib. And
Technics of experiments, 6,1986,3-27.
2.Demtroder W., Laser spectroscopy: basic
concept and instrumentation, 2003, 3-th edn.,
Springer, Germany
3.Deneva M., Uzunova P.,
Nenchev M.,,
OPT.QUANT. ELECTRON. (2007)39;193-212,
(Springer, West. Eur). Tunable subnanosecond
laser pulse generationusing an active mirror
concept;
M.Nenchev,A.Gizbrekht,
Opt.
Communs.,45(1983)266;
4.Genov L. (1991), Theoretical fundamentals of
Electrical engineering, Tecnique, Sofia
5.Koechner W and Bass M. (2003), Solid-state
laser, A graduate text. New York, Springer-Verlag.
6.Nair L.G. Prog.Quantum.Electron.,7(1982) 153268
7.Nenchev M., Stefanov V.., Bulg.J.Phys.,4,(1977),
14-21 ;, IIR Bulg. Patent No 2738/1975 (AC),
System for producing of discharge with high
intensity preliminary discharge in flash lamps
4000 Plovdiv
BULGARIA
E-mail: etehsv@gmail.com
E-mail: yankakiss777@yahoo.com
22.03.2013 .

TECHSYS 2013
BULGARIA
EXAMINATION OF THE DISCHARGE OF GASES

USED IN THE EVENT OF HIGH VOLTAGE
TANER DINDAR, N.FUSUN SERTELLER, GOKHAN GOKMEN, TAHIR CETIN AKINCI
Abstract: In this study; different gases have been studied experimentally discharge events. The market is
currently the most widely used and much of the electrical insulation gases in the atmosphere. The mostly
used gases are Nitrogen (N2), Carbon dioxide (CO2), Freon (CCl2F2) and Sulfur-hegzaflorid (SF6). In this
paper, the gas discharges are reviewed by the help of the Townsend Theory and induction coil. While, the
pressure is kept 16 mmHg and the glow discharge is observed in the tubes. During the colour spectrum of
radiation in tubes, the type of gas is identified. With the help of the gases in the tubes, built the high voltage
electrodes applied to the induction. Gas type and the emergence of a discharge voltage are according to the
size of the glittery event. According to the rays, which according to the colour of the gas so that the gases.
Keywords: Theory of High Voltage, Discharge Event, Townsend
1. Introduction
The need for electrical energy in the world
raising rapidly, the amount of electrical energy
produced is also due to the limited natural
resources, economic and consumer centres should
reach at least loss. The loss of electrical power at
long distances so that you can move, ability to move
large forces built more economically to remain
small, big powers are used to provide high voltage
conductor. In these cases, the use of high-voltage
electricity transmission and distribution.
High-voltage and high-voltage equipment
insulation is a very important problem appears to be
moving. Used in the gaseous, liquid and solid media
or the insulators are also due to the isolation of the
electrical strain problems, problems of the current
arc extinguishing and cutting. For these reasons it is
very important that the insulation. Gases, liquids
and solids are the discharge event can occur, but
these studies on the discharge phenomena have
been studied.
High-voltage devices are used as insulating
gases expected properties; high electrical
breakdown strength, coefficient of thermal
conductivity, specific heat, gas viscosity coefficient
depends on factors such as the ability of the gas
cooling and arc extinguishing to different
temperatures, high pressures, what should be the

temperature as low as possible, other gases,
liquefaction, chemical reaction, not losing, toxic
chemical properties during the ejaculation events
even in the absence of such features, Flash [1].
In this study, the purpose of the
examination of the discharge event in gasses:
Gasses will be how many volts high voltage is
applied to the number of seconds in which gas will
be the conductor and the conductor which is to
observe the glow colour to do it. Such a work is
used for gases in Geissler Tubes. The voltage
applied to the discharge of gases in the event of the
tubes, the tubes depends on the type of gas,
pressure, etc. Within the values according to the
values of this discharge is observed how the event
(in the colours of the tubes) [2].
Can Discharge events;
a) Townsend Discharge Theory
b) Induction (Rumkorf)
c) Geissler Tubes
Tension, when applied to an excellent
insulating gas undergoes a change occurs in various
reactions. Therefore the value of the voltage applied
- 252 to the electrodes with is very important. Applied

voltage is low, a small current flows between the
electrodes. However, the value of the current value
of the voltage applied to the electrodes with large
increases very quickly and discharge (electrical
conductivity) begin observing the event. Here is the
current period together with the size of the current
applied is important. This event is shown in Figure
2.1.
Fig. 1.In the puncture of the experimental narration

(Townsend Theory)
1.
Theory of Townsend Discharge
Due to the high voltage discharge in event

is quite important. So i need to know which factors
influence to discharge gases. These are all difficult
to understand if the event more than one will be
examined. Therefore, all factors should be
examined step by step.
First, there is the gas pressure p is in the
average speed of the molecules and the electrons in
an electric particle ever average thermal speeds
have been considered to be too large. Accordingly,
an electrical current, however, the surface of the
cathode electron tearing (cathode is heated or
photon of attack) is possible. This is called the
initial electron electrons the cathode surface torn
off. Starting with the cathode surface ionization
electrons in thermal or photo with zero speed is
accepted, the cathode surface current density shown
in Equation 1.
n.q
s.t
(1)
In The Equation; n, from the surface of the

cathode, which is the area for the duration of the t, s
is the number of electrons in an electron, q electrical
load.
There are a number of great importances in
the event of the discharge of electrons from
ionization device. According to this number of
acceptances to: determine the Townsend Electron
kinetic energy following the forced ionization of

gas ionization energy is small, does not occur;
greater income. He forced the electron loses all the
kinetic energy of each and the new begins with zero
movement speed. The electron moves along with
field.
2.
Electric Field
Correctly
Breakdown
Voltage
A discharge event to feed itself, the cathode

to the anode of a free electron, ' is a replacement for
the side events should occur in the electron itself.
To do this, the impact of positive ions from the
electron avalanches occur, you must reckon with.
These effects are as follows:
a) The rupture of the electron gas molecules
by collision. ( Events)
b) The surface of the cathode electron tearing.
( events)
c) Positive and negative charges photon
energy released during the reunification of
the gas molecules, electrons from the
cathode, or taken away. ( Events) These
events influence the outcome is obtained
from equation (2). This is the effect >>
events. is the very large events is true,
lightning discharges can be an example of
electrode F11.
Ud
A.U i .p.a
A.p.a
ln
1
ln 1 y
(2)
Here, Ud: Gas Breakdown voltage, A gas

constant (1/cm-mmHg), B the gas constant (V/cm.
mmHg), p: pressure (mmHg), a: between the
electrodes clearance (cm), : Fixed issue (1/50
1/5000), Ui: ionization voltage.
Can puncture voltage discharge events,
good to know how important it is to know the
discharge process is as important as the doe.
Equation (3) can the duration of the discharge.
Equation 3, a span between the electrodes, di: ion
acceleration
(cm/sec/V/cm),
E d:
Puncture
Resistance (V/cm) shows [3-5].
td
a
di .Ed
(3)
- 253 -
3.
Induction Coil
Induction coils wrapped around an iron core

are isolated from each other, the two windings.
Forming the primary circuit coil is wound and thick
at the secondary circuit is a circuit and very thin
wire-wound. Electromotive force (EMF) of the
secondary circuit is huge. This circuit is connected
to the bypass of a spark between the ends.
Under the normal conditions, not enough
free electrons and ions in gases. Therefore, do not
conduct electricity in gases. Without that gas
pressure within the tube as you lower the ions,
making it easier to reach the anode and cathode to
crop up. Thus the conductivity is achieved. Pressure
is applied to the electrodes with high-voltage
induction reduced tubules within a glow tube roller
conveyor. Without neutralization of ions in the gas
pressure inside the tube has to be deducted is easier
to reach to the anode and cathode. Thus the
conductivity is achieved. Pressure is applied to the
electrodes with high-voltage induction reduced
tubules within a glow tube roller conveyor. Gas
type and pressure changes in the tube also change
the colour of light. Induction coils applied to the
power supply 6-12 volts voltage serves as an
upgrade from 2000 to 50,000 volts [6.7].
a)
b)
Fig. 3. a) Study into the tubes (red), Figure b) work
of the tubes (light purple colour)
a)
Geissler tubes animate the gas ions and free

electrons between the electrodes (Townsend
theory). These particles during movement of ions
and electrons in the gas molecules, creating a new
avalanche start multiplying provide. The event is a
glittery discharge. Tubules are around 16 mmHg
gas pressure. The colour can be determined
according to the spectral lines of type during
irradiation. The result of the experiment of Geissler
tubes colour rays in Table 1.
Table 1
According to the colours of Gases
Purple
N (Nitrogen), Ar (Argon)
Red
Ne (Neon), Helium (He)
Orange
Ar (Argon), N (Nitrogen)
Light Purple N (Nitrogen), Ar (Argon
b)
Fig. 2. a) Induction coil, b) Geissler Tubes
- 254 Conclusion
The application study of the theory of
breakdown voltage and puncture while Townsend
by using formulas examined the discharge of gases
in the events.
In this experiment, the low pressure gas
discharge event or gases more easily penetrable and
the colours of the different gases gas drilled his
radiation to which it belongs.
Acknowledgements
This study is supported by Marmara University
Scientific Research Projects Commission.
References
1. Senoy V., A. Kuntman Engineering Science
Young Researchers Conference MBGAK 2005 17
November 19, 2005, 205-212.
2. Kueffel E., M. Lee High Voltage Engineering,
Pergamon Press, 1970.
3. Nasser E. Fundamentals of gaseous ionization
and plasma electronics, New York, WileyInterscience [1971].
4. Kreuger F.H Partial Discharge Detection in
High-Voltage Equipment, Butterworth & Co., New
York, 1989.
5. Brown, S.C Basic Data of Plasma Physics The
m.i.t. Press, 1959.
6. E Kuffel, Zaengi W.S., Kuffel J High Voltage
Engineering, Fundamentals, 2006.
7. E Kuffel, Zaengi W.S., Kuffel J High Voltage
Engineering, McGraw-Hill, 1984, USA.
TANER DNDAR
Ankara University, Nallihan Vocational
Scholl, Nallihan-Ankara-TURKEY.
N.FUSUN SERTELLER
Marmara University, Technology Faculty,
Department of Electrical & Electronics Eng.
Istanbul-TURKEY.
GKHAN GKMEN
Marmara University, Technical Education
Faculty,
Department
of
Electrical
Education, Istanbul- TURKEY.
TAHR ETN AKINCI
Kirklareli University, Engineering Faculty,
Department of Electrical & Electronics
Engineering, Kirklareli- TURKEY.
E-mail: taner.dindar@ankara.edu.tr,
E-mail: fserteller@marmara.edu.tr,
E-mail: gokhang@marmara.edu.tr,
E-mail: cetinakinci@hotmail.com
15.03.2013 .
- 255 -

TECHSYS 2013
BULGARIA

, , ,

:

.

, pH
, .
: , , PCA,
ALGORITHM OF RECOGNITION AND

VISUALIZATION OF THE RESULTS OF
DIAGNOSTIC TESTS
TANYA TITOVA, KAMELIYA VELCHEVA, VESELIN NACHEV,
CHAVDAR DAMYANOV
Abstract: The purpose of this paper is to present an approach for the formation and
visualization database collected from the experimental spectral characteristics of the diffuse
reflection of citrus juices. With the appropriate tools to examine the relationship between the
spectral properties of citrus juices and data content of total solids, pH and vitamin C, produced
using standard methods for their determination.
Key words: citrus juices, spectral characteristics, principal components analysis, regression
1.

. - ,
,
,
.

,
- .

,

.

.
-

-

.
- 256 -

-

,
.

.

- [2,8].

,

, pH
.
2.
2.1.

37
,
11
.
,

, pH .
2.2.

.

,

.
() - k

.

.
k
.
-
, -
[2,4].

OPTi Duo
Bellingham and Stanley.
( Brix)
6.9-14.0
10.45
9.5-10.2
9.85
6.0-12.8
9.4
7.3-7.8
7.55
2.3. pH
pH

. pH

.
,
pH.

[2,6].
pH
pH FiveEasy FE20
Mattler
Toledo.
pH
2.
2
pH

pH
2.95-4.14
3.55
3.7-3.87
3.79
3.07-3.39
3.23
2.25-2.33
2.29
2.4.

,
.
,
.

. -

[3].

(VWR-Hitachi HPLC System)
L-2455.

RP18 2504,6mm.

22,
- 257 -
3

(mg/dm3)
18.56-583.0
300.78
190.9-191.0
190.95
63.7-289.0
176.35
505.9-506.0
505.95
2.5.

,
,

.

,

.
UV/VIS/NIR
,
,
-

.
,
VIS-NIR
-

.
,
.
VIS-NIR
, -
, ,
,
,
[4,7].

,
,
,
, , , ,
,
,
.

USB4000 VIS-NIR
Ocean Optics.
(SpectraSuite),
.

,
,
.
,
.
. 1.

9 1 ,
.
(FO-THLS-3100, Ocean
Optics) 15 ,

10 45.
300-1100nm (. 1).
60
50
40
30
Reflection, %
0.5ml/min 1.8%
(CH2)4O 0.3%
.
244 nm .
20
10
sok portokal
portokal
greipfrut
sok freipfrut
mandarina
limon
-10
-20
400
500
600
700
Wavelength, nm
800
900
1000
. 2.

. 2

,
. . 2
,
- 258 -
E2 = E1 t2 p2T .
PCA.
35
30
Eigenvalues
25
20
15
10
8
10
12
14
Principal Component Number
16
18
20
. 3.

. 3. ,
8 .

.
,
-

(..
)
.
.

,
[5,6,8].
0.5
PC 5 (1.28%)
.
400-500 nm,
.
2.6.
(PCA)
PCA (Principal Component Analysis)

,

-.
, PCA-
,
-
.
(PC).
, PC
- ( -
),

(, PC)
..

,
( PC

).

.
A = min (n, m ) ,
(2.1)
: A - ; n ; m -
.
:
(2.2)
X = TP T + E ,
: X (n m ) ;
T -
(n A) ; P -
(m A) ; E - (nm).
:
X = t1 p1T + t 2 p 2T + ... + t k pkT + E
(2.3)
: ta -
PC (n 1) ,
a = (1... A) ; pa -
(m 1) ;

:
1. t1 p1 X .
2. PC1 X :
E1 = X t1 p1T .
3. t2 p2 E1 .
4. PC2 E1 :
-0.5
0.4
0.2
0
-0.2
PC 2 (18.49%)
-0.4
-0.4
-0.2
0.2
0.4
0.6
PC 1 (63.34%)
. 4.

2.7.
.
-
,

(rincipal Component
Regression - PCR) (Partial Least Squares
Regression - PLS).
.
(PCR)
- 259 -

Y = XB :
1
B = W (P T W ) Q T ,
(2.9)
: B

( p q ) ; W -
( p A) .
PLS
PCR : U - ;
Q - ; W - .
PLS-
PCA . ,

,
X Y .
PLS-
X Y
, T . X
Y . U
Y , X
. Q
Y - T . W
X Y ,
U - .
W ,
,
.
W
e Y .
PLS

pH ,
, . 5.
Eigenvalues and Cross-validation Results
100
95
Cumulative Variance Captured (%)
PCR ,

.

.
Y = Tq + ,
(2.4)
: Y -
; q -
( A 1) ; (n 1 ) .
:
1
q = (T T T ) T T Y
(2.5)
: q - Y -
.

:
b = P (T T T )P T X T Y .
(2.6)
A
p ,
PCR
Y = Xb ,
(2.7)
: Y -
(n 1 ) .
PCR
,
A
,
Y [1,8].
. -
(PLS).
PLS-
PCR-
Y - .
-
[1].
PLS
PCA Y X , Y X .
PLS
X ,
Y . PLS
PCA.
PLS-
.
PLS
: X = TP T + E , Y = UQ T + F
(2.8)
: Y -
(n q ) ; q -
; U - U (n A) ; A - PLS; Q - Y -
(q A) ; F - (n q ) .
90
85
80
75
70
4
5
6
7
10
. 5.

3.
PCR PLS
- 260 -
,
(RMSEC)
(RMSEP).
nv
RMSEC =
(Y Y )
i =1
(Y Y )
RMSEP =
i =1
dv
: Yi
;
Yi
; n
; nv
(,
);
d v = nv C ; ( A
d = nC A;
.)
4

PCR PLS
data1
data2
data3
6.5183
0.3035
1.6705
RMSEC
7.0367
0.3260
1.5577
RMSEP
1. ., . , LS-SVM

, . , .10, .47-53, 2008
2. Crnara M., C. Dez ,E. Torija Chemical
characterization of pineapple juices and nectars.
Principal components analysis, Food Chemistry,
Volume 54, Issue 1, Pages 93100, 1995.
3. Gazdik Z., O. Zitka, J. Petrlova, V. Adam, J.
Zehnalek, A. Horna, V. Reznicek, M. Beklova, R.
Kizek Determination of vitamin C (ascorbic acid)
using high performance liquid chromatography
coupled with electrochemical detection, Sensors,
7097-7112; DOI: 10.3390/s8117097, 2008.
4. Gomez A. H., Y. He, A. G. Pereira Nondestructive measurement of acidity, soluble solids
and firmness of Satsuma mandarin using VIS-NIRspectroscopy techniques, Jornal of Food
Engineering 77, 313-319, 2006.
0.55
5. Jolliffe I. T. Principal component analysis,

Springer, second edition, 487 p. 28 illus, 2002.
0.5
0.45
0.4
6. Karovicova J., Z. Kohajdova The use of PCA,

FA, CA for the evaluation of vegetable juices
processed by lactic acid fermentation, Czech J.
Food Sci., Vol. 20, No. 4: 135-143, 2002.
RMSEC
0.35
0.3
0.25
0.2
0.15
0.1
0.05
1
4
5
6
7
10
. 6.

4.
,

.

, ,

- ,

.
PCA,
PCR PLS
(
),

.

.
7. Li Y., H. Yang Honey discrimination using

visible and near-infrared spectroscopy, ISRN
Spectroscopy, Article ID 487040, doi:10.5402/2012
/487040, 2012.
8. Shaw P., M. G. Moshonas, B. Buslig, S.
Barros, W. Widmer Discriminant and principal
component analyses to classify commercial orange
juices based on relative amounts of volatile juice
constituents, Journal of the Science of Food and
Agriculture, Volume 79, Issue 13, pages 1949
1953, October 1999.
Department of Automatic, Information and
Control Systems
26 Maritza Blvd
4002 Plovdiv
BULGARIA
E-mail: t_titova@abv.bg
E-mail: kamelia_velcheva@abv.bg
E-mail: v_nachevbg@yahoo.com
E-mail:chavdam@yahoo.com
28.02.2013 .
- 261 -

TECHSYS 2013
BULGARIA

:
. ,
,
. ,
.
- .
: , , , ,
, , .
IDEAL GAS LAWS IN

THERMOECONOMICS AND FINANCIAL
BUBBLE FORMATION
TINKO EFTIMOV
Abstract: We discuss the validity and application of the thermodynamic model of the ideal gas
in economics. It is shown that while the well known aggregate demand curve is described by an
isothermal process, the isobaric and isochoric processes have important analogies in
economics. The relation to investment and entropy is also discussed and the important case of
financial bubbles is analyzed as well. The econophysical analysis carried out here provides a
better understanding of different economic processes.
Key words: Econophysics, thermoeconomics, ideal gas law, entropy, work, investment,
financial bubbles
1. Introduction:
The application of thermodynamic ideal gas
laws to economics is the basics of
thermoeconomics and a variety of applications of
the ideal gas equation have been presented [3].
Also the economic analogies of the first and second
laws of thermodynamics have been formulated

[3,2]. While the formation of financial bubbles and
price fluctuations of the stock markets have been
studied in econophysics [1,2], no attempt has been
made to analyze financial bubbles by means of the
ideal gas law. In this paper we provide examples of
- 262 -
different iso-process in economics characterized by

different price vs. volume of sales dependence. We
also provide empirical data showing that share
prices vs. volume of sales at stock markets are
approximately iso-price processes while during
bubble formation price-volume variations tend to
an iso-volume process.
as
2. Ideal gas law, work and internal energy

Ideal gas law formulations
We consider the ideal gas equation expressed
2
2
pV Nk B T N E k U
3
3
3
1
E k k B T Ei
2
N i
A pV
is investment in economics
The first law of econophysics thus reads
Qe dU e Ae
(1b)
3. Iso-processes in thermoeconomics
To the four iso-processes in thermodynamics
correspond the respective iso-processes in
thermoeconomics: at constant trading value
Te=const, at constant price P = const, at constant
volume of sales V = const, and with no exchange of
value (adiabatic).
constant N is the number of particles, E k is a

particles average kinetic energy and U is the
internal energy of the ideal gas, i.e. the sum of all
individual energies Ei and is proportional to
temperature.
Economic analogies of thermodynamic quantities
In thermoeconomics price P is the analogy of
pressure, volume of sales V corresponds to gas
volume, Te is economic temperature or also trading
value, N is the number of carriers of value and k is a
production of currency constant [1]
(3)
PV NkTe U e
The PV product which is referred to as the internal
value is the total financial resource which the
market of a given product can provide. If the
product is bread sold at a price P and at a sales
volume V, then PV = Ue is the total amount that the
customers can spend on bread over a given period
of time. The internal value which is the analogy of
internal energy, is proportional to the economic
temperature which, in turn, is proportional to the
purchase power of customers, the per capita GNP.
The economic temperature is also known as trading
value which reflects the fact that apart from the
purchase power the psychological perception of the
value of a product is of importance. Thus products
that enjoy a high esteem are ascribed a higher
economic temperature by the customers. We can
represent the trading value as consisting of two
parts: production content Tpr and psychological
perception Tps, i.e.
The analogy of work A in thermodynamics
(5b)
and is the resource allocated for the increase of the
number of products, production capacity, job
positions etc.
Ultimately, to the heat transferred in
thermodynamics Q corresponds entropic value
Qe, or income/expenses. The change in entropy is
then
(6)
dS Q / T and dS e Qe / T
.(1a)
where p is the pressure, V is the volume, T is the

thermodynamic temperature, kB is Boltzmans
Te T pr T ps
Ae PV
(4)
(5a)
(7)
TABLE 1. Iso-processes, work and entropy change.

Isoprocess
Isothermal
Iso-baric
Isochoric
Adiabatic
Work
Entropy
change
V
AT RT ln 2
V1
V
S T R ln 2
V1
A p pV2 V1
T
S p C p ln 2
T1
AV 0
T
SV CV ln 2
T1
V
AA iRT1 2
V1
S A 0
1
The iso-trading value process

This process takes place when the internal
value of a market or a system of economic
exchange remains constant. So is purchase power,
or per capita GNP and economic temperature is
constant meaning that the intensity of economic
exchange is constant. It follows from (3) that
NkTe U e (Te )
V
V
(8)
The P-V plot is the well known aggregate demand

(AD) curve which predicts a two-fold drop in the
price of a product if the volume of sales increases
twice at a temperature T. Under these conditions
the sales volume increases from V0 to V1 and an
investment Ae (shaded area) must be made to
increase the volume of production. Should the
- 263 -
volume of sales (and production) remain

unchanged, but the temperature rises (higher
purchase power, per capita GNP) to T, then the
price increases to P0 as shown in Fig. 1.
P
P0
saved money back in Bulgaria. Their economic

temperature (purchase power) actually increased
form T to T. Meanwhile the economic
temperature of Bulgaria has not changed and the
price levels have remained unchanged.
The iso-volume process
In this case the volume of sales does not
change and (3) becomes
T T
P0
P1
Fig. 1. Aggregate demand curve .
The aggregate supply (AS) curve can be modeled as
G.We (Te )
(V Max V )
(9)
where G is a coefficient, VMax is the maximum

possible production volume and is a power
parameter.
The iso-price process
In this case the prices remain constant, but the
trading value of a subgroup of consumers increases
from T to T which allows these customers to
increase the volume of the purchased product as
shown in Fig. 2. Since an increase in the volume
occurs, more items will have to be produced which
would need an investment to be made shown by the
shaded area (see Table 1). From (3) it follows that:
NkTe
p
(10)
so the volume of sales is proportional to the

economic temperature Te.
P
T T
P0
V0
V1
(11)
and the price P is proportional to the trading value

Te. If economic temperature increases, so will the
price. However, since the volume remains the same,
no investment is made (Table 1) but entropy
increases. The gains from price increase are because
of speculation.
V1
V0
NkTe
V
Fig. 2. Iso-price process.

Iso-price processes occur for example when
large groups of Bulgarian students take a summer
job in Great Britain or in the USA, earn
considerably more than in Bulgaria, and spend the
P
P0
T T
P0
V0
Fig. 3. Iso-volume process.

The purest example for this type of process is
bidding at an auction. The object of bidding is often
a unique piece with V = 1. During the bidding, the
perception of the item as having some special value
increases i.e. its trading value increases until the
highest possible bid. The process is shown in Fig. 3.
Similar processes can occur any time there is a
limitation on the volume of production and the
demand is much higher than the industry can
possibly provide. While a particular vendor can
achieve exrtra-profits, the heating of the auction
bidding diverts resources from other markets which
cool down. Monopoly tends to iso-volume
processes since extra-profits can be achieved
without investment to increase production volume.
Customers are thus deprived of value and their
necessities are undersatisfied.
4. Financial bubbles
We can now turn to the special case of
financial and economic bubbles. In particular, we
take examples from the prices of shares on the
stock exchange markets. When a company goes
public it offers a limited number of shares N at a
trading value of Tpr and naturally accumulates a
total financial resource of NkTpr. While the number
- 264 -
of shares remains constant N = const, the trading

value Te may vary largely not because of the
increased or decreased production content, but
because of the perception of increased or decreased
value, i.e. while the productive content may remain
constant Tpr= const, its perceived value may vary in
both positive and negative directions, i.e. Tps,
varies. Unlike produced goods which are consumed
after being purchased, the volume of sales on the
stock market does not represent volume of goods to
be produced or new shares issued. Instead, it is the
number of new exchanged possessions of a given
share. For example for N = 1000, on an active day
we can have a volume of V = 1500 which would
mean that certain shares were purchased and resold
several times. For a bubble to occur, two conditions
must be satisfied. The first is the availability of
surplus of money and the second is euphoric belief
that a certain stock share company is very
successful. A surplus of money can be achieved in
several ways the most common being the
indiscriminate printing of paper money non-backed
by goods or gold, the existence of cheap loans at a
very low or at no interest rate and the practice of
fractional reserve banking. The euphoric belief of
excessive expected future profits normally occurs in
times of technological breakthroughs that generate
new types of products outcompeting existing
technological achievements.
In the following we consider the P-V
dependences of the selling price versus the trading
volume V in the case of a normal situation and in
the case of a bubble. Fig. 4 represents the time
dependence of price P and volume V variations of
the Bulgarian company Neochim, listed on the
SOFIX stock market. It is clearly seen that while the
trading volume changes up to four orders of
magnitude, the share price P varies weakly. The PV diagram reveals a practically iso-price process in
a normal no-bubble situation. Since different points
lie on different isotherms (iso-trading value curves)
this type of stock market behavior means that the
trading value perception of the shares was varying
largely. Large volumes of hand-changing
correspond to high economic temperature, i.e. high
level of economic activity. These large variations in
Te actually indicate that the market has been
pulping the companys stocks value. Pulping
means that when potential investors try to express
their perception of an increased trading value of a
given stock through the stock exchange, they bid
but since other participants as a whole do not share
the same optimism, they do not continue bidding for
a higher price. The price thus oscillates around a
relatively stable value which reflects the productive
content of the products behind the shares of a given
company. So while the volume changes by more
than 1000 times (100 000%) the price on the

average changes by some 25%. I.e. we practically
have an iso-price process. Such a process takes
place when most of the bidders are either not
optimistic or they do not have easy money to run
the increased risk of buying at a higher price. Their
money in effect reflects a real productive content
which is why they are cautious.
The behavior starts changing once a bubble is
under formation. For a bubble to grow investors
with easy money are needed. When the perception
for a given stock sharply increases and most of the
participants bid with easy money that have no real
productive content (printed money), the higher
temperature leads to a higher price of the stock.
Fig. 5 represents the behavior of the SOFIX
index prior to, during and after the US housing
bubble from 2008 which is typical for other
bubbles.
What we find out is that prior to and after the
bubble the P-V diagrams correspond to iso-price
(isobaric) processes while price levels varies by less
than 50% the volumes varied by 3 orders of
magnitude!
Fig. 4. Neochim shares with no bubble: a) time

dependence of the price and volume; b) P-V plot.
- 265 -
Fig.5. SOFIX index prior to, during and after the

bubble: a) time dependence of price and volume; b)
P-V dependence of the SOFIX index.
During the bubble formation, the P-V diagram
appears more like a multilayer iso-price structure
which ultimately can be idealized by an iso-volume
process. While the amplitude of volume variations
shrinks from three to one order of magnitude (i.e.
by 10000 %), the price increases by almost 700%
rather than by about 50% as in the out-of-bubble
normal situations. The top of the diagram lies on an
isotherm with the highest possible economic
temperature. It appears that the market has been
pulping the price by smaller variations in the
volume and at increasing iso-price levels. The
higher the price, the smaller the volume changes.
Ultimately, it is the last fool that remains and pays
the highest price and the volume shrinks to a
minimum!. The decreased volume changes indicate
the growing pessimism of the bidders. I.e., during
the formation of the bubble the market has been
tending more and more to a fuzzy iso-volume

process.
At price maximum , the temperature T3
which reflects the non-realistic perception of high
value is much higher than T2, that characterized the
highest possible optimistic perception of the
realistic players. Realistic players bid with money
that have productive contents. Productive contents
means that a real production process involving
investment has taken place and the money is backed
by a real product priced realistically on the market.
When overly optimistic players bid with easy
money, which are not backed by a real product,
they actually reflect the fact that there has been no
real investment and no real products behind the
financial instruments in use whatever their nature.
Since no investment is made, no work has been
performed. In thermodynamics the process during
which no work is performed is the isochoric
process. Analogously, when no investment is
performed and no real production process backs the
financial instruments in use, they will tend more
and more to produce iso-volume processes which is
precisely what we observe in a bubble. Therefore,
during bubble formation, no new value is created.
It is just speculation. In reality bubbles cause
irrational investment, losses and massive debts for
the majority of the losers.
As the bubble grows the productive content
behind the stock price decreases. More and more
those who profit do so not by offering a real product
on the market, but by the acquisition of financial
resources coming from the newcomers. The system
operates more and more like a classical Ponzi
scheme and becomes a variation of a financial
pyramid.
The problem is exacerbated by globalization.
Since most of the easy credit and inflated money
supply comes for central banks as the Federal
Reserve of the USA, and since the US dollar serves
as a world currency, the collapse of one bubble
leads to the formation of another bubble somewhere
else in the world. The winners from the High Tech
bubble that burst in 2000 started blowing the next
bubble the housing bubble which burst in 2008.
Since most of the housing bubble was inflated
through the artificially low interest rates, the US
bonds and the toxic derivatives purchased by
investors worldwide, the collapse of the housing
bubble lead to a world crisis that hit the other stock
markest as the Sofix, for example. If the housing
bublled did not collapse in 2008, the Sofix index
would have been blown to even higher values and
would have imploded with even worse
consequences some time later.
What is worse is that since no official
devaluation of the US dollar is declared, new
- 266 -
bubbles will be formed in the years to come and

these may cause even greater financial and
economic losses and crisis. The dynamics financial
and economic bubbles, their growth and crash
simply proves that in economics, as in physics, the
energy conservation law (7) is observed. The
overheating of one market is at the expense of the
cooling of another market. It is just a matter of an
artificially created non-equilibrium state and no
financial or economic perpertual motion machine
can be created. The history of practical physical
perpetual motion machines is abundant with
examples of deceitful practices to demonstrate a
working perpetual motion machine. Much in the
same way the history of financial bubbles is
abundant with deceitful bookkeeping practices
maintaining the illusion of prosperous companies
headed to an illimited growth and profits.
and implosion the process tends to an isovolume one.

Ref erences
1. Econophysics Approaches to Large-Scale
Business Data and Financial Crisis, Eds. M.
Takayasu, Ts. Watanabe and H. Takayasu,
Springer, Tokyo 2010.
2. Econophysics and Sociophysics, Ed. By B. K.
Chakrabarti,
A. Chakraborti, and A.
Chatterjee, WILEY-VCH Verlag GmbH & Co.
KgaA, 2006.
3. John Bryant Thermoeconomics - A
thermodynamic approach to economics, VOCAT
International Ltd, 2011. ISBN 978-0-9562975-3-2
http://www.vocat.co.uk/TECH12012.pdf
5. Conclusions
The analysis performed in this paper allows us to
formulate the following conclusions:
1. We have shown that the ideal gas model
can describe several price vs. volume of sales
scenarios rather than the classic aggregate
demand curve.
2. Empiric data show that in normal stock
market trades the P-V diagram is practically
iso-price process. During bubble formation

Technologies,PU P. Hilendarski,
Plovdiv 4000
BULGARIA
12.01.2013 .

TECHSYS 2013
BULGARIA

, , ,
: -
.
45 -
-
.
. Matlab/Smulink e
Spectrum Digital eZdspTMF28335
TMS320F28335 ,
.

:
, - ,
HARDWARE IN THE LOOP SIMULATION OF

ROBUST HELICOPTER CONTROL SYSTEM
TSONYO SLAVOV, LUBEN MOLLOV, JORDAN KRALEV,PETKO PETKOV
Abstract: The design of a miniature helicopter integral attitude controller of 45th order is
presented in this paper. Comparing to the existing controllers the present one ensures robust
stability and performance for greater frequency independent input multiplicative uncertainties. In
addition, three simple PD controllers are designed which ensure tracking of the desired trajectory in
3D space. software in the MATLAB/Simulink environment for code generation, embedded in
Spectrum Digital eZdspTMF28335 development kit with built in DSP Texas Instruments
TMS320F28335, is developed. The results obtained by hardware-in- the- loop- simulation are
presented.
Key words: Hardware-in- the-loop simulation, controller, Miniature helicopter control
1.

.
- H
[2,9]. -
H

[4,6].
,
()
- 268 H

.

-

.
H ,
, . ,

.
- .
[10] [8]
,
[5],

10 %. [8] 28 .

(Hardware-in-the-loop-simulation)

.

[1,7].

Simulink Coder
Embedded Coder.

. 45

-
.

.
MATLAB/Smulink e
Spectrum
Digital
eZdspTMF28335
() TMS320F28335 ,
.

(),

.
,
Simulink
,

.

,

.
2.

13 ,

[3]. :
,

,
;
,
;

,
.

.
us
us
xh = Axh + B c , yh = Cxh + D c ,

dist
dist
xh = [u , v, w, p, q, r , , , , mr , i , a1 , b1 ] ,
T
us = [ col , lon , lat , tr ]T ,
(1)
dist = [uwind , vwind , wwind ]T ,

yh = [u , v, w , p, q, r , , , ]T ,
u , v, w, [m / s ] ,
p, q, r , [rad / s ] - ,
, , , [rad ] - , mr ,[rad / s ] , i ,[rad ]

c ,[rad / s], a1 , b1 ,[rad ] -
, col ,[rad ] -
, lon , lat ,[rad ]
,
tr ,[rad ] -
, uwind , vwind , wwind ,[ m / s ] -
, A, B, C , D - .
- 269 -
(1)

us ( s )
yh ( s ) = Gheli c ( s ) ,
(2)
dist ( s )

Gheli A, B, C , D .

X-Cell 60 SE

linmod MATLAB.

u trim = 0.001, v trim = 0.001, wtrim = 0, trim
mr = 167.
u trim vtrim 0 .

p trim = 0, q trim = 0, r trim = 0, trim = 0.0873,
trim = 3.3 106 , trim = 0, itrim = 29.17,

trim
= 0.0967,
a1trim = 0, b1trim = 8.48 103 , col
trim
= 0, lattrim = 0.020, trtrim = 0.2488.
lon

Z .

.
(2)
.

(.1), i < 0.25, i = 1,..., 4.
. 1.

25 %.

= diag (1 , 2 , 3 , 4 ) .

I + 044
(3)
G ( s ) = Gheli ( s ) 4
.
I 4
044

us
y = G ( s) c .
(4)
dist
G= [Gu G Gd ] ,
(4)
y = Gu us + G c + Gd dist ,
(5)
Gu , G Gd
,
. ,
c = 167[rad / s ] .
3.

-
,

.

.

.2.

, ,
ref , ref , ref ,
. 2.

u , v, w
p, q, r . ,

.
uc = [uccol , uclon , uclat , uctr ]T
- 270 -
Ws = diag ( wscol ( s ), wslon ( s ), wslat ( s ), wstr ( s )) ,

(6)
wscol ( s ) = wslon ( s ) = wslat ( s ) =
n2
s / Tz + 1
,
s / Tp + 1 s 2 + 2n s + n2
W p So Gd
W p ( So GuWs K r M )
< 1 , (10)
Wu Si K y Gd
Wu Si K r
, . 3.
tr2
w (s) = 2
,
s + 2 tr tr s + tr2
tr
s
Tz = 104, Tp = 33, n = 36, tr = 14 , = 0.5, tr = 0.6.

-

.

r
(7)
u c = [ K r K y ] = K r r + K y yc ,
yc
T
T
r = ref , ref , ref , yc = u , v, w ,,, ,
K r
K y -
.
M = diag ( wm1 , wm 2 , wm 3 ) ,
1
wm1 =
,
2 2
0.4 s + 2 0.4 0.7 s + 1
1
,
wm 2 =
2 2
0.6 s + 2 0.6 0.7 s + 1
1
,
wm 3 =
2 2
0.5 s + 2 0.5 0.7 s + 1
,
.
,

max
max
= 0.183rad , lon
= 0.096rad ,
col
(8)
latmax = 0.096rad , trmax = 0.38rad
.2
W p So Gd r
ey W p ( So GuWs K r M )
, (9)
e =
Wu Si K r
Wu Si K y Gd d
u
Si = ( I + K y Gu ) 1 , So = ( I + GuWs K y ) 1

.
- ,
(7),
:
:
,
.
:
G

. 3. -

100 Hz ,

. N d ( z )

Pd ( z ) = FU ( N d , )

.

0
: L44 , F L77 , (11)

P :=
d
0 F
ey eu
rc , d , c .
K d ,
[0, / Ts ],
Ts = 2 / f s

[ FL ( N d , K d )( j )] < 1 ,
(12)
Pd
FL ( N d , K d )
.

, ..
FU FL ( N d , K d ), P < 1
d
Pd ,
Pd
< 1 . MATLAB
dksyn
103 s + 1
0.02 s + 1
W p ( s ) = 102 2
I 3 ,Wu ( s ) = 103 4
I3 .
10 s + 1
10 s + 1
0.958,
45 .
- 271 -
. ,

,

.

. ,
Vx ,Vy ,Vz ,[m / s ]
ta

Vx = w(cos cos ) ,
Vy = w(cos ) ,
(13)
Vz = w cos cos ,
(14)
w = Z / m + g cos cos ,
Z ,[ N ] , Z
, m,[kg ] -
g ,[m / s 2 ] - .
Z ,

Tmr ,[ N ]
Z ~ Tmr
(15)
(15) (14) ,
Z
col
X Y

. (13)

U z = [ K p1 ( zref z ) K d 1Vz ] / (cos cos ),
U x = [ K p 2 ( xref x) K d 2Vx ] / (cos cos ), (16)
U y = [ K p 3 ( yref y ) K d 3Vy ] / cos ,
xref , yref , zref
, K p1 , K p 2 , K p 3 -
Kd1 , Kd 2 , Kd 3
(16)

,
X, Y
Z . U z ,U x U y
. (16)

K p1 = 0.4, K p 2 = 0.1, K p 3 = 0.3,
K d 1 = 0.3, K d 2 = 0.5, K d 3 = 0.6.
,
.
4.

.
.4.
(Host PC),

Spectrum Digital
eZdspTMF28335 TMS320F28335.
,

(32-bit).
(arget DSC)
100 Hz.
Host PC
MATLAB R2010b, Simulink V.7.6,
Simulink Coder V.7.6, Embedded Coder V.5.6,
Microsoft Visual C++ V.8.0 Code Composer
Studio V.3.3.
. 4.
Host PC arget DSC

RS232
115200 bps,
20-30 ms,

.

Protocol ,
. . handshaking Host
PC arget DSC.
,

,
,
,
.
arget DSC Host PC
(10 bytes)
, Host PC
(26 bytes).

.
Simulink
(.5).
- 272 -
(Embedded function),
.
,
,

16 (int16).

.

.

( 1).
, 12 ,
(8)
6700
.

: 1
Simulink ,

; 2 . 5;
3 ,
Host PC
,
Target DSC.
, 1
3, 2
.

. 6
deg
deg/ s
m/s
m / s2
300
50
90
50
16
16
16
16
. 100
655
364
655
.6.
.

3
v wind (t ) = v0 + Ak sin(k t ) + vg (t ) , (17)

k =1
v0 , Ak k - , k -
k - , vg (t ) - ,

vg (t ) =
vg max
2v
g max
4(sin (2 / Tg ) t 1)
,
(18)
1+ e
g -
( g = 2 / Tg ).
2.
. 5. Simulink
2

vg max Tg
v0
k
Ak
k
[m / s]
[ Hz ] [ m / s ] [m / s ] [ s ]
1
0.5
0.1
0.2
10
1
2
0.7
0.25
3
1
0.3
1
0.6
-0.05
0.8
15
Y
0.5
2
1
-0.1
3
1.5
-0.3
1
0.1
-0.2
0.5
20
Z
2
2
1
-0.25
3
1.5
-0.1
- 273 -
25
20
y [m]
20.2
15
20
10
19.8
20
22
24
26
28
30

Simulink
-5
-10
10
20
x [m]
30
40
50
60

X ,
.
a
Y
,
. 12 13

. ,
.

.
. 7 X Y
0.05
-17.5
0.04
z Simulink
z
-18
Simulink
trim

0.03
-18.5
0.02
[rad]
-19
z [m]
-19.5
0.01
-20
-20.5
-0.01
-21
-0.02
-21.5
-0.03
0
20
40
60
80
100
120
time [s]
140
160
180
200
. 8 Z

20 m. . 7
XY
, .8
Z . ,

.
Y
.
40
60
80
100
120
time [s]
140
160
180
200
x 10
3
2
1
0
-1
Simulink
trim

-2
-3
0.12
20
-3
0.13
Simulink
trim

.10
[rad]
-22
-22.5
20
40
60
80
100
120
time [s]
140
160
180
200
. 11
0.11
0.15
col Simulink
col [rad]
[rad]
0.1
0.09
0.08
col trim
0.1
col
0.05
50
0.07
100
time [s]
150
200
-3
0.05
20
40
60
80
100
120
time [s]
140
160
180
200
.9
9-11
.

Simulink . -
lon [rad]
0.06
x 10
lon Simulink
lon trim
-2
lon
-4
50
100
time [s]
150
200
. 12

- 274 -
-3
lat [rad]
x 10
lat Simulink
lat trim
lat
-2
50
100
time [s]
150
200
tr [rad]
-0.22
-0.24
tr Simulink
-0.26
tr trim
-0.28
tr
-0.3
50
100
time [s]
150
200
. 13

5.

.

.
1. Cai, G., Chen, B.M., Dong, X., Lee, T.H.

Design and implementation of hardware-in-the-loop
simulation system for small-scale UAV helicopters.
Mechatronics, 19(7), p.1057-1066, 2009.
2. Castillo-Effen, M., Castillo, C., Moreno, W.,
Valavanis, K.P.Control Fundamentals of Small
Miniature Helicopters - A Survey. In: Valavanis,
K.P. (Ed.): Advances in Unmanned Aerial Vehicles,
p. 73-118. Springer, 2007.
3.Gavrilets, V., Mettler, B., Feron. E. Dynamic
model for a miniature aerobatic helicopter.
Technical report, MIT-LIDS Report LIDS-P-2580,
2003.
4. La Civita, M., Papageorgiou, G., Messner,
W.C., Kanade, T. Design and flight testing of a
high-bandwidth H loop shaping controller for a
robotic helicopter. Journal of Guidance, Control,
and Dynamics, 29(2), p. 485-494 ,2006.
5. Mettler, B. Identification Modeling and
Characteristics of Miniature Rotorcraft. Kluwer
Academic Publishers, Boston, 2003.
6. Poslethwaite, I., Prempain, E., Turkoglu, E.,
Turner, M.C., Ellis, K., Gubbels, A.W. Design
and flight testing of various H controllers for the
Bell 205 helicopter. Control Engineering Practice,
13(3), p.383-398, 2005.
7. Shen, N., Su, Z., Wang, X., Li, Y. Robust

controller
design
and
hardware-in-the-loop
simulation for a helicopter. 4th IEEE Conference on
Industrial Electronics and Applications (ICIEA
2009), Xi'an, Shaanxi (China), p.3187-3191, 2009.
8. Shim, H. Hierarchical flight control system
synthesis for rotorcraft-based unmanned aerial
vehicles, p. 126-137. PhD Dissertation, University
of California, Berkeley, 2000.
9. Wang, X., Zhao, X. A practical survey on the
flight control system on small-scale unmanned
helicopter. In: Proceedings of the 7th World
Congress
on
Intelligent
Control
and
Automation(WCICA 2008), Chongqing, China,
p.364-369 ,2008.
10. Yuan, W., Katupitiya, J. Design of a synthesis controller to stabilize an unmanned
helicopter. Proceedings of the 28th Congress of the
International Council of the Aeronautical Sciences,
Brisbane, Australia, Paper number ICAS 201211.5.2, 2012.
Department of Systems and Control
8 Kliment Ohridski Str.
1000 Sofia
BULGARIA
E-mail: ts_slavov@tu-sofia.bg
E-mail: l_mollov@abv.bg
E-mail: jkralev@yahoo.com
E-mail: php@tu-sofia.bg
25.02.2013 .

TECHSYS 2013
BULGARIA

FLASH
, ,
: ,
,
.
, -01/10.
.
,
.

.
: ,
EDUCATIONAL MULTIMEDIA APPLICATION

WITH SOFTWARE SIMULATIONS
VALENTINA VOINOHOVSKA, SVETLOZAR TSANKOV, RUMEN RUSEV
Abstract: This paper is related with a research developed in collaboration by three higher
education institutions University of Ruse, New Bulgarian University and South-West
University "Neofit Rilski". The research is supported by National fund: Bulgarian science fund,
project -01/10. The major objective of this project is to research innovative methods
for assessing competence in e-learning environments. The specific article presents a multimedia
learning application with software simulations. It is developed for the course Multimedia
Systems and Technologies and it is used like additional didactic tool.
Key words: multimedia, education
1.

,

.

.
.
,

[3, 5, 6].

,

.
- 276

.
,
.

: 1)

2)

.

.
[1, 2, 4] ,

,

.

.

. -
. ,

.
, ,

.
: 1)
2)

.

,
.
2.

Flash
.

.
Lynda.com - Flash Professional CS5
Essential Training
Flash Professional CS5 Essential Training,
, Flash
Professional CS5

.

- ,
ActionScript.
.
,

.
,
:
Flash Professional CS4 Essential
Training;
Flash Professional CS3 Essential
Training;
KelbyTraining Flash CS4 Basics;
LearnFlash.com:IntroductionToFlashCS
4;
Learning Macromedia Flash MX 2004;
Learning Flash CS3 Dynamic Learning;
.
, Flash
, Flash

Macromedia
FlashMX2004.

:
;
;

;

;
;

;

.

AdobeFlash.
AdobeFlashCS3 Adobe
Captivate. -

.

.
- 277 , , ,

.
,

,
-
.
()
:
1) ,

, ..
.
2)

, (
).
3)
:
.
4) ,
.
5) (
,
-
).
6)
.

,

.
6)
.
7)
.
8)

, ,
.

.
:
1) .
2) .
3) .

(. 1).
. 1.

, .

,

(. 2).
. 2. ,

.

Adobe
Flash

.
,

(. 3).
,

(. 4).

:

.
,
, .
- 278 -
. 3.
. 4.

,

. :
1)
.
2) ,

.
3) ,
,
.
4) ,
.
5)
.
6) ,

.

.
7) ,

.
8) ,

.

Adobe Flash CS3
,
.
Adobe Captivate.
Captivate
Flash- HTML-.
Adobe Captivate,

Flash-.
Adobe Captivate
:

;
, ,
,
;
;
SWF ,
, PDF .

:
1)
,
.
2)
.
3)
,
Captivate.
4)
( ).
5)

Adobe Flash.
3.
,
.
,

,
.

- 279
.

: 1)

,
,
,

? 2)

?

,
.
, ,
,
,

. ,
.

,
.
,

.
,
- .
,

.
,
,

Adobe Flash.
,

.

.
,
.

.

:

, ;

.

, , ,
,
.
,

.

. ,
,
.

.

,
,

, .

,
.
.

.

,
.
.
-
, -
.

. -

. ,

(
) , .
, -
,
- 280 ,
.
,

,
,

.
based training. London. Kogan Page, 1992.
4.
,

.

. ,
,
.

, .

,

.

:
1)
;
2)
;
3)

;
4)
6. Underwood, J. Linguistics, computers, and the

language teacher: A communicative approach.
Rowley, MA: Newbury House, 1984.
1. Alessi, ., S. rollip. Multimedia for learning.

Boston: allyn and bacon, 2001.
2. Becker, K. Teaching with games: the
minesweeper and asteroids experience. The Journal
of Computing in Small Colleges, Vol. 17, 2001, 23
33.
3. Dean, Ch., Q. Whitlock. handbook of computer
4. Grice, R., L. Strianese. Learning and building

strategies with computer games. In Proceedings of
2000 Joint IEEE International and 18th Annual
Conference
on
Computer
Documentation,
Cambridge, MA, USA, 2000.
5. McConnell,
D.
Implementing
computer
supported co-operative learning. London. Kogan
Page, 1994.
Department of IIT
University of RousseRousse
8 Studentska St.
7017 Rousse
BULGARIA
E-mail: voinohovska@ami.uni-ruse.bg
E-mail: stzancov@ami.uni-ruse.bg
E-mail: rir@ami.uni-ruse.bg
28.02.2013 .
- 281 -

TECHSYS 2013
BULGARIA
,
:
(Hf)
()
. ,
, .
.

.
: (), ,
CRITERIA FOR REFINING OF HAFNIUM SCRAP

USING ELECTRON BEAM MELTING
VANIA VASSILEVA, KATIA VUTOVA
Abstract: Values of the coefficients for relative volatility of metal impurities in hafnium (Hf)
and values of the free energy for possible chemical reactions at electron beam melting and
refining of Hf scrap are estimated. Factors controlled refining processes are investigated and
criteria, measured the possibility for refining go are selected. Obtained experimental data are
presented and discussed. Technological scheme for EBMR of hafnium scrap is proposed for
concrete electron beam equipment.
Key words: electron beam melting and refining (EBMR), impurities, hafnium, scrap
1.
(
10-3 )
,

.
,

.

,
,
.

.
()

,
,

[1-3,6].
- 282 -

.

-
:
,
[3].

,
,
.
,
,

,
:
i = pR
pi
Mi
MR
(1)
R pi a
, MR Mi
,
. [1]

.

- ,
.. F. F

:
F = F0 RT ln pn ,
,
.

. ,

(Hf)

,

()
.
2.

-60

.

60 kW,

.
(2)
F0 F

; R
8.31 kJ/kmol,
[]; p ,
( 10-3
), n ,
.
F<0.
-

,
.
,
. 1.
Hf (-60)
(. 1)

,

60 mm 10 mm.
() e 12 kW ;
15 kW 17 kW , a 2
min 4 min.

(. 1)
- 283 -
QP-31

. e

ICP MS .
3.
. 2
i

2500 3000 . [5]
1,00E+08
2500
1,00E+06
2700
1,00E+04
2900
1,00E+02
3000
1,00E+00
1,00E-02
, i
1, ,

.

-

.
1,00E+08
1,00E+06
1,00E+04
1,00E+02
1,00E+00
1,00E-02
1,00E-04
1,00E-06
MgMnCr Pb AgAs Al Fe Ni Ti CoMo V Hf Zr Ta W
1,00E-04
1,00E-06
. 2.

2500 3000
, ,

-
,
. ,
,

/

-
(
) (
).
2500
2700
i
Hf ( (1))
. 3. i
-
10-9 105.
,
(. 3), i << 1

.
,
, i >> 1

.
MgMn Cr Pb Ag As Al Fe Ni Ti CoMo V Hf Zr Ta W
1,00E-08
1,00E-10
2500
2700
. 3.
i Hf
2500 2700
ICP MS

, 1,

,

(. 2).
,

.
(,
, Hf . 2, 3)

.
,

.
-
,

.

10-3 . [4]
2.
- 284 -
.

Ci [ppm]
P
T
GHf
O
Zr
Ta
W
Co
Pb
Ti
kW
K
min
g
1
12 2500
2
2
12 2500
4
3
15 2700
2
4
15 2700
4
5
17 2900
2
6
17 2900
4

503,4 544
418,5 34,7
843,9 26,8
950,3 12,0
654,3 168
642,2 33
524
5226 6,8
1620 2,4
1865 2,0
2856 3,9
1972 3,5
2535 6,8
1200 3,29
2
F
,
Hf 2500 2700
F2500K,
F2700K,
kJ/kg
kJ/kg
+1937
+2150
2Pb+ O2 2PbO
+4805
+5464
2Co+ O2 2CoO
+2603
+3429
2Mn+ O2 2MnO
Hf+ O2 HfO2
-894
-540
-2964
-2248
4Ta+ 5O2 2Ta2O5
-2226
-1534
Zr+ O2 ZrO2
-3342
-1958
4Cr+ 3O2 2Cr2O3
-4762
-3927
4Nb+ 5O2 2Nb2O5
-2885
-1886
Ti+ O2 TiO2
. 4

. ,
, - 97.7%
= 4 min
.
,
(F >0
).

.
-
( -
)
-
. . 5

Hf .
11,3
6,1
3,7
17,2
7,3
48,3
2,09
0,27
4,00
0,02
0,18
0,13
1,4
2,46
0,33
0,08
0,55
0,16
0,07
0,09
3,29
Cr
Nb
Mn
41,4 5,21 8,89 0,1

56,4 6,70 4,4 0,6
3,03 0,00 3,3 0,33
4,33 31,20 12,8 1,6
23,5 0,90 0,39 0,18
33,3 7,90 6,2 3,96
1,46 16,00 2,46 2,01
. 4. 2

Ta, Zr, Cr, Nb Ti F < 0

,
.

.

.
,
,
.

. . 5
- ,
Hf.
- ,
,
.
,
= 2 min

(.. ).

- 285 -
,
.

Ti, ppm
12kW
15kW
17kW
120
80
40
0
12kW
15kW
17kW
, min
, min
C Ti, ppm
C Pb, ppm
Pb, ppm
. 5. i

4.

Hf
,
,
.

,

. -

(=2min).

(>2min)

,
.
, Hf
-
,
,

0.1% .

(97.7%)
,
, ,
Hf
.
,
, :
.
-
99.6 %.,

99.7
%.
.
-60,
99.8%,
92.4%.

02-127/2009 (BIn-5/2009),

-
,

.
1. .. .
- ,
, : ,
, (19), 6, . 11-17,
2011.
2. . .

, -, , 2012.
3. . ,
. .,
, 2009.
- 286 -
4.
..,
..
65 ,
, , ,
., 1965.
5. Samsonov G, Chemo-physical Properties of
Elements, Kiev, Naukova doumka Publ. House,
145, 1965.
6. Sharma I.G., N. Krishnamurthy, and A.K.
Suri, Electron beam melting of reactive and
refractory metals and alloys, Proceedings of IndoBulgarian Workshop on Electron Beam
Technologies and Applications, ed. A.K.Das,
Mumbai, India, 42-50, 2004.
Laboratory Physical problems of Electron

Beam Technologies
Institute of Electronics, BAS
72 Tzarigradsko shosse blvd.
1784 Sofia
BULGARIA
E-mail: vvvania@abv.bg
E-mail: katia@van-computers.com
vutova@ie.bas.bg
01.03.2013 .
- 287 -

TECHSYS 2013
BULGARIA

,
:
()
144
.

. , ,
,
95%
.
,
40 Gb/s.
: ,
FIELD MEASUREMENT OF POLARIZATION

MODE DISPERSION OF COMMUNICATION
FIBERS USING THE FIXED ANALYZER
METHOD WITH POLARIZATION AVERAGING
VANYA PLACHKOVA, TINKO EFTIMOV
Abstract: By averaging over a sequence of random input polarization states, we have
determined the polarization mode dispersion (PMD) and obtained the DGD probability
distribution in an optical cable line made of 144 stacked fiber pieces. The fixed analyzer method
in a transmission mode has been used. Due to the insufficient dynamic range of the optical
spectrum analyzer the measurements could not be performed in reflection mode because of the
accumulated losses. The field tests performed at a fiber cable installation company show that the
empiric distribution of experimental results can be fitted by a Maxwell distribution with 95%
confidence level. Reference to the international communication standards shows that the fibers
under test are convenient for building 40 Gb/s high bit-rate systems.
Key words: polarization mode dispersion , communication fibers ,
- 288 -
1. Introduction:
High-speed high-capacity communications
are the basis of contemporary global
communication networks using single mode optical
fibers [1]. In these fibers pulse broadening is
caused by chromatic and polarization mode
dispersion [2, 3]. When the former can be
compensated and is nullified the only limitation to
high bit-rate communication is the polarization
mode dispersion. While modern fibers are
manufactured with more stringent control on fiber
quality [2], older brands already installed and in
use feature more imperfections and exhibit higher
PMD. For the present day communications systems
the PMD is the key limiting factor for achieving
10Gb/s and 40 Gb/s
The accurate determination of the PMD is
of paramount importance for the characterization of
the transmission properties of high-speed (above 2.5
Gb/s) long-haul (above 100 km) optical
communication lines. A number of methods have
been developed to measure PMD and the groupdelay statistics and several types of PMD meters are
commercially available [4-6].
In the present paper we apply the well
known fixed analyzer method in combination with
input polarization changes to generate a large
number of independent measurements which enable
statistical treatment of the effect of random
perturbations on the PMD..
2. Pulse broadening in single mode optical
fibers
In optical fibers the differential group delay
between the fastest and the slowest components
of a light pulse is given by
2
2
2 chr
PMD
of oscillation of the electric vector E and the

type of trajectory followed by its end linear,
circular or, in the general case elliptical.

45
-45
4. Uniform and random birefringence in

optical fibers
An ideal conventional optical fiber preserves the
state of polarization when in a single-mode regime
of operation because the propagation constant of the
X- and Y-polarized fundamental mode are the same
and are thus degenerate. Minute external
perturbations imposed upon the fiber introduce
birefringence and break the degeneracy due to
which the X- and Y-polarizations have different
propagation constants x and y accumulate a phase
difference . In result the polarization changes from
linear to elliptic and back to original along the fiber.

(1)
where chr is the broadening due to the chromatic
dispersion, while PMD is the broadening caused

by the different delays of the X and Y polarizations
of the fundamental mode. The corresponding
dispersion is physically the intermodal dispersion
between two orthogonal polarization modes known
as polarization mode dispersion (PMD). Since
chromatic dispersion can be nullified or
compensated, i.e. chr 0 , the only limitation to
bit-rate is the PMD.
3. State of polarization in optical fibers

Polarization
is
the
property
of
electromagnetic waves related to the direction
Fig.1. Polarization components of the fundamental

mode in a linearly birefringent fiber with core
ellipticity.
Artificial birefringence thus causes a difference
in the time delays and a differential group delay
(DGD) between X- and Y-polarized fundamental
mode arises which we dente as g. Differential
group delay arises because of polarization mode
dispersion (PMD) which is the DGD between the
two polarizations per unit length of the fiber. To
define PMD we assume that the propagation
constant difference
= x - y is frequency
(wavelength) dependent and expand it into a Taylor
series as:
- 289 -
0 g 0 .....
(4)
where 0 is the propagation constant

difference at a center frequency 0. When =
const along the fiber we have uniform birefringence

and if varies along the fiber the birefringence is
random.
5. Birefringence in optical fibers and PMD
State of polarization in optical fibers
L yields the total phase difference accumulated

between the two polarization modes.
L 0 L g 0 L
In (4)
g g L
(4)
(5)
is the DGD, while g (in ps/km) is the polarization

mode dispersion (PMD).
The PMD thus is the DGD per unit length. If we
The state of polarization in optical fibres is

determined by the input polarization state and by
the existence of birefringence in the fibre. Uniform
birefringence can be either linear, circular or,
generally, elliptical. Core/cladding ellipticity,
lateral stress and bending cause linear birefringence
and the fibre preserves linear polarization launched
along the X or Y axis. Twists, helical winding and
Faraday effect cause circular birefringence so the
fibre preserves circular polarizations. Combining
linear and circular birefringence produces elliptical
birefringence. Birefingence can be unform and
random. Uniform birefringence is caused by a
constant and uniformly distributed perturbations
imposed upon the fibre. When the strength and the
position of the perturbations vary randomly so is the
birefringence.
measure the average group delay g , then the
Uniform birefringence and PMD
is the PMD coefficient measured in ps/km,

and Lc is a critical fiber length over which
polarization mode mixing is complete.
So the PMD coefficient which describes
polarization mode dispersion in randomly perturbed
fibres is calculated as
We consider uniform linear birefringence along a

fibre caused, for example, by an elliptical core. The
propagation constants for the X- and Y-polarized
fundamental modes LP01x and LP01y . x and y are
different and the propagation constant difference
at a central frequency 0
n n n
(2)
0 x 0 y 0 0 s 0 f 0
c
c
c
is nonzero. In (1) n ns n f n x n y is the
fiber modal birefringence where n s and n f are the
effective refractive indices along the slow and the
fast axes, denoted as X and Y, and the refractive
indices become n x and n y which are different.
This causes the different polarizations to delay
differently and hence a differential group delay
(DGD) g.
We now expand (1) in series and restrict to the
first order term
0 g 0
(3)
where 0 is the propagation constant difference at
0. Multiplying (2) on both sides by the fiber length
average PMD will be:
g
L
ps / km
(6)
Random birefringence and PMD coefficient

With random birefringence fast and slow
axes interchange so the slow and the fast modes
interchange optical power. This causes slow modes
to move faster and fast modes more slowly on the
average, so DGD does not grow linearly with fiber
length bus is proportional to
g g Lc .L . g ,coeff L
(7)
In (7)
g ,coeff g Lc
g ,coeff
g
L
ps / km
(8)
(9)
So (6) is the PMD for distances L << Lc, while

(8) holds for long fibers for which L>Lc.
6. The fixed analyzer method
Basic scheme
The fixed analyzer method is one of the
standardized methods for the measurement of PMD
and the PMD coefficient. The method is based on
the classical polarimetric scheme in which the
birefringent medium is placed between crossed
polarizers with X-axis oriented at 45 with respect
to the polarizer transmission axis.
It can easily be shown that in the case of a
non-polarized source, the response of the fixed
analyzer scheme is
S0
1
1 cos 1 1 cos(L)
4
4
(10)
- 290 -
Or also, after reworking (4) the spectral dependence

of (9) can be written as
1 cos 0 2c g 2 z (11)
where is the mean wavelength and 0 0 L

For a 2 phase shift of the above cos response the
S0
spectral period is found as:
2
g L.c
(12)
From (11) we can calculate the DGD and the PMD
g g z
2
c
(13)
Polarization averaging
For randomly birefringent long-haul fiber lines
(L>Lc) the response (9) does not have a constant
period . Instead we observe a response with
randomly varying periodicity. The particular pattern
depends on the input polarization. By varying the
polarization prior to the test fiber we can achieve a
variety of random period patterns and accumulate
statistics on the distribution of the DGD. To vary
input polarization we use the scheme in Fig. 1.
P (45)
A (-45)
Polarization
controller
Fiber under
test
Fig. 1 Schematic representation of the fixed

analyzer method with input polarization controller.
In Fig. 1 the polarization controller is represented as
a phase plate with a phase difference , rotated
trough an angle Using Mueller matrices the
response in this case is
1
1 cos sin 2 2 cos2 2 cos
4
(14)
sin cos 2 sin
S0
Fig. 2 Schematic representation of a fiber with

randomly oriented birefringence.
7. Significance of PMD for optical fiber
communication systems
Polarization mode dispersion is of fundamental
importance for high speed communication links o
transmitting at rates Rb > 2.5 Gb/s. This PMD is
particularly high for older brands of single mode
fibers whose PMD value was not a parameter to
minimize in earlier communication systems. In
present-day high bit-rate systems PMD is the only
limiting dispersion once the chromatic dispersion is
compensated.
8. Field measurement of PMD in
transmission mode
Field measurements of the PMD were realized
with the DC Corporation using a multi-fiber cable.
The individual fibers were spliced in series and a
total length of 57,752 km, measured with a OTDR
(Optical Time Domain Reflectometer). The
experimental set up is shown in Fig. 3. By varying
the orientation of the polarization controllers coils
placed after the polarizer we create different input
polarizations and generate responses with
randomly varying spectral periods .
Broadband
source
JW3107 ASE
FO polarizer
Spool with
144 optical
fibers
Optical
Spectrum
Analyzer
AQ6331
FO analyzer
If we vary and the amplitude and the phase of

the spectral response change. If the fiber is with
randomly varying birefringence as shown in Fig.2,
then by varying the state of polarization prior to the
test fiber, the state of polarization may in certain
sections of the fiber may be its eigenpolarization so
this section will not contribute to the periodicity in
the spectral response. In this way different statistical
realizations can be observed and a sufficient
number of measurements can be made.
Fig. 3 Schematic representation of the experimental

set-up for the field measurement of the PMD in
transmission mode.
Fig. 4 presents a spectral dependence of the
polarimetric response.
Fig. 5 shows the statistical distributuon of the
measured DGD a Maxwellian fit [7]
4
32 ( ) 2
f ( ) 2
e
( )3
(14)
- 291 -
-10
lmax2
lmax1
lmax3
-12
-14
, (dB)
-16
-18
-20
lmin3
lmin2
lmin1
-22
-24
-26
-28
-30
1530
1540
1550
1560
1570
1580
1590
1600
1610
Fig. 4. Spectral response in transmission mode.
TABLE 1.
For = 3 degrees of freedom and for a 95%

confidence level the 2 test yields
2 3.157 23 7.82
(15)
The average DGD was measured to be

0.188 ps and the average PMD was found as
g 0.0034 ps / km while the PMD coefficient

is g ,coeff 0.025 ps / km .
From (8) we find the critical length Lc as:
g ,coeff
Lc
g
Fig. 6 OTDR trace to measure the losses obtained

for a fiber length of 57.752 km.
(15)
and the value obtained is Lc 54 km.
0.188 ps
0.148 ps
Bit-rate
SDH format
SONET format
40 Gb/s
STM-256
OC-768
Equivalent time
slot
25.12 ps
DGD limit
PMD coefficient
for 400 km
0.125 ps/km
2.5 ps
The results obtained for the PMD and PMD

coefficient show that the fibers can be used for
building 40 Gb/s communication systems for the
international standards shown Table 1 below.
9. Conclusions
The results obtained permit us to formulate
the following conclusions:
1. The tested schemes of the method used here
are mobile, easy to apply for long-haul
communication lines in field conditions
2. The use of polarization averaging by
changing the input polarization provides responses
with varying spectral periodicities and helps
generate statistical data.
Fig. 5. Statistics of the field measurement of DGD

by the fixed analyzer method in transmission mode.
In reflection mode the total losses in one
direction measured by an OTDR were of the order
of 22.95 dB. With a double pass losses become
approximately 46 dB which surpasses the dynamic
range of the OSA so measurements become
impossible.
3. Measurement in reflection mode demands

either more optical power from the ASE or greater
dynamic range of the OSA.
Ref erences
1. Brandon Collings, Fred Heismann, Gregory
Lietaert, Advanced Fiber Optic Testing HighSpeed Fiber Link and Network Characterization,
Reference Guide to Fiber Optic Testing, JDSU,
2010
2. Gregory Lietaert, Product Manager Testing
Polarization Mode Dispersion (PMD) in the Field,
JDSU , 2006.
- 292 -
3. Daniel A. Nolan, Member, Xin Chen,

Member, and Ming-Jun Li, Fibers With Low
Polarization-Mode Dispersion, Journal of
Lightwave Technology, Vol. 22, Issue 4, pp.
1066 (2004)
4. Paul Hernday, Dispersion Measurements
in Fiber-optic Test and measurements,
Dennis Derickson, ed. Prentice Hall, 1998.
5. Poh-Boon Phua, Deterministic Approach to
Polarization Mode Dispersion, PhD Dissertation,
MIT, USA, 2004
6. Fedor Mitchke, Fiber Optics Physics and
Technology, Heidelberg, Springer, 2010.
7. Sergey Ten, Merrion Edwards, An
Introduction to the Fundamentals of PMD in
fibers,White Paper WP5051, Corning, 2006

Technologies,PU P. Hilendarski,
Plovdiv 4000
BULGARIA
E-mail: vania_plachkova@abv.bg
04.01.2013 .
- 293 -

TECHSYS 2013
BULGARIA
,
:
() ,
.
-.
( ,
, .),
, .
(GUI).

:
,
.
,
. ,

.
: ,
, ,
APPLICATION SOFTWARE BASED ON NONSTATIONARY HEAT MODEL FOR SIMUALTION

OF THE HEAT PROCESSES DURING
ELECTRON BEAM MELTING OF METALS
VELIKO DONCHEV, KATIA VUTOVA
Abstract: Application software for simulation of the heat processes during Electron Beam
Melting (EBM) of metals is developed. It is based on a developed by us non-stationary heat
model and a modified numerical scheme of Pismen-Rekford type is used. The input data are
technological parameters of EBM (the power and radius of the electron beam, heating time,
casting velocity, etc.), thermo-physical parameters of the material, dimensions of the cylindrical
metal ingot, etc. These data are imported via the developed graphical user interface (GUI). The
application software simulates the changes of the temperature fields in the metal ingot during
the heating and in the same time systemizes data about the changes of important parameters
such as: the heat streams through the different ingot boundaries, the geometry of the liquid pool
- 294 -
in the cast block, etc. At the end of the performed numerical experiment, data about the
variation of important main parameters characterizing the EBM process is systemized. The
obtained results give opportunity for studying, investigation and recommendation of suitable
EBMR regimes for process optimization and obtaining of pure metals.
Key words: electron beam melting and refining, non-stationary mathematical heat model,
simulation, visualization, application software
1.
()

.

.[1-5]
, ,
. 1.
,

( )

( / ).

.

,

,
.
2.
. 1. ,

G1 ;
G2 /;
G3 /;
G4 /
()
,

( ) .1.

.
(G1),
,
.
-
(,
)
,
, -

[6-9]
1 T
2T V T C p T
T
(1)
(r
) 2
k
r r r
z
a z
t
t

{( r,z,t )|0 r R,0 z H ,0 t L }

(2)
T( 0,r,z ) T0 ( r,z )
:
T
4
G1:
Psurf (T 4 Troom
) C p .W .T (3)
z
T
T
T
(4)
G2 :
2 water
r
wall
T
4
(5)
G3 :
(T 4 Troom
)
r
T
T
T
(6)
G4 :
2 water
r
under
V T

a z
(.
1). a
. ,[ kW / m2 ] C p ,[W .s / g.K ]

- 295 -
. Psurf

. C p .W .T
4
, (T 4 Troom
)
(
-). 2
.
Twater ;
wall under

(. 1).
,
-
[6]. :
Wh1 ,h2 , {( ri ,z j ,tn )| ri ih1 ,z j jh2 ,t n n ,

i 0,N , j 0,M ,n 0,K }
Ti n, j
( ri ,z j ,tn ) . ,
T n M.N ,
n.
, .. -
()
n- (n+1)-

(. 2).
n+1/2,
. , n
:
( A) k
1
2
1
n
Tn
1T 2 2T n
0.5
T n 1 T
() k
0.5
1
2
1T
1
2
2T n 1
1 2 ,

1 T
2T V T
.
(r
)
z 2 a z
r r r
. 2.
() ()

. (A, )
(r, z)

,
.
3.

():
3.1 () ,
() ;
3.2 GUI ,

;
3.3 (S):

.
() (main)
. ()

ELITE-60 (
- ) ,
.
GUI
, (M)
,
(.2).
()
,
(, ) .,
(S). ,
()
Wh1 ,h2 , .
, , (M)
GUI

(.1),
.
GUI ,
(. 3):
- ,
(. 3);
- ,
;
- 296 -
, (
)

.
(. 4)

(
. 3),
.

.

,
,
(M) ,
GUI (
)

(. 4). ,

/
(G2),
(.4).
, .

( snapshot)
.
. 3. Screenshot GUI ,
- 297 -
. 4. Screenshot GUI
(S) ,

,

:
- 2D

;
- 2D
(G1);
- 3D

( ,
);
-

,
, ;
- ,
-
/ (
) ;
-
( )
(.
1),
;
- ,

,
.
4.
,

-

,
, , .,
.
,
.

( )
.
5.

02-127/2009 (Bin-5/2009),
- 298 -

-
,

Laboratory Physical problems of electron

beam technologies
Institute of electronics, Bulgarian
Academy of Sciences
72 Tzarigradsko shosse
1784 Sofia
BULGARIA
E-mail: velikod@gmail.com
E-mail: katia@van-computers.com
1. , .

, .
, , 2009
2. Bakish R. Electron Beam Melting 1995 to 2005,
Proc. of the 7th Intern. EBT Conference, Varna,
2003, 233-240
3. Bellot JP, Floris E, Jardy A, Ablitzer D. In:
Bakish R. editor. Proc. of the Intern. Conf. Electron
Beam Melting and Refining. State of the Art 1993,
Reno, Nevada, NJ, USA, 1993, 139-152
4. Mitchel A, Wang T. Electron beam melting
technology review. In: Bakish R editor. Proceedings
of the Conf. Electron Beam Melting and Refining.
State of the Art 2000, Reno, Nevada, NJ, USA.
2000, 2-13
5. Mladenov G., Koleva E., Vutova K., Vasileva
V. Experimental and theoretical studies of electron
beam melting and refining, Chapter in a special
review book Practical Aspects and Applications of
Electron Beam Irradiation, ds.: M.Nemtanu,
M.Brasoveanu,
publ.
Research
Signpost/
Transworld Research Network, 2011, 43-93
6. Vutova K., Donchev V. Proc. of 10-th Intern.
Conf. EBT12, Varna, J. Electronics and Electrical
Engineering, Vol.47, N 5-6, 2012, 273-279
7. Vutova K., Donchev V., Vassileva V.
Mladenov G., Thermal processes at electron beam
treatment of metals, accepted for publication in
Journal Metal Science and Heat Treatment,
Springer, 2012
Mladenov G. Influence of process and thermophysical parameters on the heat transfer at electron
beam melting of Cu and Ta, Supplemental
Proceedings: v.1: Materials Processing and
Interfaces, TMS 2012, Wiley, USA, 2012, 125-132
Investigation of electron beam drip melting by a
time-dependent heat model, Proc. of the Intern.
Conf. on High-Power Electron Beam Technology,
Reno, Nevada, USA, 2012
01.03.2013 .
- 299 -

TECHSYS 2013
BULGARIA

, , ,

: ,

.
,
, , .

.
: , , ,
TECHNOLOGY PLATFORM FOR SPECTROMETRIC

ANALYSIS OF WHISKEY
VESELIN NACHEV, PLAMENA MADZHAROVA, TANYA TITOVA,
CHAVDAR DAMYANOV
Abstract: Standardization and quality monitoring of alcoholic beverages is an important
issue in the liquor production industry. Various spectroscopic techniques have proved useful for
the analysis of food and beverages in tasks such as determining the geographic origin, safety,
authenticity, etc. The article focuses on the possibility of identifying specific indicators of
whiskey through a synthesis of the spectral feature space information.
Key words: whisky, authenticity spectral analysis, quality evaluation
1.
20 ,

,
[1].

.
, ,
.

,
.

(
) .

- 300 -
( , ,
.).

[2]. ,

(),
.
2.
.
,

, 19%
2011.).

.
"uisge beatha" , ,
,
,
,
.

, ,
, .
,

[2].
[8].

.
, ,
, , . ,
,

.
,
.
,
.
,
.
,
( , , ,
, .),
[5].
, -
:
.
[10],
.

,

. :
,
,

.

.

, -
[9].

.
,
,
,

. 1%
,
, .

,
, .

.
[5] e

.
,
NIR
,
,
e-nose [10]
.

.

,

.

,

,
.
3.
32
- 301 -
,
.
20 C.

VARIAN, Cary100.
(Cary
WinUV),
.

,

,
.
,
.

( 2/3
).
200 900 nm.

(Ocean Optics USB 4000)
300-1100nm.
.

. .1,

, .
5.5
1.4
5
1.2
4.5
4
3.5
Abs
Abs
1
0.8
0.6
0.2
1.5
0
300
400
500
800
700
600
Wavelength, nm
900
1000
1100
1
200
.2.

( x ), ( xi ) F .

:
K ( x , xi ) = ( ( x ) ( xi )) .

.
SVM (one-class
SVM) :
1 l
(1)
min LP = i=1 i j K ( xi , x j )
2
1 l
: 0 i , i = 1,
vl i=1
: xi ;
i ; v
.
:
3
2.5
0.4
SVM

F
( : X F ).
220
240
260
Wavelength, nm
280
300
320
)
)
.1. NIR 3501100nm () UV ,200-320nm, ()

.2.

,

.

SVM-
.

Brand(i,j), :
i = 1,2 ,.., M , j = 1,2 ,.., mi
i (.2).

.
sv
g 1SVM = i K ( xnew , xi ) ,
(2)
i =1
: xnew ; sv -
.

: ;
- 302 -

(
),
;

( );
.
.3

.
.
7
. .3,,

.
,

v (2)

2
K ( x , x i ) = exp( x x i
Teachers one-class-SVM
2.5
2.5
Teachers
Ballantines
BlackRam
/ ).
Decision function score:

0.7
1.5
0.6
1.5
0.5
PC2
PC2
0.4
0.5
0.5
-0.5
-0.5
-1
-1
-1.5
-2
-1
PC1
-1.5
0.3
0.2
0.1
-2
-1
)

0.25
0.25
0.2
1.5
0.2
0.5
0.15
0.1
-0.5
PC1
PC2
Ballantines one-class-SVM
2.5
1.5
0.15
0.5
0.1
0
-0.5
0.05
-1
-1.5
PC1
BlackRam one-class-SVM
2.5
-2
-1
PC1
0.05
-1
-1.5
-2
-1
PC1
.3.

4.

, ,
(UV-VIS-NIR)
.
,

,
.
1.
.
. .
2001.
2. ., . ., .

. , , 2000.
3. Harrison B.H., F.G. Priest Composition of
peats used in the preparation of malt for Scotch
whisky production influence of geographical
source and extraction depth, Journal of Agricultural
and Food Chemistry 57, 23852391, 2009.
4. Jackson M. Malt Whisky Companion, 6th ed.,
Dorling Kindersley Ltd., London, 2010.
5. Lee K.Y.M., A. Paterson, J.R. Piggott, G.D.
Richardson Sensory discrimination of blended
Scotch whiskies of different product categories,
Food Quality and Preference 12, 109117, 2001.
6. MacKenzie W.M., R.I. Aylott Analytical
strategies to confirm Scotch whisky authenticity.
Part II: Mobile brand authentication, Analyst 129,
607-612, 2004.
7. Moubarik, A., R. Benkirane, R. Soulaymani
Toxicol. Lett. 1998, 95, 71
8. Mosedale J.R., J.L. Puech Wood maturation of
distilled beverages, Trends in Food Science and
Technology 9, 95101, 1998.
9. Nasicmento E. C. L., M. C. U. Araujo, R. K.
H. Galvao A flow-batch analyzer for UV-Vis
spectrophotometric detection of adulteration in
distilled spirits, J. Braz. Chem. Soc. Vol. 22 no. 6
Sao Paulo, June 2011.
10. Novakowski W., M. Bertotti, T.R.L.C. Paixao
Use of copper and gold electrodes as sensitive
elements for fabrication ofan electronic tongue:
discrimination
of
wines
and
whiskies,
Microchemical Journal 99, 145-151, 2011.
Department of Automatic, Information and Control
Systems
26 Maritza Blvd
4002 Plovdiv
BULGARIA
E-mail: v_nachevbg@yahoo.com
E-mail: plamena_madjarova@abv.bg
E-mail: t_titova@abv.bg
E-mail: chavdam@yahoo.com
28.02.2013 .
- 303 -

TECHSYS 2013
BULGARIA
A COMPARATIVE STUDY OF DTW-BASED

METHODS FOR MISSING VALUE ESTIMATION
IN GENE EXPRESSION TIME SERIES
VESELKA BOEVA
Abstract: In this paper, two imputation methods particularly suited for the estimation of
missing values in gene expression time series data are studied: an integrative imputation
approach (IDTWimpute), which uses information that is contained in multiple related data sets,
and a conventional algorithm (DTWimpute) based solely on a single matrix. The both methods
use Dynamic Time Warping (DTW) distance in order to measure the similarity between
different time expression profiles. They are compared one to another by studying their influence
on the quality of gene clustering by using k-means and k-medoids clustering algorithms and two
different cluster validation measures. The integrative DTW-based imputation algorithm is
shown to ensure gene clustering quality performance that is comparable to that of the
conventional DTWimpute algorithm. In addition, the both imputation methods are demonstrated
to be an appropriate imputation solution in case of k-means clustering analysis.
Key words: DTW distance, clustering algorithm, gene expression time series, k-means, kmedoids, missing value estimation
1. Introduction
Microarray experiments frequently produce
multiple missing values due to flaws such as dust,
scratches, insufficient resolution or hybridization
errors on the chips. Unfortunately, most of the
analysis, mining, and classification methods for
gene expression data require a complete matrix of
gene array values. Therefore, the accurate
estimation of missing values in such datasets has
been recognized as an important issue. Various
imputation algorithms have already been proposed
to the biological community. Most of these
approaches are not particularly suitable for time
series expression profiles. In view of this,
Tsiporkova and Boeva have proposed an imputation
algorithm, named DTWimpute, which is specially
suited for estimation of missing values in gene
expression time series data [17]. The algorithm
utilizes Dynamic Time Warping (DTW) distance in
order to measure the similarity between time
expression profiles. DTW has been shown to be a
much more robust distance measure for time series
than classical distance metrics as Euclidean or a
variation thereof since it allows similar shapes to
match even if they are out of phase in the time axis.
Further Kostadinova at al. have developed an

imputation algorithm, referred to Integrative
DTWimpute (IDTWimpute), that is not only suited
for the estimation of missing values in gene
expression time series data but it also uses
information that is contained in multiple related
data sets [8]. The rationale behind the latter is that if
a set of genes exhibit expression similarity to a gene
with missing values over multiple related data sets,
then their profiles may be used in order to derive
more accurate and robust estimation. This would be
useful for estimating data sets with a limited
number of samples, where the information is not
sufficient to select neighbour genes accurately.
Initially, the IDTWimpute algorithm identifies an
appropriate set of estimation matrices by using the
DTW distance in order to measure similarities
between gene expression matrices. Next it employs
the same distance measure to evaluate the similarity
between gene expression profiles and further
identify estimation genes.
Gene clustering is one of the most
important microarray analysis tasks when it comes
to extracting meaningful information from gene
expression profiles. A strong assumption is that
- 304 -
genes with similar expression profiles are likely to

be correlated and/or co-regulated and thus involved
in the same or similar biological processes [13]. de
Brevern et al. [3] have shown that the imputation
method used affects the final clustering, even at a
low rate of missingness. Therefore, choosing an
appropriate imputation method is a crucial step in
the analysis of gene expression data. Motivated by
this, we study herein how the two aforementioned
missing value estimation methods [17, 8] influence
on the output of cluster algorithms and whether they
succeed in restoring the correct gene correlations.
2. Related Work
A series of missing value estimation
methods have been published in the bioinformatics
literature. They can be classified in two broad
categories: ones that use local information and
others that utilize global information. Troyanskaya
et al. have proposed two algorithms, known as
weighted K-Nearest Neighbours (KNNimpute) and
Singular Value Decomposition (SVDimpute),
which are the respective pioneers in these two
categories [16]. The KNNimpute algorithm uses
Euclidean distance to measure the similarity
between two gene profiles and selects for each gene
with missing values a preliminary fixed common
number of candidate genes for estimation. The
SVDimpute finds the dominant components by
summarizing the entire expression matrix and then
predicts the missing values in the target gene by
regressing against the found prevalent expression
patterns. KNNimpute is shown to perform well on
non-time and noisy time series data, while
SVDimpute is better on time series data sets with
low noise level. Few other imputation methods have
also been introduced, either exploiting the local
similarity structure, such as Local Least Square
techniques (LLSimpute) [7], or using the global
correlation among all the genes to impute the
missing entries, e.g., Bayesian approaches in [10].
Recently, several new and innovative
methods that exploit additional information have
been developed. For instance, Oba et al. have
examined the effect of directly merging two
expression data sets from the same study, and found
that this achieved some improvement in the
estimation [10]. An approach that explicitly utilizes
external information is proposed in [19]. It exploits
the functional similarity information embedded in
the Gene Ontology database to enhance the
neighbour gene selection and increase the
imputation accuracy. Further, Hu et al. have
presented an integrative Missing Value Estimation
method that incorporates the information from
multiple reference microarray data sets [6]. This
method first derives a consistent neighbour-gene list
for each gene with missing data, and further uses

the co-expressed profiles for estimation. Gan et al.
have also addressed the missing value imputation
issue by considering the biological characteristics of
the data [4]. Another imputation method that
extracts biological knowledge of the gene using
fuzzy relation has been proposed in [11]. Missing
values are predicted and optimized based on the
extracted knowledge.
3. Methods
Time series expression profiles are expected
to vary not only in terms of expression amplitudes,
but also in terms of time progression since
biological processes may unfold with different rates
in response to different experimental conditions or
within different organisms and individuals.
Therefore, any classical distance metric (Euclidean,
Manhattan, ) which aligns the i-th point of one
time series with the i-th point of the other will
produce a poor similarity score. In view of this, two
methods for missing value estimation of time series
data based on dynamic time warping techniques [1,
14] have been proposed in [17, 8]. The DTW
distance produces a more intuitive similarity
measure allowing similar shapes to match even if
they are out of phase in the time axis (see Fig. 1).
Fig. 1. Time series alignment, classical distance

metric (left) versus DTW distance (right).
3.1. Dynamic Time Warping Algorithm
The DTW alignment algorithm was
developed originally for speech recognition [14]
and it aims at aligning two sequences of feature
vectors by warping the time axis iteratively until an
optimal match (according to a suitable metrics)
between the two sequences is found. Because of its
flexibility, DTW is widely used in many scientific
disciplines and business applications.
Let us consider two sequences of feature
vectors: A = [a1, a2,, an] and B = [b1, b2,, bm].
The two sequences can be arranged on the sides of a
grid, with one on the top and the other on the left
hand side, see Fig. 2. Both sequences start on the
bottom left of the grid. Inside each cell a distance
measure can be placed, comparing the
corresponding elements of the two sequences. To
find the best match or alignment between these two
sequences one needs to find a path through the grid
P = p1,, ps,, pk, where ps=(is, js) and P, referred
to as the warping function, minimizes the total
distance between A and B (see Fig. 2). Thus the
- 305 -
procedure for finding the best alignment between A

and B involves finding all possible routes through
the grid and for each one compute the overall
distance, which is defined as the sum of the
distances between the individual elements on the
warping path. Consequently, the final DTW
distance between A and B is the minimum overall
distance over all possible warping paths:
dtw( A, B)
1
min dist(i s , j s ) .
n m P s 1
Fig. 2. The DTW grid with the optimal warping

path through it.
It is apparent that for any pair of
considerably long sequences the number of possible
paths through the grid will be very large. However,
the power of the DTW algorithm resides in the fact
that instead of finding all possible routes through
the grid, the algorithm makes use of dynamic
programming and works by keeping track of the
cost of the best path at each point in the grid. A
detail explanation of DTW algorithm can be found
in [14, 15].
3.2. DTW-based Imputation Algorithm
Tsiporkova and Boeva have developed an
imputation algorithm, referred to DTWimpute,
which is specially suited for the estimation of
missing values in gene expression time series data
[17]. The algorithm utilizes DTW distance in order
to measure the similarity between time expression
profiles, and subsequently selects for each gene
expression profile with missing values a dedicated
set of candidate profiles for estimation. The latter is
employed for generating an estimation list for each
gene profile with missing values, consisting of
genes with expression profiles which exhibit at least
minimum relative (preliminary defined) similarity
in terms of some distance measure (in this case,
DTW distance) to the gene profile that best matches
the expression profile of the gene with missing
values. These profiles are consequently used for

missing value estimation.
It has been shown that the DTWimpute
offers a more robust and accurate missing value
estimation in gene expression time series than the
KNNimpute developed in [16]. A detail explanation
of DTWimpute algorithm can be found in [17].
3.3. Integrative DTW-based Imputation
Algorithm
An integrative method for missing value
estimation of gene expression time series data,
referred to IDTWimpute, has been proposed in [8].
It is based on the DTW distance [14] and hybrid
aggregation algorithm discussed in [2, 18]. The
considered method imputes missing entities in a
microarray time series experiment using
information that is contained in multiple related
data sets. A detail explanation of the IDTWimpute
algorithm can be found in [8]. Therefore the
description following below is restricted to the main
steps of the algorithm.
Assume that a particular biological
phenomenon is monitored in a few high-throughput
experiments under n different conditions. Thus a set
of n different data matrices M1, M2,, Mn will be
produced, one per experiment. Each experiment is
supposed to measure the gene expression levels of
m genes in a number of different time points. In
order to estimate missing values in any matrix Mi a
set of estimation matrices, which is identified by
considering all n matrices, is used. The estimation
process consists of four distinctive steps:
1) Initial rough imputation of the involved
microarray datasets by filling in the missing values
in a straightforward fashion. For instance, the mean
expression over the respective row or the average of
the two nonmissing neighbours in the row can be
used.
2) Select the estimation matrices based on
the calculated DTW distances between the target
matrix Mi and the rest datasets. For instance, a set of
estimation matrices all at a maximum R-DTW
distance (R is preliminary defined) from matrix Mi
needs to be identified by considering all involved
datasets. The contribution of each estimation matrix
is weighted by the degree of its similarity to the
matrix Mi.
3) Calculate the overall DTW distance
matrix across the estimation datasets. Initially, a
quadratic DTW distance matrix for each estimation
time expression matrix is built by applying the
DTW algorithm described in Section 3.1. Then we
can apply the hybrid aggregation algorithm
described in [2, 18], which after a final number of
iterations will transform the built DTW distance
- 306 -
matrices into a single matrix, consisting of one

overall DTW distance per gene pair.
4) Impute the missing values in the target
matrix. Thus in order to estimate missing values in
any location of gene j of matrix Mi, a set of genes
all at a maximum r-overall DTW distance (r is
preliminary defined) from expression profile gj
needs to be identified. In this process, we use the
overall DTW distance matrix calculated at the third
step and generate an estimation list for each gene
profile with missing values, consisting of genes
with multiple expression profiles (one per
estimation matrix). All these profiles are
consequently used for missing value estimation.
are in the same cluster and 1/j otherwise. Then for a

particular clustering solution of matrix M, which
contains the expression values of m genes in n
different time points, the connectivity is define as
Conn(M) im1 nj1 ximi ( j ) . The connectivity has a
value between zero and infinity and should be
minimized.
5. Results and Discussion
In this section, the impact of the two studied
imputation methods on the quality of gene
clustering is evaluated by using k-means and kmedoids clustering algorithms [9], and Connectivity
and SI validation measures [12, 5].
4. Experimental Setup
4.1. Datasets
The two studied missing value estimation
methods are evaluated on gene expression time
series data obtained from a study examining the
global cell-cycle control of gene expression in
fission yeast Schizosaccharomyces pombe [13]. The
study includes the following 9 different expression
sets: elu1, elu2, elu3, cdc25-1, cdc25-2.1, cdc252.2, cdc25-sep1, elu-cdc10-br, elu-cdc25-br.
A special test data corpus was created as
follows. Initially, all rows containing missing values
were removed from each of the nine original gene
expression datasets. Further, a set of 665
overlapping genes were found across the
transformed original datasets and the time
expression profiles of these genes were extracted. In
this way nine new matrices, which are our complete
datasets, were built. Subsequently, test datasets
were generated by deleting at random 1%, 5%,
10%, 15%, and 20%, respectively, of the data points
from each complete data matrix.
4.2. Cluster Validation Measures
Silhouette Index (SI) [12] is a cluster
validity index that is used to judge the compactness
and separation of a clustering solution. Suppose ai
represents the average distance of gene i to the other
genes of the cluster to which the gene is assigned,
and bi represents the minimum of the average
distances of gene i to genes of the other clusters.
Then the SI of matrix M, which contains the
expression profiles of m genes, is defined as
s(M) 1 / m im1 (bi ai ) / max{ai , bi }. The values of
SI vary from -1 to 1 and higher value indicates
better clustering results.
Connectivity captures the degree to which
genes are connected within a cluster by keeping
track of whether the neighboring genes are put into
the same cluster [5]. Define mi(j) as the jth nearest
neighbour of gene i, and let ximi ( j ) be zero if i and j
Fig. 3. SI values generated by applying the k-means

algorithm on the imputed test matrices estimated by
the DTWimpute and IDTWimpute, respectively.
Fig. 3 depicts the SI values generated by
applying the k-means algorithm on the imputed test
matrices with 1%, 5%, 10%, 15% and 20% missing
values estimated by using the DTWimpute and
- 307 -
IDTWimpute, respectively. It can be observed that

the SI scores obtained on the imputed matrices
evaluated by the IDTWimpute algorithm are very
close to (or slightly worse than) those imputed by
the DTWimpute. Similar results have been
produced on the imputed test matrices under the
Connectivity index.
Fig. 4. Connectivity values generated by applying

the k-medoids algorithm on the imputed test
matrices estimated by the DTWimpute and
IDTWimpute, respectively.
Further the impact of the studied missing
value imputation algorithms on the quality of
produced clustering solutions by using k-medoids
clustering algorithm on the test data matrices has
been investigated. Fig. 4 represents the Connectivity
scores obtained by using the k-medoids algorithm
on all the imputed test matrices evaluated by the
DTWimpute and IDTWimpute, respectively. As can

be noticed the obtained results support those
generated by the k-means clustering algorithm.
Evidently, the integrative DTW-based imputation
approach (IDTWimpute), which uses information
that is contained in multiple related data sets, can
ensure performance with respect to the quality of
gene clustering that is comparable to that of the
conventional DTWimpute algorithm based solely
on a single matrix.
Fig. 5 and Fig. 6 benchmark the SI and
Connectivity scores generated by applying the kmeans and k-medoids algorithms on the nine test
matrices with 15% missing entries estimated by the
two imputation algorithms. It can be noticed that the
cluster solutions produced by the k-means in most
experiments have better performance than the
corresponding ones obtained by the k-medoids for
the both imputation algorithms under the both
validation indices. Similar Connectivity and SI
results have been produced on the test matrices with
1%, 5%, 10% and 20% missing rates.
Consequently, the both studied imputation
algorithms may be considered as an appropriate
imputation solution when the subsequent cluster
analysis of gene expression data will be performed
by the k-means clustering.
Fig. 5. SI values generated on the imputed test

matrices with 15% missing values by applying the
k-means and k-medoids algorithms, respectively.
6. Conclusion
In this paper two imputation methods suited
for the estimation of missing values in gene
expression time series data have been benchmarked
one to another by studying their influence on the
output of clustering algorithms. The integrative
DTW-based imputation algorithm has been shown
to ensure gene clustering quality performance that is
comparable to that of the conventional DTWimpute
- 308 -
algorithm. In addition, the both imputation methods

have been demonstrated to be a more appropriate
imputation solution in case of k-means clustering
analysis than that performed by the k-medoids.
Imputation Method for Gene Expression Time

Series Data, The 6th IEEE Int. Conference on
Intelligent Systems, 2012, 258-263.
9. MacQueen Some methods for classification and
analysis of multivariate observations, Fifth Berkeley
Symp. Math. Stat. Prob., Vol. 1, 1967, 281-297.
10. Oba S. et al. A Bayesian missing value
estimation method for gene expression profile data,
Bioinformatics, Vol. 19, 2003, 2088-2096.
11. Paul A., J. Sil Estimating missing value in
microarray gene expression data using fuzzy
similarity measure, Fuzzy Systems (FUZZ), 2011
IEEE International Conference, 2011, 1890-1985.
12. Rousseeuw P. Silhouettes: a graphical aid to the
interpretation and validation of cluster analysis,
Journal of Computational Applied Mathematics,
Vol. 20, 1987, 53-65.
Fig. 6. Connectivity values generated on the

imputed test matrices with 15% missing values by
applying the k-means and k-medoids algorithms,
respectively.
REFERENCES
1. Aach J., G. M. Church Aligning gene
expression time series with time warping
algorithms, Bioinformatics, Vol. 17, 2001, 495-508.
2. Boeva V., E. Kostadinova A Hybrid DTW based
Method for Integration Analysis of Time Series
Data, Proc. in ICAIS'09, Austria, 2009, 49-54.
3. de Brevern A. G. et al. Influence of microarrays
experiments missing values on the stability of gene
groups
by
hierarchical
clustering,
BMC
Bioinformatics, Vol. 5, 2004, 114.
4. Gan X., A.W. Liew, H. Yan Microarray missing
data imputation based on a set theoretic framework
and biological knowledge, Nucleic Acids Research,
Vol. 34, No. 5, 2006, 1608-1619.
5. Handl J. et al. Computational cluster validation
in post-genomic data analysis, Bioinformatics, Vol.
21, 2005, 3201-3212.
6. Hu J. et al. Integrative Missing Value Estimation
for Microarray Data, BMC Bioinformatics, Vol. 7,
2006, 449.
7. Kim H. et al., Missing value estimation for DNA
microarray gene expression data: Local least
squares imputation, Bioinformatics, Vol. 21, 2005,
187-198.
8. Kostadinova E. et al. An Integrative DTW-based
13. Rustici G. et al. Periodic gene expression

program of the fission yeast cell cycle, Nat.
Genetics, Vol. 36, 2004, 809-17.
14. Sakoe H., S. Chiba Dynamic programming
algorithm optimization for
spoken
word
recognition, IEEE Trans. on Acoust. Speech, and
Signal Proc., Vol. ASSP-26, 1978, 43-49.
15. Sankoff D., J. Kruskal Time Warps, String
Edits, and Macromolecules: The Theory and
Practice of Sequence Comparison, Addison Wesley,
Reading Mass., 1983.
16. Troyanskaya O. et al. Missing value estimation
methods for DNA microarrays, Bioinformatics, Vol.
17, 2001, 520-525.
17. Tsiporkova E., V. Boeva Two-pass imputation
algorithm for missing value estimation in gene
expression time series, J. of Bioinformatics and
Comp. Biology, Vol. 5, No. 5, 2007, 1005-1022.
18. Tsiporkova E., V. Boeva Fusing Time Series
Expression Data through Hybrid Aggregation and
Hierarchical Merge, Bioinformatics, Vol. 24, No
16, 2008, i63-i69.
19. Tuikkala J. et al. Improving missing value
estimation in microarray data with gene ontology,
Bioinformatics, Vol. 22, 2006, 566-572.
Dept. of Comp. Systems and Technologies

4000 Plovdiv, BULGARIA
E-mail: vboeva@tu-plovdiv.bg
13.02.2013 .

TECHSYS 2013
BULGARIA

IPTV

:
.

IPTV .
.
.
: IP ,
WORKABILITY OF MULTIMEDIA PLAYER FOR

BUILDING A HOME IPTV NETWORK
VLADIMIR DIMITROV
Abstract: This paper examines the workability of standard multimedia player for transmission
and reception of streaming audio and video content. Settings and performance of the
application have been tested in real environment and it can be used to build a home IPTV
network. It also provides interoperability between different operating systems and multimedia
players. Thus extends the standard range of options of a media player.
Key words: IPTV, multimedia player
1.

IPTV ,

.
,
IPTV
- IPTV
IP
.
IPTV : [1]
- IPTV (data )
-
- IPTV
.

IP .
-
, ,
.

IPTV ,

. -
Set-top box-.
- 310 IPTV
IPTV
:

,
700$.

, 15 000$
.
.
Set-top box ,
Android
(
50$ ,
). Set-top box
70$
,
.
Android Mini
PC -
,
-
.
, Set-top
box Electronic program guide (EPG
), ,
VideoOnDemand (VoD )
.
2. IPTV
.1 ,

:
-

.
VLC
. VLC

,
, DVD,
/
. [2]

.
- Android Mini PC
Android
Wi-Fi . HDMI

.
BSPlayer FREE
Android. BSPlayer FREE
Android .
[3]
-

Android Mini PC-.

Wi-Fi 802.11g.
. 1. IPTV
3.

VLC Media => Stream (. 2).
(. 3)
,
,
(
).
4,36 GB
1280x720 .
. 2.
VLC
- 311 -
. 5.

. 3. ,

(. 4) H.264.

DVD-
,
IPTV [1].

BSPlayer
URL
.
: http://192.168.7.108:8080 (. 6)
. 4.

(. 5).
HTTP.
(URL) .
. 6. URL BSPlayer

(. 7 . 8).

.

,
25 .
- 312
1. G. O' Driscoll Next Generation IPTV Services
and Technologies, John Wiley & Sons, 2008.
2. http://www.videolan.org/vlc/
3. https://play.google.com/store/apps/details?id=co
m.bsplayer.bspandroid.free&hl=bg
4. Robinson, D. C., Jutras, Y. and Craciun, V.
Subjective Video Quality Assessment of HTTP
Adaptive Streaming Technologies. Bell Labs Tech.
J., 16: 523.2012
5. Dongyu Qiu, On the QoS of IPTV and Its Effects
on Home Networks, International Journal of Digital
Multimedia Broadcasting, vol. 2010
. 7.
6. Lloret, J., Garcia, M., Atenas, M. and

Canovas, A., A QoE management system to
improve the IPTV network. Int. J. Commun. Syst.,
24: 118138, 2011
Department of Computer Systems

Technical UniversitySofia
8 Kliment Ohridski Blvd.
1000 Sofia
BULGARIA
E-mail: vldimitrov@tu-sofia.bg
27.02.2013 .
. 8.
4.
IPTV
.

,

.
5.
,
,
-2012.
- 313 -

TECHSYS 2013
BULGARIA
DESIGN AND FUNCTIONAL DESCRIPTION OF

EXPERIMENTAL SMART GRID
ZDENEK BRADAC, FRANTISEK ZEZULKA, ZOLTAN SZABO,
ZDENEK ROUBAL, PETR MARCON
Abstract: The paper contains a description of an Experimental Smart Grid. The principle of the
proposed design consists in improving stabilization of current grids and enabling maximum
exploitation of electric energy produced by renewable sources (hydroelectric stations,
photovoltaics, wind power plants). In this respect, the authors present a model where the
existing grid can be interconnected with renewable energy plants in such a manner that the
highest possible volume (ideally the total) of electric energy from renewable sources finds
effective application. A block description of the proposed experimental grid is provided,
including functional characterization of the blocks. The most detailed description is given for
the elementary block denoted as Control Block (in the paper, the block is marked as CB1),
which serves both to dispatch energy to the grid and to charge the smart grid batteries. In order
to secure correct operation and functioning of the CB1, the authors applied a step-up converter
and designed a special step-down converter
Key words: control block, energy storage, experimental smart grid, modeling, optimal control
1. Introduction
The Current development of the electrical
energy production and distribution leads from
central production in big classical as well as nuclear
power plants towards combined centralized and
decentralized power sources. The development has
been started with a massive expansion of the
renewable energy source such as photovoltaic, wind
farms and partly with small distributed energy
sources such as cogeneration units in houses, small
energy sources on the basis of the fuel cells
(electrodes are not consumed, fuel as hydrogen,
methanol, etc.), small water turbines on small rivers
and streams with a time limited power production.
All of these small energy sources have a character
of renewable energy sources and their expansion is
a part of the common energy policy of Europe and
all developed countries. This way is on the one hand
very useful and advanced. It also saves on natural
energy resources (cool, water, nuclear fuel, gas, oil)
and reduces impact on the environment, but on the
other hand it brings problems.
The most serious problems in power supply
are prices of electrical energy, significant initial
investments, a difficult prediction of their
efficiency, a stability of the electrical grid which are

problems to control the classic power plant in order
to compensate non-predictable power from
photovoltaics (PV) and wind farms, non-existing
sophisticated monitoring and control systems (smart
metering) in many countries.
In this work the authors make a step in the
experimental smart grid framework. In their
investigation, authors considered revealing of the
National Institute of Standards and Technology [1]
to [5], that there are following the most important
smart grid areas that have to be taken into
consideration by a design of smart grids: bulk
generation, transmission, distribution, customers,
operations, markets and service. These areas contain
many technical disciplines such as Data
Management System, Smart Metering systems,
Communication System, Distributed Automation
Systems, Advanced Metering Infrastructures (AMI),
Sensor Networks (SN), Networked Control Systems
(NCS), SCADA, FACTS and M2M communication
in the near future.
An ambitious final authors goal is a design
of a methodology to stabilize already existing
electrical grids and at the same time to utilize the
- 314 -
maximum of electrical energy from renewable

energy sources. Authors of this article thing, that the
first steps towards a methodology is a representative
modeling of electrical grid with connected the most
usual renewable as well as natural energy sources.
2. Experimental electrical network (E.E.S.)
description
When we make a smart energy network
research (smart grids) with a large share of the
renewable sources, we have to provide partially
physical network modelling. A creating network is
called E.E.S. (Experimental Energy Network).
Small power plants utilizing renewable energy
sources are connected to a grid. Some of these
power plants are present in the Laboratory
Automation in Brno University of Technology,
where the E.E.S. is located [6]. Firstly, there is a PV
panel and small wind power plant which are
situated on a roof of the building. Small
hydropower plant must be placed in other locations
from practical reasons. A hydrogen fuel cell will be
connected to the smart grid in the near future.
Electrical energy of these remote energy sources are
realized and connected into the E.E.S. by means of
RES (Remote Energy Sources) boxes, see Fig. 2.
The RES box is controlled by a CompactRIO and
LabView control system according information
from GPRS wireless data logger which is situated
near the remote energy source to measure its
power Pi.
The physical realization of the E.E.S. is
shown in the Appendix. The thick lines represent
material and energy flows, dashed lines represent
wireless communication links. Each power control
source communicates with the CompactRIO and
LabView control system physically through
input/output circuits via wired connections in the
laboratory. These non-power control connections
are not indicated in Fig.1 for better clarity.
The control system, including hardware
CompactRIO and software LabView, reserves in
individual sources flows of energy and realize the
consumption control. This system can determine
which resources are currently supplying the power
to the grid and what amount of the power. The
E.E.S has to work according the
energy law:
Pi = 0, I = 1, 2 n
(1)
Where Pi is a power delivered by a power source or

consumed by appliance.
The resources supplying energy to the
E.E.S. are a wind and a water turbine, a fuel cell,
solar panels and possibly other renewable sources (a
gas turbine, etc.). Each of them supplies to the
E.E.S. network approximately 200 - 300 W peak.
Given the simplicity of physical realization, safety

experiments and low cost, the E.E.S. works with
24 V DC.
The authors are convinced that this
simplification is acceptable for a generality of the
control algorithm smart grid methodology design.
An object of interest is a switching and controlling
strategy for amount of energy from various energy
sources considering consumption fluctuations and
stochastic changes in a wind, sunshine and partly
changes in a water flow. The designed E.E.S. has at
its disposal number of batteries for an each
individual power source but centralized battery
storage is proposed. Although the storage of
electrical energy into hydrogen seems to be a less
promising, a hydrogen fuel cell (FC) (one metalhydride container as a hydrogen storage) will be
integrated into the E.E.S. in order to realize a peak
energy source [7] .
Physically connected PCs via a simple
DC/AC converters without a need for strict
adherence to phasing and frequency built one
important appliance of the E.E.S. Other appliances,
physically connected to the E.E.S., are heaters,
battery chargers for cars, possibly also electrolyser
(EL) as a backup to store energy by over generation.
The energy consumption is controlled by the
CompactRIO, LabView control system and by the
Control Block (CB2), see Appendix. At this point,
the authors simplified theirs situation for
switching and controlling sources into a common
energy grid the CB2 controls the current (and
power) only because the E.E.S. is a DC grid.
Authors are convinced that the optimal control
strategy, developed for the DC E.E.S., will be
equally valid like if it was developed for a threephase AC grid, where the resources are connected
when synchronization level was entered with a grid
frequency and with a proper phasing. A reason for
such conviction is the fact, that the control of each
energy source frequency and its proper phasing is a
standard question and a standard technical solution
of each AC energy source. These features dont
correspond with higher strategic algorithms of
distribution,
production,
consumption
and
optimality (time optimum, energy and fuel optimal
real time control) of the grids in general. Hence the
higher strategic algorithms which will be
developed, simulated and evaluated in experiments
with the E.E.S., will be appropriate for any other
smart grid.
3. Control and Communication
As already mentioned, most of electrical
energy sources in the E.E.S. [6][7] (the hydro and
wind power plant) will be placed in areas with
favorable weather conditions and information about
- 315 -
their instant power will be transmitted to the control

system (LabView). Based on this information, the
same power as if these sources were physically
connected to the E.E.S., will be physically delivered
to the E.E.S. This finding has two important aspects
for the E.E.S. realization. Firstly, it is necessary to
transmit information about an instant power of these
remote sources into a central control workstation
(CWS), secondly, to control them remotely (connect
and disconnect) and to realize theirs physical
replacement in the E.E.S.
The data transfer from the remote energy
sources is realized with a GSM data logger. Power
meters of these sources are located near the water
and wind power plant and instant power data (when
power changes) is transmitted wirelessly (GPRS) to
the central control and communication system with
CompactRIO and LabView. The laboratory uses
wired connection RS232 (RS 485) or Ethernet
among each controlled source of power (CB1) and
control appliance blocks (CB2), see appendix.
Design of power (electrical) part of E.E.S. is shown
in Appendix.
In the Fig. 1 you can see a physical model
of renewable energy sources (RES).
source. The 24 VDC smart grid is used as the

supporting source. The DC/DC converter is
controlled depending on both the current sensed by
a HY10 sensor and the voltage determined on a
divider; the current and the voltage are measured at
the CB1 input. The applied DC/DC converter is a
special Linear Technology circuit controlled by a
D/A converter via the LabView environment.
The second mode of CB1 allows power the
24 VDC smart grid from local 12 VDC batteries.
For this purpose, we used a Vicor DC/DC
converter, whose output voltage is controlled via
LabView. In general terms, the controlling is
performed through a D/A converter in
CompactRIO; the output of the converter is
connected to the controlling input of the changer.
Here, the controlling is also realized depending on
the output current and voltage, and sensor types
similar to those applied within mode 1 are utilized.
Mutual switching between the two modes is
performed using the LabView controlling interface;
thus, the activity of only one converter (the step-up
or the step-down) at a given moment is secured.
Fig. 1. Physical model of RES

This RES model [6][7] consists of an
AC/DC converter and I/U sense block. The RES
represent E.E.S. physical source (with the exception
of PV source, which is directly physically
connected to the E.E.S.). Local batteries are
supplied with simple uncontrolled AC/DC sources.
The LabView control system manages battery
charging current, based on the information from
voltage and current sense sensor, using voltage
control. For the purposes of a power management of
the actual RES source into E.E.S. is behind RES
block included a controlled block 1- CB1 [6][7]
(see Fig 2). The CB1 can operate in two different
modes:
The first mode of CB1 enables to charge the
local 12 VDC batteries (whose voltage may range
between 11 VDC and 14.5 VDC depending on the
battery charge status). The batteries are charged via
a DC/DC converter, which behaves like a current
Fig. 2. Diagram of controlled power controller CB1

Another physical E.E.S. source is a fuel cell
(FC); it can be connected remotely or physically. In
the case of physical connection, FC is connected via
its own voltage regulator and a power regulator to
the E.E.S. In case of the remote function mode, the
FC is located in different laboratory and supplies
power to the E.E.S. via the CB1 power controller.
4. Appliance control blocks [6][7]
The last one of the power E.E.S. blocks is a
CB2 appliance control block. For a block diagram
see a Fig. 3. A DC current from E.E.S. is controlled
using the LabView and if needed (for AC
appliances) it is converted by a low-cost unmanaged
converter to 230 V AC.
For financial reasons we do not realize a
sale of an energy over generation in the public
network since the corresponding converters are
- 316 -
significantly more expensive and for their own

functions and an E.E.S. role as an experimental
network for development and testing of
management strategies of renewable energy sources
and prediction their performance, this option would
provide no additional problem to be considered in
our investigation.
REFERENCES
1. N. A. Hidayatullah, Stojcevski B., A. Kalam:
Analysis of Distributed Generation Systems, Smart
Grid Technologies and Future Motivators
Influencing Change in Electricity Sector, Smart
Grid and Renewable Energy, 2011, 2, pp.216-229 2.
2. G
rid
2030A
National
Vision for
Electricitys Second 100 Years U.S Department
of Energy, Office of Electric Transmission and
Distribution,
July
2003
Available
from
http://www.oe.energy.gov/DocumentsandMedia/Ele
ctric_Vision_Document.pdf
3. National Institute of Standart and Technology,
NI
ST
and
the
Smart
Grid,
2010.
http://www.nist.gov/smartgrid/nistandsmartgrid.cfm
Fig. 3. Control unit CB2 of appliance

5. Conclusion
The whole E.E.S. model aims to simulate a
real power network with a large share of renewable
sources and will serve mainly for the development
and validation of advanced control algorithms of
power system, utilizing the most comprehensive
information on the temporal power distribution of
different types of renewable resources in an island
smart grid.
Research will be focused on optimizing the
volume of information transmitted from remote
energy sources and from monitoring system of
energy consumption to CWP (a cooperation
between the smart grid and the smart metering) and
on the control strategy design for energy storage
and control of the network stability using both
resources management and retail consumption
control (households, institutions, retailers). As result
should be also a feasibility study of a smart grid
stabilization (a stand-alone operation as well as the
entire network segment with a significant share of
renewable resources).
Acknowledgements
This work was supported by Operational
Program Research and Development for
Innovations; project CVVOZE Center for
Research and Utilization of Renewable Energy
Sources no. of contract CZ.1.05/2.1.00/01.0014 and
Brno University of Technology.
04.03.2013 .
4. European Commission, Eur

opean Smart Grid
Technology PlatformVision and Strategy for
European Electricity Networks of the Future, 2010.
http://www.smartgrids.eu/web/node/28/default.asp
5. Office of the National Coordinator for Smart
Grid Interoperability, National Institute of
Standard and Technology, U.S. Department of
Commerce, N
IST Framework and Roadmap for
Smart Grid Interoperability Standard, Real-ease
1.0, NIST Special Publication 1108 on the January
2010.
6. ZEZULKA, F.; BRAD, Z.; SAJDL, O.;
EMBERA, J. Experimental Smart Grid. In
Proceedings of 11th IFAC/IEEE International
Conference on Programmable Devices and
Embedded Systems. Brno: IFAC-PapersOnLine /
Elsevier, 2012. ISBN: 978-3-902823-21- 2.
7. VESEL, I.; ZEZULKA, F.; EMBERA, J.;
SAJDL, O. Problems of energy saving in Electrical
Experimental Network/Smart Grid. In Advanced
Batteries Accumulators and Fuel Cells - 13th ABAF
Book of Proceedings. 2012. s. 1-8. ISBN: 978-80214-4610- 6.
CVVOZE Department
Department of Control and Instrumentation
Comunication
Brno University of Technolgy
Technicka 12
61600 Brno
CZECH REPUBLIC
E-mail:zezulka@feec.vutbr.cz
E-mail:bradac@feec.vutbr.cz
E-mail: szabo@feec.vutbr.cz
E-mail: roubal@feec.vutbr.cz
E-mail: marcon@feec.vutbr.cz
- 317 -
Appendix: E.E.S. block diagram
CB1
CB2
CL
SI
BAT
RES
EL
S
FC
ST
G1
G2
G3
Heat
Cont 1
Cont 2
CWP
source block (controlled power source 0 300 Wss)

control block of energy consumption of individual appliance (LabView controlled)
current limiter (d-Space control of accumulator charge)
sensor of DC current
accumulator of DC energy
physical model of remote energy source
electrolyzer
valve to switch and control the hydrogen flow to the fuel cell (FC)
fuel cell
suppliers stand for e-cars
wind turbine generator 12 Vss, 160 W
water turbine generator 230 V, 50 Hz, 500 W
virtual model of a gas turbine
electric heater
metal-hydride container of hydrogen
pressure flask of hydrogen
control work place
- 318 -
- 319 -

TECHSYS 2013
BULGARIA
X-RAY DIFFRACTION ANALYSIS OF RESIDUAL

STRESSES IN LASER WELDED STEEL SHEETS
ZDENEK PALA, KAMIL KOLAK, NIKOLAJ GANEV
Abstract: Advent of high power diode laser has substantially contributed to the popularity of
laser welding in industry where its virtues such as low heat input and good weld strength are
highly appreciated. However, one of the drawbacks of the laser welding is distortion of the
welded bodies that is closely linked with the generation and/or redistribution of residual
stresses in the vicinity of the weld. In this contribution, we have striven to bring insight not only
into the distribution of surface residual stresses after laser welding, but also perform qualitative
assessment of the surface structure by means of X-ray diffraction.
Key words: laser welding, residual stresses, X-ray diffraction
1. Introduction
The cost effectiveness, non-contact
character, high degree of automation and the
resultant good reproducibility make laser welding
ever more attractive for automotive and aeronautics.
Industrial applications of laser welding using highpower diode laser (HPDL) are becoming more
numerous since the usage of this type of laser was
firstly published more than 20 years ago [1].
Multiple beam interaction of HPDL and its shorter
wavelength in comparison with conventional CO2
and Nd:YAG lasers is responsible for smaller heataffected zone, better absorption of the beam energy,
fewer cracks and lower porosity or in general more
compact microstructure [2]. On the other hand, the
HPDL beams are divergent and, hence, problematic
to be focused into small beams which would be
beneficial for welding of small components.
The fatigue life of an object with a laser
weld or, more universally, its performance depends
on several parameters of structure, most notably on
the residual stresses which are considered as
important criterion of the weld quality [3]. The aim
of the laser welding process is not only to produce a
joint, but also to avoid generation of tensile residual
stresses that would significantly speed up crack
propagation and to minimalize the distortion of the
final object. It has been found that distortion and the
macroscopic residual stresses after welding are two
mutually affected phenomena and, therefore, upon
controlling the residual stresses by the means of
laser beam energy and speed of the weld creation,

one can influence the undesirable distortion as well.
Determination of residual stress distribution
in the vicinity of laser welds can be performed by
employing by one of X-ray diffraction (XRD)
methods, when the inter-planar lattice spacings are
measured and then used for stresses calculation with
the help of generalized Hookes law [4] and
appropriate elastic constants. However, there are
some limitations to the applicability of XRD for
establishing the state of residual stress in
polycrystalline materials. Namely occurrence of
crystallographic texture and too coarse material [5],
the absolute extreme being the case when the
irradiate volume lies in a single crystallite, belong
to the state of the material when the standard
methods fail to give reliable results.
In this contribution, mapping of surface
macroscopic residual stresses in two directions, i.e.
parallel and perpendicular to the welds, were
performed for two bodies. The first contained a
weld created with the HPDL beam speed of
2 m/min and the second with the speed of
15 m/min. Our aim was to compare not only
resulting fields of residual stresses, but also to
perform qualitative assessment of the possible
presence of crystallographic texture and gain a
qualitative apprehension about the grain sizes in the
vicinity of a laser weld joining two steel sheets.
Moreover, diffraction line broadening as a
- 320 -
parameter of the degree of plastic deformation was

computed from the measured profiles.
2. Experimental
The analysed bodies were manufactured by
welding of two sheets made form S355 steel. Power
of the HPDL was set to 3.5kW; the speed of
welding was 2 m/min and 15 m/min. The lengths of
the welds were approximately 180 mm.
Distortions of both bodies were measured in
six equidistant points in the direction perpendicular
to the welds; the obtained values were, thus, 6
angles of deflection.
The structure of the welds and the adjacent
areas of the steel sheets were qualitatively
characterized by 2D diffraction patterns, or more
precisely by Debye rings of {211} planes of -Fe
obtained in the backscattering layout of DebyeScherrer method. For this purpose, ISO
DEBYEFLEX 3003 apparatus, non-filtered
radiation from X-ray tube with chromium anode,
cylindrical primary slit with 1 mm in diameter and
image plate detector were employed.
Surface distributions of macroscopic
residual stresses were established perpendicularly to
the welds in eleven areas mutually shifted by
0.5 mm. Each analysed area had a rectangular
shape with dimensions 10 0.5 mm2; the longer
sides being parallel with the welds. Stresses were
determined in the directions parallel with the welds,
longitudinal stresses L, and perpendicular as well,
transverse stresses T.
We assumed biaxial state of stress and used
the sin2[6] method coupled with WinholtzCohen least squares fitting procedure [7] to
compute both longitudinal and transverse stresses.
The measured diffraction profile of -Fe {211}
planes has for the used filtered CrK radiation it
maximum at 2 156. Detected doublets were
separated by Rachinger method [8] to the
contributions from CrK1 and CrK2. Diffraction
profile corresponding to CrK1 radiation was fitted
by Pearson VII function. Maxima of this function
for all measured profiles served as input data for
inter-planar lattice spacings calculations. In the
generalised Hookes law, we used X-ray elastic
constants s1 = -1.25 TPa-1 and s2 = 5.76 TPa-1
obtained with the help of Eschelby-Krner
model [9]. Eventually, the diffraction profile
corresponding to -Fe {211} planes parallel with
the surface was characterized by FWHM (Full
Width at Half Maximum) profile parameter which
represents another parameter, often described as
degree of plastic deformation, because the
diffraction profile broadening can be related to such
phenomena as grain size, microscopic residual
stresses or dislocation density whose evolution is

closely connected with plastic deformation.
Fig. 1. In-house designed versatile positioning

system for XRD analyses of real objects-.
Diffraction measurements were carried out
on vertical - XPert PRO MPD diffractometer
equipped with monocapilary optics in the primary
beam, i.e. the beam impinging on the analysed
sample was pseudo-parallel. Positioning of the
measured objects to the coveted locations was done
by combining versatile positioning system seen in
Fig. 1 which with six degrees of freedom and laser
triangulation for precise surface position
determination with accuracy of approx. 5 m.
3. Results
Prior to the XRD analyses of residual
stresses by sin method, diffraction patterns
capturing {211} -Fe Debye rings were obtained.
Table 1 shows the selection of three patterns: one
form the initial state of the steel sheets prior to laser
welding and one from each object 2 mm far from
the welds boundaries. The six angles of deflection
for each object are summarized in Tab. 2. Surface
distributions of macroscopic residual stresses can be
seen in Figs. 2 and 3 as well as corresponding
FWHM parameters in Fig. 4 and 5.
- 321 -
1,80
600
15 m/min
FWHM, 2
, MPa
1,70
200
0
-200
tranverse direction
longitudinal direction
Fig. 2. Surface distribution of residual stresses in

the vicinity of the laser weld created with the HPDL
beam speed of 0.25f m/s.
600
, MPa
1,60
1,50
Fig.5. Surface distribution of FWHM of -Fe {211}

diffraction profile of planes parallel with the sample
surface; measured in the vicinity of the laser weld
created with the HPDL beam speed of 2 m/min.
2 m/min
0
-200
-400
2 m/min,
2 mm from
the welds
boundary
distance from the weld's boundary, mm
Fig. 3. Surface distribution of residual stresses in

the vicinity of the laser weld created with the HPDL
beam speed of 2 m/min.
15 m/min,
2 mm from
the welds
boundary
3,2
3,0
transverse direction
15 m/min
FWHM, 2
2,8
2,6
2,4
2,2
2,0
Table 1 Backscatter diffraction patterns with Debye

rings of -Fe {211} planes.
Sample,
Diffraction pattern
area
400
1,65
1,55
200
1,75
400
-400
2 m/min
state prior to
laser
welding
Fig. 4. Surface distribution of FWHM of -Fe {211}

diffraction profile of planes parallel with the sample
surface; measured in the vicinity of the laser weld
created with the HPDL beam speed of 15 m/min.
- 322 -
Table 2 Deflection of both analysed bodies in seven

equidistant points in the direction perpendicular to
the welds.
Deflection of
Deflection of
Point
sample
sample 2 m/min
15 m/min
1
-1 4
2 20
2
-1 3
2 25
3
-1 3
2 25
4
-0 8
2 25
5
-0 7
2 20
6
-0 8
2 20
4. Conclusions
Considering all the results from the analysis
of two objects with laser welds prepared with
HPDL laser and two distinctive beam speeds
entitles us to make the following conclusions.
The structure of the material in the vicinity
of welds was suitable for residual stress
determination by means of XRD.
From the comparison of the Debye rings
from both samples (Table 1) emerges the fact that
the structure in the vicinity of the weld created with
lower speed is distinguished by larger grains of
b.c.c. iron. It is, thus, coarser-grained than the areas
adjacent to the weld done with much higher speed
of 15 m/min (15 m/min). Taking the effect of
temperature-related phenomena, the performed
measurements affirm that laser welding with lower
beam speed leads to more pronounced heat impact
of the structure.
On the surface of both the objects,
anisotropic biaxial state of stress was established,
i.e. T L. Residual stresses L in the direction
parallel with the welds are tensile in all measured
areas and decrease in value with larger distance
from the welds boundaries. The decline has
monotonous character for the sample 2 m/min, but
oscillations between neighbouring areas occur in the
case of the body with 15 m/min weld.
Whereas
character
of
L
surface
distributions is similar, the stresses in perpendicular
direction to the welds are qualitatively different for
both bodies.
The stresses T are within comparatively
narrow interval from -70 to -13 MPa for the sample
2 m/min which is in stark contrast to the other
sample where the stresses rise from approx. 300
MPa to approx. 150 MPa within the 1.5 mm wide
area fast beside the welds boundary.
From the comparison of all four surface
distributions of {211} -Fe FWHM parameter, it is
visible that higher values are seen for faster beam
speed, see Fig. 4 versus Fig. 5. Both areas closest to
the welds show apparent increase in this parameter,
but only for diffraction profiles measured in

longitudinal directions, i.e. parallel with the welds.
Larger distortion of the body with a laser weld is
exhibited by the sample manufactured with
approximately 8 times bigger speed of HPDL beam.
This sample is in the immediate vicinity
characterized by substantial compressive residual
stresses in the direction perpendicular to the weld.
Acknowledgements
This research was carried out in the frame of
research projects TA02011004 (Technology
Agency of the Czech Republic) and FR-TI3/814
(Ministry of Industry and Trade of the Czech
Republic).
References
1. Li L., Optics and Lasers in Engineering 34
(2000) 231-253.
2. Lawrence J, Li L. J Phys D 32 (1999) 10751082.
3. Olabi A.G., Benyounis K.Y., Hashmi M.S.J.
Strain 43 (2007) 37-46.
4. Kraus I., Ganev N.: Residual Stress and Stress
Gradients, In: Industrial Applications of X-Ray
Diffraction. New York: Marcel Dekker, 2000, p.
793-811.
5. Leoni M., Scardi P., Thin Solid Films 345
(1999) 263-269.
6. Macherauch E., Mller P., Zeitsch. angew.
Physik 13 (1961) 33 38.
7. Winholtz R.A., Cohen J.B., Aust. J. Phys. 41
(1988) 189-199.
8. Rachinger W.A., J. Sci. Instrum. 25 (1948) 254 259.
9. Eshelby, J. D., Proc. Roy. Soc. A 241 (1957)
376 396.
Department of Solid State Engineering
Faculty of Nuclear Sciences and Physical
Engineering
Czech Technical University in Prague
Trojanova 13
120 00 Prague 2
CZECH REPUBLIC
E-mail: zdenek.pala@fjfi.cvut.cz
E-mail: kamil.kolarik@email.cz
E-mail: nikolaj.ganev@fjfi.cvut.cz
15.02.2013 .

TECHSYS 2013
BULGARIA
PHYSICAL LUNG SIMULATOR FOR EX VIVO

MEASURING OF AEROSOL DEPOSITION IN
LUNGS
ZEZULKA F., DRAUSCHKE A., BURE Z., KREJ .I, BALCAR J., PROCHZKA M.
Abstract: With the current exposure to aerosols, nanoparticles and fine dust the cases of
pulmonary diseases increase. Nowadays there is still little information about the distribution of
inhaled particles in the lung itself.
The presented lung simulator i-Lung 2.0 is an active mechanical lung simulator, which offers
the use of different lung equivalents, like a primed porcine lung or latex bags. The use of live
lungs from pigs is proposed to be used as well. The i-Lung uses a non destructive aerosol
measurement system for measuring the size and amount of in- and exhaled particles that were
produced beforehand.
Mechanical, electronically as well as microcontroller based control and communication
systems are briefly presented in the contribution..
Key words: control, lung simulator, aerosol, nanoparticles, vacuum
1. Introduction
With the current exposure to aerosols,
nanoparticles and fine dust the cases of pulmonary
diseases increase. Nowadays there is still little
information about the distribution of inhaled
particles in the lung itself. However, this
information is important for pharmaceutical
industry providing inhalable diagnostics and
therapeutics. The presented lung simulator i-Lung is
an active mechanical lung simulator, which offers
the use of different lung equivalents, like a primed
porcine lung or latex bags. The i-Lung uses a non
destructive aerosol measurement system for
measuring the size and amount of in- and exhaled
particles that were produced beforehand.
This lung simulator is a first step into the
direction of replacing laboratory animals for
inhalation test as ordered by the EU REACH
regulation.
One possible solution to reduce the amount
of needed animals is the here presented lung
simulator i-Lung 2.0. This lung simulator uses a non
destructive aerosol measurement system and is
designed to work as a passive or active lung
simulation tool. The i-Lung can simulate different
physiological and pathological breathing patterns
and measure inhaled and exhaled particles using a
white light aerosol sensor. Furthermore, the i-Lung

can use different lung equivalents such as a isolated
primed porcine lung and has the foundations to be
used as an ex vivo aerosol measurement.
The contribution deals with mechanical
realization, design and realization of electronics for
signal processing from sensors and the design of the
micro - controller based control system.
2. Device overview
The i-Lung 2.0 is an active physical lung
simulator. Due to the use of different modules its
use and setup are flexible. With these i-Lung
modules different pathological breathing situations
can be simulated and inhaled and exhaled particles
can be produced and detected. In this text the iLung module, sensor network and the used
electronics will be presented, see Figure 1.
2.1 The i-Lung module
The core element of the i-Lung module is
the "thoracic chamber". Chamber is made from
plexiglass. A negative pressure is created within the
chamber to inflate the lung equivalent. Lung
equivalents can be mounted via the connective
flange. The chamber is connected to a bellows
system and a vacuum pump. The vacuum pump
creates a constant negative pressure in order to
- 324 achieve a more realistic anatomical and

physiological breathing situation. The constant
negative pressure represents the adhesion of the
human lung to the thorax through the pleura. The
bellows system induces differences of the pressure
within the thoracic chamber due to the movement of
the bellows. The bellows movement of compression
and extension is in direct correlation to the ballscrew in the middle of the bellow. A motor rotates
the ball-screw which has the effect that the bottom
of the bellows is moved downward or upward. The
pressure changes within the thoracic chamber cause
a inflation and deflation of the used lung equivalent.
Speed of bellows expansion or contraction
is continuously controlled by comparing the air
flow F to or from lung equivalent with the desired
course of respiratory curve F = f (t) (Figure 2).
System is regulated and controlled by a controller
on the basis of flow sensors, the differential
pressure inside the chamber. Continuously are
scanned also temperature and humidity in the
chamber and in the ambient. UV sensor provide
auxiliary function measurements during sterilization
by UV radiation.
2.2 Electronics
Electronics consist of single board
computer (SBC03), display and keyboard. For more
complex setting of parameters and drawings of
graphs PC can be used. PC is connected to SBC03
via Ethernet cable.
The SBC03 contains a ARM9 (Linux OS)
which is connected to a Cortex M3 (real time
processor) via SPI. The ARM9 gets input from the
PC or display and forwards the according signals to
the Cortex M3. As a result the Cortex M3 sends
signals and commands to the Driver of the motor
and the power switch of the vacuum pump.
Additionally the Cortex M3 collect all data from the
sensor network and the limit switches of the
bellows. According to the incoming signals of these
sensors and switches the M3 again send signals to
driver and vacuum pump as well as data to the
ARM9 for further processing. This data then can be
displayed numerical or as graphs on the PC and
display.
2.3 Sensor network
Normal respiratory cycle, ie, inhale and
exhale for a period T = 2 s, ie, the frequency of 0.5
Hz. For sampling during the reference breathing
curve, according to which the regulation will take
place, has been selected 100 samples per cycle, ie,
Ts is the sampling frequency of 50 Hz.
Fig. 2. Curve of physiological breathing

The sampling frequency allows implement
appropriate modulation signals that occur in
pathological breath, a frequency less than 25 Hz.
The reference respiratory curves are stored in the
data memory of control computer as flow
dependence on the time from which will be released
according to the type of breathing. From memory
the data will be released with the mentioned
sampling frequency, which will also be sampling
frequency regulation loop. From this sampling
frequency also implies the need for equally quick
sampling real the flow of air into the lungs of the
model. Other auxiliary variables (temperature,
humidity, pressure, etc.) are characterized as slow,
and therefore these data do not have to be captured
at each sample frequency (sampling rate 1 sec).
This can be used for time-division multiplexing of
samples to reduce the capacitive load for line. For
this purpose RS485 communication protocol was
used with baud rate of 115 kbaud.
The sensor network currently consists of
seven sensors:
Box Flow meter : Honeywell
AWM720P1
Box Temperature : Heraeus WEXK6
Box
Differential
Pressure
:
Freescale MPXV2010DP
Box Humidity : sensirion SHT75
Ambient Temperature : Heraeus WEXK6
Ambient Pressure : Freescale
MPX5100DP
Ambient Humidity : sensirion
SHT75
All of them are connected to the Cortex M3
using the RS 485 communication protocol and
power with either 12V or 5V.
- 325 -
3. Control and sensor part

3.1 Introduction
Two-stage solution of the physical model of
the artificial lung was used, the first stage consists
of main control board processor based on Cortex
M3 and the second one consist of intelligent sensors
connected to the network via a standard RS-485
line.
3.2 Control part
Measurement of air flow in and out of the
lungs is a key variable according the speed and
direction of the motor is regulated during
respiratory cycle. For each direction of flow one
sensor is used because of their big non linearity in
opposite flow directions. Sensors are connected in
series.
Subordinated processors are based on lowpower MCU units TI MSP430FE2xx (Figure 4),
which in themselves have implemented 24-bit AD
converter and a universal asynchronous serial
channel type UART.
3.3 Sensor part
Using communications physical layer RS
485 is an open system for expansion requirements
for connecting additional sensors or actuators,
implementing
auxiliary
control
systems
(stabilization in temperature or humidity, etc.). Data
layer will form efficient communication protocol
that ensures sufficient throughput communication
lines as well as for expansion of the system. The
device will be designed so that both could
independently control the operation of artificial
lungs and also to cooperate with the master system.
Control and regulation loop is based on the
airflow sensors. As an airflow sensor is used a
Honeywell AWM720 microbridge mass airflow
sensor, which provide in-line flow measurement
(Figure 3) with a specially designed bypass flow
housing. The sensors measure flow as high as 200
standard liters per minute (SLPM) while inducing a
pressure drop of 1 inch H2O, typically. The
AWM700 has a high flow range capability in a
small package. The AWM720 has a 6 millisecond
response time, requires a 10 Vdc supply, but
consumes only 60 mW of power. It has analog
output from 1V (0 SLPM) to 5V (200 SLPM)
Conclusions
Contribution deals with a physical
realization of a lung model. The i-Lung 2.0 is an
active physical lung simulator. The goal of
investigation is in research of influence of aerosols
on human body (lung) by breathing. Particular
attention by such an investigation is in replacing
laboratory animals for inhalation test as ordered by

the EU REACH regulation.
Authors describe electro - mechanical
model of the i-Lung 2.0 system, as well as the
electrical and particularly sensor, communication
and microcontroller based control sub - systems.
The i-Lung 2.0 system enhances the previous
models of artificial lung was equipped by a pump to
insure vacuum around the lung. The system i-Lang
2.0 uses for the negative pressure a bellows, which
has more appropriate relation weight power.
Nevertheless it is necessary to compensate leakages
in the bellows by more smart control strategies of
additional air pump.
Acknowledgement
This work was supported by cross-border
project of the College Of Polytechnics Jihlava and
University of Applied Sciences Technikum Wien
ELBIK.
References
1. Forjan M, Stiglbrunner K, Zbynek B,
Drauschke A: Overview of the "i-Lung" as
developing active lung simulator including
respiration aerosol measurement. Pittsburgh: ACTA
Press, 2011
2. European Commission: REACH. European
Commis-sion, 2012.
Available
at:
http://ec.europa.eu/environment/chemicals/reach/rea
ch_intro.htm
3. Ari A, Hess D, Myers T, Rau J: A Guide to
Aerosol Delivery Devices for Respiratory
Therapists. 2nd Ed. American Association for
Respiratory Care, 2009
4. Forjan M, Stiglbrunner K, Steiner T, Zbynek
B, Drauschke A: Sensor System Development for
the Novel Spontaneous Active Breathing Lung
Simulator, i-Lung. Vienna: UAS Technikum Wien,
2011
Department of Electrotechnics and Informatics
College of Polytechnics Jihlava
Tolsteho 14
58601 Jihlava
CZECH REPUBLIC
E-mail: frantisek.zezulka@vspj.cz
E-mail: zbynek.bures@vspj.cz
E-mail: ivan.krejci@vspj.cz
E-mail: mprochazka@rwetc.com
E-mail: jbalcar@nbox.cz
E-mail: andreas.drauschke@technikum-wien.at
04.03.2013 .
- 326 -
Fig. 1. Schematic diagram of the i-Lung 2.0 setup
- 327 -
Fig. 3. Voltage Output vs. Flow
Fig. 4. Schematic wiring converter RS 232 to RS 485

interface for connection to a sensor system model of the
artificial lung. The converter uses the TI processor
variant with two asynchronous serial UART
- 328 -
- 329 -

TECHSYS 2013
BULGARIA
SMART GRID SMART METERING SYSTEM

ZEZULKA FRANTISEK, BRADAC ZDENEK, SAJDL ONDREJ, SIR MICHAL,
VESELY IVO
Abstract: This paper deals with the state of the art in a energy consumption area of small and
medium energy users. Existing systems for energy consumption monitoring are intended mostly
for big companies, institutions, buildings, technologies with a great energy consumption. The
paper deals with an idea of energy savings and simultaneously about a smart grid stabilization
with an important deal of renewable energy sources. Utilizing a cheap flexible individual smart
metering system could small and medium companies and households move individually energy
consumption into a more available time zone. It will lead not to only lower a bill for energy in
households and by other small energy consumers, but also to a better grid stabilization. A
prerequisite for successfulness of such a system is an implementation of a dynamic charge per
energy unit.
Key words: smart grid, smart metering, energy saving, dynamic tariff, renewable energy
sources
1. Introduction
A stabilization of an energy grid could be
done by both, a control of energy sources and a
control of energy consumption. The basic principle
of any electrical grid is as follows: sum of power
from all energy sources must be equal the sum of
power consumption of all connected appliances in
any moment. In the present time and still more in
very near future, existing energy grids will be
negatively influenced by non regular energy
production and their control. The reason is the
continuously growing ratio of connected renewable
energy sources. Wind and PV farms produce a
stochastic ratio of energy. It is very dangerous for
existing control strategies of electrical grids. The
classical power plans cannot be effectively
controlled to compensate the stochastic energy ratio
of renewable energy sources. But a second
possibility for the grid stabilization could be done
by control of energy consumption. One form of
such a system is in many countries already realized
by the planned lower energy fees during night and
apart from the peak energy request period of the day
and also by a quarter-hour maximum. On the other
hand it would be possible to utilize still a new
control mechanism by control of a huge amount of
households and lower energy consumers. This

mechanism should be more dynamic and could
really stabilize grids with very dynamic energy ratio
from renewable energy sources. The most important
step to motivate a huge amount of individual
household and small energy consumers should be
an implementation of a dynamic tariff of electrical
energy. Such dynamic tariff should offer a lower
price for energy by a big energy production from
wind and the PV farms and increase the price for
energy unit by energy deficiency.
The second prerequisite for the proposed
stabilization methods is a massively implemented
cheap, effective and user friendly measuring system
in households and by lower energy consumers. Such
smart metering systems are already developed and
used in several countries, but mostly for metering
purposes of big energy consumers. These systems
are not developed for purposes of grids stabilization
but mostly for energy saving only. Our intention is
to develop and to implement a smart metering
system for households and to connect it by smart
control strategies by use of a prediction system,
weather forecast internet sources, supplementary
sensor networks and etc.. This smart metering
system will deliver sufficient actual information
- 330 -
into individual smart decision system in individual

households and by smaller energy consumers to be
used for smart control of the individual energy
consumption.
2. State of the art in the area of smart
metering systems
As mentioned in the previous chapter, there
are several smart metering systems in the market.
Let us introduce some of them.
ESA2000 wireless power consumption meter
dealer: Tipa [1]
System enables to measure the whole
consumption of the household in the electric meter,
a day comparison with the previous year,
visualization of actual price and simple
commissioning of the ESA 2000. The
communication with the basic station of the user is
WLS. The daily data of consumption are available.
It does not enable to store data on SD card or to
store them in a PC.
DT23 wireless power consumption dealer:
Elektro Palouek [2]
It is a simple system intended to put into a
wall socket with two WLS measuring models and
with one central station. No connection and storage
in SD card or in a PC.
System VOLTCRAFT Smart Metering VSM-120
80A HS manufacturer VOLTCRAFT [3]
The system is smart enough for current
measuring, and energy counting. Actual the whole
energy consumption is available. Data from the
meter are by WLS connected to a PC to be
processed by an existing user program.
HOBO U30 Series Energy Monitoring Systems
manufacturer: Onset (USA) [4]
Measuring system for general outdoor
measuring or measuring of an electrical energy
(current, voltage) and some supplementary values
(pressure, differential pressure, concentration of O2
and other for purposes of energy monitoring not
valuable values) It has several communication
interfaces (Ethernet, WIFI, USB, GSM). A WEB
software
interface
enables
configuration,
monitoring and saving of energy and others.
Wi-LEM - Wireless Energy Meter manufacturer:
Synetica Limited (UK)
The most sophisticated system from this
list, WLS connection sensors with the basic station.
Basic station contains a web server and data can be
processed and visualized via internet (e-mail and
others). Single sensors can be simple attached on a
wire by a clip and send wirelessly process data to

the basic station. The sensor network topology is
the mesh type.
3. HW solution of the SMART ENERGO
system
In the Fig. 1 is shown the complete project
conception of the system for monitoring, planning
and energy saving with stabilization of a grid with
important ratio of renewable energy sources [7].
The concept of the system goes out from the
fact that households are equipped mostly by one or
three phase electrical installation with one central
distributor. There are current or voltage protectors
and energy meters. Next the electrical installation is
split into more light and sockets networks, which
are protected by one or three phase protectors. From
the electricity meter the electrical installation is user
available. The system SMART - ENERGO uses two
types of energy monitors [7].
The first one is the SM1 (Smart Meter 1)
[7] and it makes monitoring of the whole energy
consumption of the flat or house (households) or the
whole energy consumption of a small energy
consumer (SMS, institutions, offices, etc.). The
SM1 is situated at the distributor behind the energy
meter and monitors the whole energy consumption
in one or three phases. The second element [7] of
the system SMART - ENERGO is the SM2. It is an
element for energy consumption monitoring of an
electrical appliance that is connected in one socket.
The SM2 can by realized as a special wall socket
(with additional measuring sensors, processing
electronic and communication interfaces). A
communication of SM2 is done by WLS or via
power line. In this epoch when households are not
totally equipped by internet we decided to realize
SM2 in these two options. The SM2 can be realized
on demand as a pass-through module or as a special
wall socket with embedded SM2 functionality
additionally to the standard power functions of the
socket.
One new option of the system SMART
ENERGO for internet based monitoring (PDA,
netbook, notebook, PC, tablet) is its total
distribution. That means, data of energy
consumption are written in local memories of each
SM2s and because of the fact, that SM2s are
developed by WIFI interface, data can be observed
directly in PDAs, netbooks, notebooks, tablets, PCs
in WLS intranet of the households or enterprises,
institutions etc. System would enable a distance
monitoring by users via internet in a near future. As
shown in the Fig.1, SM1s are equipped by internet
interface (Ethernet, GPRS and other interfaces). The
world first internet based system which is equipped
by such a service is Google energy monitoring
- 331 -
system. Users have access to individual users

database with historical data of their individual
energy consumption from anywhere.
The recommended PWL (Power Line)
option for SM2 (recommended by energy
distribution firms for this temporary time period)
enables direct translation data from SM2s to the
SM1 without WLS (Wireless) interfaces in SM2s
via power lines connecting any wall socket for
energy purposes. The price for such a solution is
not cheaper in general and authors believe on a final
solution in the WLS option.
3.1 SM1 module for monitoring of the
whole energy consumption of a user
This module (Fig. 2) is intended to be
situated and installed in energy distributors of
households or in central energy distributors of
smaller energy consumers. The SM1 is only one
device for a consumer. The SM1 will monitor the
total energy consumption continuously over the all
day, week, and years. Module is supplied directly
from a power line. Module SM1 is connected to a
power lines directly in the main distributor via
current measuring transformer and enables to
measure the total current and power of individual
smaller energy consumer. The installation of SM1 is
as simple as possible to motivate consumers to
accept it. System is equipped by a simple display
for visual monitoring and indication of the function.
As mentioned above the SM1 will collect
data from SM2s and send them to the user or to the
database of a provider via internet.
SM2 module for monitoring of energy
consumption of individual appliances
The block diagram of the SM2 is described
in the Fig. 3 and Fig. 4. The module enables
generate and translate data of actual current,
voltage, power and energy of appliances, connected
in the socket.
SM22
WiFi
PWL
Internal net
Interfaces
P
Current
sensor
Voltage
sensor
f1
NO
Appliance
Fig. 4. Block diagram of SM22

The SM2 [7] can by realized as a pass
through module (with measuring sensors,
processing
electronics
and
WLS/PWL
communication interface), or as a special wall
socket (with additional measuring sensors,
processing
electronics
and
communication
interfaces to its standard power functionality).
Communication of SM2s is done by WLS or via
power line. In this epoch when households are not
totally equipped by internet we decided to realize
SM2 in these two options SM21 (Fig. 3) and SM22 (
Fig.4).
3.2 SM3 module for monitoring and
distance control
The principal block diagram of the module
SM3 [7] is in the Fig. 5.
SM3
WiFi
PWL
Internal net
Interfaces
P
SM21
WiFi
Current
sensor
PWL
Internal net
Interfaces
Voltage
sensor
f1
NO
Appliance
Breaker
Current
sensor
Voltage
sensor
f1
NO

Appliance
This diagram is very similar to the SM2s

diagram, but its function is smarter. By means of
energy breaker (just in the power socket) the
electric current into the socket is switched on and
switched off. These actions are controlled by en
embedded microcontroller. The SM3 is controlled
- 332 -
by individual user (from the central SM1 and / or

intranet / internet) and / or central control station of
the smart grid system. Communication of the SM3
in the household framework is via WLS and / or
PWL and no direct connection of SM3 towards
central control station of the smart grid and towards
direct internet based access of user is planned. The
basic higher level communication of the SMART
ENERGO system via internet is possible from the
SM1 module only.
Thanks to the module SM3 the SMART
ENERGO system realizes an important step from a
pure smart metering system towards a smart-grid
functionality. Thanks the module SM3 and its future
enhancements, an individual user could influence an
individual energy consumption according the lower
or higher tariff for energy and in the same moment
to contribute to the stabilization of the grid with
important share (ratio) of renewable energy sources.
The first generation of the SM3 assumes to switch
of electricity in the wall socket or in the connecting
module (Fig. 5). A new generation of the SM3
should take into account also appliances equipped
by an embedded individual control systems which
dont
enable
switching
by
simple
connection/disconnection of the power line in the
socket.
4. SW for data processing and control
The measured data from the SM1 and
individual SM2s will be stored in individual data
storages of SMs with defined size. Historical data
will be stored in the central database of the smart
grid/smart metering system for marketing and
management as well as for research purposes.
Another important function of the SW will
be information about the dynamical tariff of costs
for energy. The tariff is proposed to correspond
with a excess or lack of energy in the smart grids.
And the excess or lack of energy should
dynamically depend on the weather situation in area
of FV and wind farms. The SMART ENERGO
system expects to utilize precise weather forecasts
from internet sources (Nordic servers and others).
Next generation of the system will be
equipped by more sophisticate SW with advisory
functions to help users to economize their energy
consumption according adaptive and learning
algorithms.
5. Conclusion
Contribution deals with the idea of a system
for individual economical control of energy
consumption by small energy consumers. Presented
system SMART ENERGO enables saving money
for energy by small energy consumers and at the
same time to contribute to the stability of electrical
grid with a great ratio of renewable energy sources.

System is modular and enables not only monitoring
but will more and more enable an individual
control of bigger electrical appliances in households
and by smaller energy consumers in order to save
money for electrical energy. By proposed
dynamical tariff for energy, SMART ENERGO
will contribute to a better stabilization of grids with
great ratio of renewable energy sources by
following mechanism: price for energy unit will
dynamically correspond a excess or lack of energy
from renewable sources ( PV and wind farms).
There are described block diagrams of
SMART ENERGO hardware elements SM1, SM2
and SM3 and their functions specification in the
contribution, as well as the basic specification of an
users SW. The SMART ENERGO system is
conceived to utilize internet weather forecast
services to predict an excess or a lack of energy
from renewable energy sources to plan strategy for
energy consumption by households and by smaller
energy consumers.
Acknowledgements
This work was supported by Operational
Program Research and Development for
Innovations; project CVVOZE Center for
Research and Utilization of Renewable Energy
Sources no. of contract CZ.1.05/2.1.00/01.0014 and
Brno University of Technology.
References
1. http://www.tipa.eu/cz/meric-spotreby-elektrickeenergie-v-bezdratovywattmetr-esa2000/d-119040/
2. http://www.elektropaloucek.cz/elektromaterial/merice-spotrebyelektriny/meric-spotreby-elektricke-energie-dt23bezdratovy
3. http://www.conrad.cz/system-voltcraft-smartmetering-vsm-120-80a-hs.k125454
4. http://www.onsetcomp.com/products/energy_log
ging_systems
5. Zezulka F., Sajdl O., Bradac Z., Sembera J.:
Experimental Smart Grid. Proceedings of 11th
IFAC/IEEE International Conference PDES 2012,
Brno, IFAC-PapersOnLine / Elsevier, 2012. ISBN:
978-3-902823-21- 2.
Experimental Network/ Smart Grid. In Advanced
- 333 -
7. ZEZULKA, F.; SAJDL, O.; EMBERA, J.;

VESEL, I. System for measurement, prediction
and energy save. In Annals of DAAAM for 2011 &
Proceedings. Vienna AT, DAAAM International
Vienna, TU Wien. 2011. p.1387-1388. ISBN978-3901509-73-5.
CVVOZE Department
Department of Control and Instrumentation
Comunication
Brno University of Technolgy
Technicka 12
61600 Brno
CZECH REPUBLIC
E-mail:zezulka@feec.vutbr.cz
E-mail:bradac@feec.vutbr.cz
E-mail: szabo@feec.vutbr.cz
E-mail: roubal@feec.vutbr.cz
E-mail: marcon@feec.vutbr.cz

Experimental Network/ Smart Grid. In Advanced
01.03.2013 .
Internet
(GSM/cable/...)
SM1
f1
f2
f3
NO
Main distributor
with electrometer
Main
switch
SM3
SM22
Appliance
A
Appliance
B
SM21
Appliance
C
Appliance
D
Fig. 1. Complete project conception
WiFi
SM1
Voltmeter
Internal net
Interfaces
PWL
GSM/GPRS
LAN
Current
sensors
External net
Interfaces
f1
f2
f3
NO
PWL
WiFi
- 334 -

Journal V19 Book1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Journal V19 Book1

Uploaded by

Copyright:

Available Formats

ISSN 1310-8271

OF THE TECHNICAL UNIVERSITY - SOFIA

Volume 19, Book 1, 2013

Journal of the Technical University Sofia

Journal of the Technical University Sofia

2 ALBENA TANEVA, MICHAIL PETROV, GEORGI SOROKOV ... 23

16 GEORGI IVANOV, DIMITAR SPIROV . 91

OPERATING AT VARIABLE ELECTRICAL MODES

53 VANYA PLACHKOVA, TINKO EFTIMOV ... 287

Journal of the Technical University Sofia

QUANTUM ELECTRONICS AND OPTICAL

and aerosols of the atmosphere, in differential

can be obtained on the illuminated object by

perpendicularly to their bisectrises, and thus the

form to a cickle-shaped one as it is shown below.

simple non-selective flat-flat miror resonstor with

Fig.2.The actual photograph of the two-coaxial

The principle of the applicatin of the developed

with a short focal length lens L1. The focal point is

Solution with free

sequential laser pulses

Fig.5. Schematic of the canalizing-accelerating

switch between DPr and the first selective prism

optical element in laser resonator design

beam illumination. On the base of our mathematical

Fig.6. Schematic presentation of a basic principle

Fig.7. Calculated transmission IT and reflection IR

version of the Fizeau Interferometer is very

- e.g. = /F, where F is the fines factor,

Fig.8 The measured transmission of the 5 m,

Fig.9. Transmission and reflectionof

Fig. 11. Schematic of the new composite wedged

For sandwich type structure the thickness is

Fig. 12. Computed transmission of the composite

resonance of the thick wedge at high spectral

In the start position we measured the resonant

Fig.16. Principle of the device, based on IW, for

Fig.13. The experimental set-up presents the

Fig.17.Calculated dependence of the resonance

Fig.14. Visualization of the wavelength separation

We have proposed also a new utilization of the

Three systems with thick IWs (IW, IWS1

Fig.17. Selection and tuning of a single

two systems are convenient for application in the

Fig.18. Selection and continuous tuning of a single

wedge in its plane [13,12,19]. Calculation of the

The important, found by us, property of the

Fig.20. WDM multi-wavelength laser resonator

Fig.19. An important, found by us, property of the

laser beam. As we have discussed in Sec.2, the

in one of the channels. To obtain lasing also in the

characteristics theory, experiment, for this laser

Fig.24. Tuning characteristics of the twowavelength CW Yb:YAG laser. One wavelength- 2

Fig.22. Tuning curves (experimental) for the

Fig.25. Two-wavelength emission from diodepumped CW Yb:YAG crystal-oscilloscope traces.

pumped volume (application to CW Yb:YAG and

frequency of the selected mode (due to the change

Fig.26. Schematics of easily realizable twowavelength cw Yb-YAG (or diode) laser. M1 is a

Fig.27(c). Fabry-Perot spectrograms of a typical

The shown schemes are very simple and cipper. The

Fig. 27(a),(b). Schematic of continioously tunable

Fig.28. The separating layer surface for habitual

Using an RIW with mirror of R1=0.94 and R2= 0.74

4.1. Scheme with implementation of