This action might not be possible to undo. Are you sure you want to continue?
https://www.scribd.com/doc/101051819/DCLinkVoltageControl
07/20/2014
text
original
Institute of Energy Technology
Power Electronics and Drives
DC Link Voltage Control
PED101013a
Institute of Energy Technology Aalborg University PED10 Spring
2008
Pontoppidanstræde 101 9220 Aalborg
TITLE: DC Link Voltage Control
THEME: PED 10
th
Master Project
PROJECT PERIOD: 4. February 4. June 2008
PROJECT GROUP: PED101013a
Author:
Stefan Partyka
Supervisors:
Stig MunkNielsen, Associate Professor, AAU
Mads Krogsgaard, SauerDanfoss
No. of copies:3
No. of pages:94
Completed: 4. June 2008
The project purpose is to control an In
terior Permanent Magnet Synchronous
Generator in the way that DC link
voltage is kept constant within +/ 10%
limits, regardless of the applied load
and rotor speed.
The converter topology selection was
performed and the PWM rectifier topo
logy was selected for further investiga
tion. Then several methods of the field
oriented control methods for generator
were presented and at the end Constant
Torque Angle Control (CTAC) method
was selected for further studies.
The simulations of the generator sys
tem were made. The selected control
method was evaluated and verified. The
reliability of the control method is
proven also.
The selected control strategy was im
plemented in DSP then. Implementa
tion platform is also described.
However due to time constraints in the
project work, implementation is done in
period between project submission
deadline – June 4
th
and final exam
date, June 19
th
.
Preface
The presented report is the documentation of my project work in field of the Electrical Engineering,
specialization Power Electronics and Drives performed at the Aalborg University. Institute of Energy
Technology.
I would like to acknowledge Associate Professor, Stig MunkNielsen for supervising my project, valu
able advices and patience during my project work. I would like also acknowledge Mads Krogsgaard
from Sauer Danfoss for given important clues regarding my project.
I would give special thanks to Szymon Beczkowski for crucial hints during the laboratory work and
and supportive advices for writing of the report.
Finally I would like give also special thanks to Torben Matzen for help in laboratory work.
The references to the literature used in the project are in [ ] brackets.
Acronyms used in the report may be found in the list of acronyms. A list of the nomenclature used in
the report can be found in the list of nomenclature. In the back of this report, a CDROM is attached.
Aalborg University, 4 June 2008
Stefan E. Partyka
Table of Contents
Chapter 1. Preliminaries........................................................................................................................................1
1.1. Introduction................................................................................................................................................1
1.2. Problem Analysis.......................................................................................................................................1
1.2.1. Generator with sixpulse diode rectifier........................................................................................2
1.2.2. Generator with boost type chopper converter..............................................................................2
1.2.3. Generator with boost PWM rectifier..............................................................................................3
1.3. Control Method..........................................................................................................................................4
1.3.1. Constant Torque Angle Control (CTAC).......................................................................................6
1.3.2. Maximum Torque per Ampere Control.........................................................................................7
1.3.3. Unity Power Factor Control.............................................................................................................8
1.4.Comparison and Selection of the Described Control Methods............................................................9
1.4.1. Analysis .............................................................................................................................................9
1.4.2. The Method Selection.....................................................................................................................10
1.5. Scope on the Project.................................................................................................................................10
1.6. Objective....................................................................................................................................................10
1.7. Resources and Tools.................................................................................................................................11
1.7.1 Resources...........................................................................................................................................11
1.7.2. Tools..................................................................................................................................................12
1.8. Initial problem..........................................................................................................................................12
1.9. The Structure of the Project.....................................................................................................................12
Chapter 2. Modelling...........................................................................................................................................15
2.1 Modelling...................................................................................................................................................15
2.1.1. Permanent Magnet Machine Model.............................................................................................15
2.1.2. DC Link Model................................................................................................................................18
2.1.3. Converter Model.............................................................................................................................18
2.1.4. Modulation Scheme........................................................................................................................20
2.2. Summary...................................................................................................................................................24
Chapter 3. Control System..................................................................................................................................26
3.1. Field Oriented Control of the Generator...............................................................................................26
3.2. Field Oriented Control.............................................................................................................................27
3.2.1. Field Oriented Control for Base Speed.........................................................................................27
3.2.2. Flux weakening mode....................................................................................................................28
3.3. Design of the Control System.................................................................................................................32
3.3.1. Design of the q axis controllers ....................................................................................................32
3.3.2. Design of the daxis controllers.....................................................................................................47
3.4. Summary...................................................................................................................................................54
Chapter 4. Simulations Results and Implementation....................................................................................56
4.1. Simulations................................................................................................................................................56
4.1.1. Method of Testing............................................................................................................................56
4.1.2. Obtained Results..............................................................................................................................57
4.1.3. Model verification............................................................................................................................61
4.1.4. Summary...........................................................................................................................................61
4.2. Implementation........................................................................................................................................61
4.2.1. Digital Signal Processor...................................................................................................................61
4.2.2. Signal Sensing...................................................................................................................................63
4.2.3. Laboratory Setup Layout...............................................................................................................64
4.3. Summary...................................................................................................................................................66
Chapter 5. Conclusions and Future Work........................................................................................................69
5.1. Conlcusions...............................................................................................................................................69
5.2. Future Work..............................................................................................................................................69
Acronyms..............................................................................................................................................................72
Nomenclature List................................................................................................................................................75
Bibliography..........................................................................................................................................................78
Appendix A...........................................................................................................................................................81
Appendix B............................................................................................................................................................83
Appendix C...........................................................................................................................................................88
Appendix D..........................................................................................................................................................92
Chapter 1. Preliminaries
In presented chapter the introduction to the project will be performed. Then
the initial problem will be stated. Also used scientific method and use
resources used in the project are given.
1.1 Introduction
The recent development of Permanent Magnet PM machines provides new solutions for industrial
and vehicle applications. The application of permanent magnet synchronous generator increases
energy efficiency of the whole system. A traditional, electrically excited synchronous generator has
lower efficiency because of excitation losses. Moreover PM generator has more compact mechanical
structure and lower mass in compare with conventional synchronous generator [5].
In the automotive application where speed of the prime mover is variable, the constant bus voltage
irrespective of speed and load is essential. To maintain wide speed range operation the field
weakening mode is needed. Therefore IPM generator is introduced because this type of machine is
mechanically robust and provides control in field weakening region up to high speed . These IPM
features and proper design of the machine allows to achieve constant power operation of the machine
in wide speed range. [22], [16]. However the fully successful system control depends on the
appropriate control algorithm that can fully utilize the performance of the machine [16]
Introducing Energy Storage Unit (ESU) increase reliability of the system. Additionally presence of the
ESU enables to use smaller diesel engine. So looking at the presented electrical system description it is
seen that robust control design for PM generator is crucial [5].
1.2 Problem Analysis
As it was stated above the main goal of the project is to control power flow from generator to DC load
by means of Field Oriented Control (FOC) method. Bus DC voltage has to be constant under variable
load conditions.
1
Generator used in considered system is Interior mounted Permanent Magnet Synchronous Generator
(IPMSG). Output DC bus voltage is controlled by AC/DC power electronics converter installed
between generator and DC link.
The main schemes of threephase PM generator and AC/DC converters system are:
● Generator with classic sixpulse diode rectifier
● Generator with boost type chopper converter
● Generator with boost PWM rectifier
1.2.1 Generator with sixpulse diode rectifier
The diode rectifier is very common device in power electronics systems. It has simple topology and
does not need sophisticated control devices. However diode rectifier generates harmonic currents to
the ac side and also allows only one side power flow [3]. Moreover DC link voltage control in system
where diode rectifier is attached to PM generator (so excitation current is constant in linear region of
work) is only possible by means of varying generator speed. Therefore when generator is running in
variable speed mode it is problematic to keep constant DC link bus voltage in this solution.
Fig 1.1.Uncontrolled diode rectifier coupled generator (BLDC in that case) [6].
1.2.2 Generator with boost type chopper converter
To overcome disadvantages of diode rectifier many techniques have been proposed. The converter
presented below consists of diode rectifier and PWM boost chopper transistor [3]. This converter gives
maximum output DC voltage (opposite to diode rectifier where output voltage is always lower then
supply voltage) [3].
2
Figure1.2.Boost chopper rectifier topology connected to onephase grid inverter [8].
The boost chopper is responsible for :
1. Controlling the ac side current to be sinusoidal with unity power factor.
2. Controlling of the DC link capacitor voltage Vd which is always higher than than the peak ac
line voltage.
Also in this topology reversed power flow from load to generator is not possible.
However advantages of this topology are that because of the less DC link voltage ripple the PWM
characteristics are improved in compare to diode rectifier. Also control of of the DC link voltage is
possible without changing generator speed [3].
The main disadvantages of this topology besides one direction power flow are high stresses of the
components Also high peak currents carried by diodes causes EMI problems [3].
1.2.3 Generator with boost PWM rectifier
Fig.1.3.PWM Rectifier topology [7]
This topology is simple reversed PWM inverter so power from ac side is converted to DC side.
3
Therefore bidirectional power flow is allowed with this topology. The DC voltage is higher than peak
ac voltage. The control strategy of PWM rectifier provides sinusoidal ac current and unity power
factor [3].
The main disadvantages of this converter topology are: requirements for sophisticated control devices
and high switching frequency losses [1].
To compare all of these briefly described topologies, their main features were put together in table1.1:
Table1.1.Features of Threephase Rectifiers [1].
Diode rectifier Boost chopper PWMrectifier
Regulation of DCoutput voltage  + +
Low harmonic distortion of ac current —  +
Near sinusoidal current waveforms —  +
Power factor correction  + +
Bidirectional power flow   +
From table1.1 it is seen that PWM rectifier is the most promising topology for the project purpose.
Moreover the PWM rectifier allows to use sophisticated field oriented control methods oriented in
rotating dq reference frame. Also fact that PWM rectifier topology provides bidirectional power flow
is promising feature for automotive applications where starter/alternator units are demanded
(however bidirectional power flow is not considered in this project) [16]. Hence the PWM rectifier
topology it is chosen for further studies.
1.3 Control Method
To obtain high dynamic control, method based on classic Field Oriented Control for Permanent
Magnet Synchronous Motor (PMSM) was taken for consideration and will be further investigated. The
basic FOC scheme for PMSM is shown on figure 1.1:
4
Figure 1.4. Field Oriented Control scheme for PMSM [4].
The FOC method is performing realtime control of torque variations and control of angular speed
and phase currents in transient and steady state. To achieve robust control threephase variables are
transformed into dq rotating reference frame. Therefore these variables are in DC domain and hence
control of system which consists of PI controllers is more reliable.
In this project FOC was adopted to generator mode of PM machine. So speed reference was replaced
by DC link voltage reference.
In figure 1.5 real system scheme is shown. It consists of generator run by diesel engine. Generator
provides constant DC link voltage in steady state. Energy Storage Unit (ESU) is added to keep
constant DC voltage under dynamic system conditions.
Figure.1.5. Real system layout.
The real system had to be adopted for laboratory conditions. Therefore some modifications were done.
5
On the figure 1.6 The modified system is shown. As a prime mover the PMSM motor was adopted.
The variable resistor acts like load fan motors.
Figure.1.6. Modified system layout.
There are several methods of field oriented control of PMSM:
● Constant Torque Angle Control (CTAC)
● Maximum Torque per Ampere Control
● Unity Power Factor Control
1.3.1 Constant Torque Angle Control (CTAC)
In this method the torque angle is kept constant (α=π/2). The torque angle is between flux linkage
oriented in daxis and the current phasor, in the dq rotating reference frame. The π/2 angle is
maintained by forcing
i
d
=0
. The advantage of this method is that it is easy to implement because the
torque and current relationship is linearized [15], [1].
The PMSM torque is expressed:
T
e
=0.75
p
2
\
fd
i
s (1.1)
where:
i
q
=i
s
(1.2)
i
d
=0
(1.3)
6
The
i
s
current for a given torque
T
e
can be calculated:
i
s
=
T
e
0.75
p
2
\
fd
(1.4)
The air gap flux linkages can be described as:
\
m
=(\
fd
2
+L
d
2
i
d
2
)
0.5
(1.5)
Because the i
d
=0 this method is common for Surface Permanent Magnet Synchronous Machines
(SPMSM) . For IPMSM the reluctance torque cannot be utilized in this method. Hence for IPM with
high saliency ratio this is not recommended to use this method [15].
Zero daxis current also occurs in the higher air gap flux linkage and higher back emf in result [15], [1]
1.3.2 Maximum Torque per Ampere Control
In this method the stator current is minimized for given torque. Therefore copper losses are
minimized. The
i
q
and
i
d
currents are optimized by predefined functions derived from maximum
torque per stator current curve (figure 1.7).
7
Fig.1.7. Maximum torque/amp trajectory (motoring and generating mode) [3].
1.3.3 Unity Power Factor Control
Unity power factor can be achieved by maintaining following equation:
L
d
i
d
2
+L
q
i
q
2
+\
fd
i
d
=0
(1.6)
This control strategy provides maximum possible torque in the system [1].
Deriving
i
q
from (1.6) and putting into torque equation, differentiating and and equating it to zero the
expression for
I
dm
is obtained [1]
The
I
dm
is the daxis current for the maximum torque in the system,
T
em
and can be derived from the
following equation:
8
o I
dm
2
+ß I
dm
+¸=0
(1.7)
where:
o=4( L
d
L
q
) L
d
(1.8)
ß=3( L
d
L
q
) \
fd
+2 \
fd
L
d
(1.9)
¸=\
fd
2
(1.10)
Putting
I
dm
into (1.7) yields
I
qm
at the maximum torque point. Inserting these two currents into
torque equation yields maximum torque for UPFC strategy [1].
1.4 Comparison and Selection of the Described Control Methods
1.4.1 Analysis
It is important to notice that generator operates in constant power region (figure 1.8). So machine
operates near the base speed and above. It is shown in [1] that these three methods described above
most vary in below the base speed operating region. The base speed is defined as a speed at which the
backemf reaches the maximum
Fig. 1.8.Torquespeed curve [3]
possible value. Beyond the base speed flux weakening is necessary. In the [18] there are waveform
9
plots shown. In these waveforms it is seen that the important factors such as torque, power factor,
back emf and current are different for each method mainly below the base speed. Beyond the base
speed the waveforms of each methods are close to each other. Therefore for the studied generator
system purposes the leading selection criterion is the easy way to implement the control algorithm in
DSP.
1.4.2 The Method Selection
Taking into account conclusions written above the most promising method to use is CTAC. However
the machine used in the project is interior permanent magnet type. Therefore it seems that this method
is not suitable for this application. However as it was written before it depends on the machine
saliency ratio.
The saliency ratio of the machine used in the project is:
Lq
Ld
=
47.2 uH
28.7uH
=1.64
(1.11)
So the saliency ratio is not that high (some machines has saliency ratio around five)
Therefore regarding to the low saliency factor of the used machine and fact that CTAC method is easy
to implement , this method was selected for further studies.
1.5 Scope on the Project
The PWM rectifier converter topology and FOC control method were selected in this project. FOC
method which utilizes PWM rectifier will be studied during a project work.
The FOC method also will be implemented in Digital Signal Processor (DSP) by means of TI C2000
Simulink toolbox and Code Composer Studio software to control the generator in scaleddown
laboratory setup.
1.6 .Objective
The objective of this project was to study ,optimize and implement selected method which is reliable
for DC Link Voltage robust control.
The project does not deal with mechanical noise of the machine, thermal problems, EMC problems,
influence of the environmental factors on the system.
10
Magnetic saturation problems are also neglected despite fact that these problems are significant
phenomena during operating the IPM machine [16].
The reason of neglecting magnetic saturation is described in [18] in higher speeds region this effect
has not that sufficient influence on working IPM characteristics (in this project generator is running
close to base speed and above) [18].
1.7 .Resources and Tools
1.7.1 Resources
The IPM generator is a three phase prototype designed by SauerDanfoss. It is 12
pole machine, designed to be supplied by a 24V power supply and it has rated power 2,2kW.
The available BPI is also from SauerDanfoss and its target applications are
battery powered mobile vehicles. It has a rated maximum continuous
power at 7kW and its nominal supply voltage is 48V, but is only supplied by
24V due to the voltage rating of the generator. The BPI is a modified
type BPI 5435.
The modification is due to a mounted interface board, which
accepts optical PWM signals from an external source. The BPI has a built in
dead time generator to provide protection against shoot through.
At the shaft of the IPM generator an encoder is mounted, so the rotor position is obtained. The
encoder gives 10:000 Quadric Encoded Pulses (QEP) per revolution.
Th machine used in project as prime mover is a threephase brushless servo permanent magnet motor
made by Siemens. It is supplied by threephase 380 voltage and it
has rated torque Tn=12 Nm and rated speed n=4500 rpm. The motor is controlled by Siemens 6SE7022
inverter.
To implement a realtime control and measure drive system performance, the TMS320F2812 Digital
11
Signal Processor (DSP) from Texas Instruments is used. The selected DSP is designed for industrial
control applications. It has 32 bit fixed point core, 12 bit ADC converters (16 channel), and 6 PWM
outputs.
1.7.2 Tools
The model of FOC control of IPM generator was built and tested in Simulink software environment.
The Simulink model has been performed to understand and predict behavior of the controller .
The controller model then has to be discretized and the new model is built in Simulink toolbox TI
C2000 DSP (where control system tested in Simulink is used and the whole system using TI C2000
DSP blocks is built) .
The Simulink model is enclosed in the attached CDROM .
1.8 Initial problem
As it was written before the design of the proper control method for IPM generator is the goal of this
project.. Therefore the initial problem could be written as:
Design of a DC Link Voltage control strategy for IPMSG.
1.9 The Structure of the Project
This report is a documentation of performed project work. The report contains a main report,
appendices and enclosed CDROM.
In order to be able to control the IPM generator a motor model based on mathematical equations was
performed.
To reduce complexity of the model, effects such temperature variations, saturation or nonsinusoidal
winding distribution are neglected.
In chapter 1 necessary introduction to project is presented. Description of the project goals and
available resources are written. The review of the available converter topologies are described and one
was selected for further investigation. Also some of FOC methods were presented and also one was
chosen .
In chapter 2 the principles and theory background of the models used in project are presented. All
models used in simulations are briefly described.
12
In chapter 3 the selected FOC method is explained. This chapter contains also description of PI
controllers design.
In chapter 4 the simulations results, hardware setup and the implementation of the controller in DSP
is described. The FOC control method is tested in laboratory. Then performance of the method is
evaluated.
The conclusions and future work tasks are written in chapter 5.
13
14
Chapter 2. Modelling
In this chapter mathematical models of the system parts such as generator,
AC/DC converter , DC link and load have been developed and built. The
SVM principles are also described here.
2.1 Modelling
This project deals with IPMSG which provides electric power in DC voltage fan drive system. The
load fan motors are BLDC type controlled by simple sensorless scheme and are supplied by ac/dc
converters. However for simulations purposes concentrated on dynamic behavior of the whole system
under variable load conditions simulation is focused on generator model and DC link side. The load
fan motors are replaced by simple variable resistor. In the real system as prime mover the diesel
engine is used. For modeling purposes the diesel engine is replaced by Permanent Magnet
Synchronous Motor (PMSM) controlled by means of Field Oriented Control.
2.1.1 Permanent Magnet Machine Model
In the simulated system there are two permanent magnet machines used one as prime mover and one
as generator. That for motor is surface magnet mounted type and that for generator is interior magnet
mounted type. Therefore the same model could be used for both machines. However for generator
sign conversion has to be taken into account.
In the figure 2.1 Permanent Magnet Synchronous Machine is presented. It is seen that the daxis is
aligned with aaxis.
.
15
Figure 2.1 .Stator abc and dq equivalent windings [2].
The machine model voltage equations in dq rotating reference frame where daxis is fixed to the
permanent magnet rotor flux direction, are expressed as follows:
sdq m
sdq
sdq s sdq
j
dt
d
i R v λ ω
λ
+ + =
(2.1)
where:
sq sq fd sd sd sdq
i jL i L + + = λ λ (2.2)
The real and imaginary parts of the (2.1) numb and (2.2) are:
sq m
sd
sd s sd
dt
d
i R v λ ω
λ
− + = (2.3)
and
sd m
sq
sq s sq
dt
d
i R v λ ω
λ
+ + = (2.4)
fd sd sd sd
i L λ λ + =
(2.5)
and
16
sq sq sq
i L = λ
(2.6)
Putting (2.5) to (2.3) and (2.6) to (2.4) we can obtain:
sq sq m
sd
sd ds s ds
i L
dt
di
L i R v ω − + = (2.7)
v
sq
=R
s
i
sq
+L
sq
di
sq
dt
+o
m
L
sd
i
sd
+o
m
\
fd
(2.8)
where the electrical rotor speed
o
m
is related with mechanical rotor speed
o
mech
by expression:
o
m
=
p
2
o
mech (2.9)
The expression of electromagnetic torque is written by equation:
T
em
=
p
2
3
2
(\
sd
i
sq
\
sq
i
sd
)
(2.10)
Putting (2.5) and (2.6) into (2.10) the torque equation becomes:
T
em
=
p
2
3
2
(\
fd
i
sq
+( L
sd
L
sq
)i
sq
i
sd
)
(2.11)
It is seen that IPMSM contains torque component because d and q inductances are not equal.
The acceleration of the machine is expressed by mechanical equation:
1
J
d
o
mech
dt
=T
em
T
L
(2.12)
Where
T
L
is the load torque.
In the generating mode power flows in reversed direction. Therefore to keep order in signs conversion
generator model has to be adopted and its equations were changed:
17
v
sd
=R
s
i
sd
L
sd
di
sd
dt
+o
m
L
sq
i
sq
(2.13)
v
sq
=R
s
i
sq
L
sq
di
sq
dt
o
m
L
sd
i
sd
+o
m
\
fd
(2.14)
The mechanical equation for generating mode is:
1
J
d
o
mech
dt
=T
mech
T
em
(2.15)
where
T
mech
is torque produced by prime mover and
T
em
in this case is produced by electrical load of
the generator.
2.1.2 DC Link Model
DC link is modeled by relation:
CU
dc
dt
=i
dc
i
L
(2.16)
where
i
dc
is DC link current and i
L
=
U
dc
R
L
where
R
L
is load resistance.
On figure 2.2 DC link model is shown.
Figure 2.2 DC link model.
2.1.3 Converter Model
The generator is controlled by means of AC/DC converter. Therefore DC link is related with machine
via converter.
18
The converter is a voltage source converter (VSC) working as a rectifier in generation mode where
power flows to the DC link.
Opposite to motoring mode in generation mode, model of the VSC is necessary because DC side
variables and AC side variables are controlled. Because of fact that system includes mechanical
components which time constants are significantly bigger than converter switching times, the average
VSC model is used without disturbing the final result.
The generator windings are star connected. Generator is connected to VSC.
Fort he star type connection the generator voltages are calculated using following equations [14]:
V
A0
=V
AN
V
0N
=s
A
V
DC
V
0N
=
V
DC
3
(2s
A
s
B
s
C
) (2.17)
V
B0
=V
BN
V
0N
=s
B
V
DC
V
0N
=
V
DC
3
(s
A
+2s
B
s
C
) (2.18)
V
C0
=V
CN
V
0N
=s
C
V
DC
V
0N
=
V
DC
3
(s
A
s
B
+2s
C
) (2.19)
Where
V
0N
is sum of line to neutral voltages (
V
AN
,
V
BN
,
V
CN
)
divided by 3:
V
0N
=
(V
AN
+V
BN
+V
CN
)
3
=
V
DC
3
( s
A
+s
B
+s
C
) (2.20)
The relations between generator voltages are shown in figure 2.3
19
Fig. 2.3. Voltage Source Converter (VSC) attached to star connected generator [14].
The relation between DC current and AC currents is:
i
DC
=( s
A
i
A
+s
B
i
B
+s
C
i
C
)
(2.21)
The model of VSC is obtained by putting (2.17)(2.19) and (2.21) into a matrix form [14].
2.1.4 Modulation scheme
The proper modulation technique is needed to control the inverter/rectifier.
In [2] the comparison of the PWM (Pulse Width Modulation) method and SVM (Space Vector
Modulation) is performed. This comparison shows that that SVM method better utilizes the DC bus
voltage and provides 15 percent higher voltage factor (compared to PWM method) [2].
Therefore SVM method is chosen for further investigation.
The stationary αβ reference frame is used in SVM method and reference space stator voltage
vector is written as:
¯ V
ref
=
2
3
(v
aref
+av
bref
+a
2
v
cref
)
(2.22)
Where: a=
1
2
+ j
.3
2
(2.23)
20
And
v
aref
,
v
bref
and
v
cref
are reference voltages in 3 phase abc reference frame.
The figure 2.4 shows reference voltage vector projection.
Fig. 2.4. Basic voltage vectors [12].
It is seen that there are eight possible combinations of the switches position (six active states, two zero
states).
Therefore to create desired voltage vector in preset sector within the reference frame (fig 2.4), it is
essential to combine adjacent vectors of the reference voltage and to modulate them in proper time
period [15].
In the figure 2.4 reference voltage vector is located in sector 1 and adjacent state voltage
vectors are
¯
V
100
,
¯
V
110
. As it is seen in figure 2.4 the adjacent voltages are modulated by
specified time factors to obtain desired voltage vector.
In figure 2.5 the cycle for sector 1 is shown:
21
Fig. 2.5. Waveforms in sector 1 [12].
The reference voltage vector is expressed :
¯
V
ref
=
¯
V
100
t
1
T
PWM
2
+
¯
V
110
t
2
T
PWM
2
+
t
0
+t
7
T
PWM
2
⋅0
(2.24)
where (as in figure 2.5) :
T
PWM
2
=t
1
+t
2
+t
7
+t
0
(2.25)
and:
t
1
T
PWM
2
+
t
2
T
PWM
2
+
t
0
+t
7
T
PWM
2
=1
(2.26)
It can be written:
¯ V
ref
=
2
3
(V
DC
s
A
+a V
DC
⋅s
B
+a
2
V
DC
⋅s
C
)
(2.27)
Hence:
22
¯ V
100
=
2
3
(V
DC
⋅1+aV
DC
⋅0+a
2
V
DC
⋅0)=
2
3
V
DC (2.28)
¯
V
100
=
2
3
(V
DC
⋅1+aV
DC
⋅1+a
2
V
DC
⋅0 )=
2
3
V
DC
(
1
2
+ j
.3
2
) (2.29)
The equation (2.24) can be expressed in complex form:
¯
V
ref
=V
ref
(cos ¸+sin ¸)
(2.30)
Further it is possible to distinguish real and imaginary part:
ℜ
¯
V
ref
¦=V
ref
cos ¸=
2
3
V
DC
2t
1
T
PWM
+
2
3
V
DC
2t
2
T
PWM
cos
n
3
(2.31)
ℑ
¯
V
ref
¦=V
ref
sin ¸=
2
3
V
DC
2t
2
T
PWM
sin
n
3
(2.32)
From (2.31) and (2.32) the time durations for active states in sector 1 can be calculated:
t
1
=.3
V
ref
V
DC
T
PWM
2
sin (60
o
¸) (2.33)
t
2
=.3
V
ref
V
DC
T
PWM
2
sin (¸) (2.34)
From waveforms in the figure 2.5 , the duty cycles for control converter can be calculated.
s
A
=
t
1
+t
2
+t
7
T
PWM
2
(2.35)
s
A
=
t
2
+t
7
T
PWM
2
(2.36)
23
s
A
=
t
7
T
PWM
2
(2.37)
The procedure to obtain these times for each sector is similar [1].
2.2 Summary
In this chapter mathematical models of the system components were derived and presented. Also
SVM scheme for AC/DC converter was described. The system models are realized in continuous
domain and are used as connected together to simulate behavior of the whole generator system.
24
25
Chapter 3. Control System
In this chapter the selected field oriented control method is explained. Also
the flux weakening algorithm is described. In the second part of the chapter
the machine controller design is performed.
3.1 Field Oriented Control of the Generator
A constant speed IPM generator is controlled via PWM rectifier. The vector oriented control is applied
to the rectifier. The selected vector control method is presented on figure 3.1:
Fig. 3.1. IPMSG control system layout (without flux weakening control block).
The method is based on a controllers which operates in dq synchronous rotating frame. The voltage
26
controller gives i
sq
as a reference to the rectifier controller. Because CTAC method was applied the
i
sd
is equal to zero. In result the phase current is in a phase with back emf (BEMF) [9].
To eliminate the coupling terms between d and q, the decoupling network has been introduced.
3.2 Field Oriented Control
3.2.1 Field Oriented Control for Base Speed
The Field Oriented Control (FOC) method provides robust control of the PMSM. In Chapter 2 the dq
model of PMSM was presented. The reason of development dq model is that the FOC method is
oriented in dq synchronous rotating reference frame. The electrical equations for PMSM dq model
(generator convection) are as follows:
v
sd
=R
s
i
sd
L
sd
di
sd
dt
+o
m
L
sq
i
sq
(3.1)
v
sq
=R
s
i
sq
L
sq
di
sq
dt
o
m
L
sd
i
sd
+o
m
\
fd
(3.2)
Electromagnetic torque is expressed:
T
em
=
p
2
3
2
(\
fd
i
sq
+( L
sd
L
sq
)i
sq
i
sd
)
(3.3)
Putting equations (3.1), (3.2) into sdomain (replacing d/dt by Laplace's operator s) , the relations for
i
sd
and for i
sq
can be derived:
v
sd
=R
s
i
sd
L
sd
si
sd
+o
m
L
sq
i
sq
(3.4)
v
sq
=R
s
i
sq
L
sq
si
sq
o
m
L
sd
i
sd
+o
m
\
fd
(3.5)
Defining stator electrical time constants:
T
sd
=
L
sd
R
s
(3.6)
27
T
sq
=
L
sq
R
s
(3.7)
Equations (3.4) and (3.5) could be written:
o
m
L
sq
i
sq
v
sd
=R
s
(1+sT
sd
)i
sd
(3.8)
o
m
\
fd
v
sq
o
m
L
sd
i
sd
=R
s
(1+sT
sq
)i
sq
(3.9)
And currents are expressed as:
i
sd
=
(v
sd
+o
m
L
sq
i
sq
)
( R
s
(1+sT
sd
))
(3.10)
i
sq
=
(v
sq
o
m
L
sd
i
sd
+o
w
\
fd
)
( R
s
(1+sT
sq
))
(3.11)
3.2.2 Flux weakening mode
The maximum voltage which can be delivered to the machine is limited by the DC link voltage:
V
s
=2V
dc
/n (3.12)
where:
V
s
=
.
(v
d
2
+v
q
2
) (3.13)
where:
v
q
=o
e
L
d
i
d
+\
fd
(3.14)
v
d
=o
e
L
q
i
q
(3.15)
(Resistance drop is neglected)
Substituting equation (3.15), (3.14) and (3.12) into (3.13) [3]:
28
(o
e
L
d
i
d
+o
e
\
fd
)
2
+o
e
2
L
q
2
i
q
2
=
4V
d
2
n
2
(3.16)
(3.16) can be rearranged [3]:
(i
d
+
\
fd
L
d
)
2
(
2V
d
no
e
L
d
)
2
+
i
q
2
(
2V
d
no
e
L
q
)
2
=1 (3.17)
It is seen that (3.17) is ellipse equation [3]:
(i
d
C)
2
A
2
+
i
q
2
B
=1 (3.18)
where:
A=
2V
d
no
e
L
d
(3.19) is the length of the semi major axis
B=
2V
d
no
e
L
q
(3.20) is the length of the semi major axis
C=
\
fd
L
d
(3.21) is the offset of the ellipse center on the daxis [3].
Voltage limit ellipse is shown on figure 3.2.
29
Fig. 3.2. Voltage limit ellipse (motoring mode) [3].
It is seen from (3.18) and from figure 3.2. that when the speed increases the ellipse shrinks.
The current is limited by the circle (not shown on figure 3.2.) according to :
I
s
=
.
(i
d
2
+i
q
2
) (3.22)
which can be expressed as circle equation where Is is a circle radius:
I
s
2
=i
d
2
+i
q
2
(3.23)
Therefore if the demanded torque requires value of current which is placed beyond the voltage ellipse
the controller will saturate and reliability of the system drastically falls down. [19].
So in order to maintain proper control of the machine, the Is current should be kept within voltage
ellipse
Hence the field weakening strategy control was introduced to provide wide constant power region.
The implemented algorithm keeps Is current within voltage ellipse by means of reducing the angle of
30
the commanded current (figure 3.3) [16].
Fig. 3.3. Flux weakening action by means of reducing angle (generating mode) [16].
The controller scheme is shown in figure 3.4:
Fig. 3.4. Flux weakening scheme [16].
The control scheme from figure 3.4 was implemented in another set up also [16], so the max
torque/amp method was used there but in this project zero d axis strategy was used.
31
To get current angle dq currents are transformed into the polar coordinates where the angle is
multiplied by β which is :
0ß1 (3.24)
β is a result of the error between calculated and referenced modulation index (which is around 1). In
the project instead integrator ,PI controller with antiwind up was used. Current angle is changed by
multiplying it by β and the result seen on figure 3.3. is obtained.
Modulation Index Calculator calculates modulation index [16]:
M=
.
( v
q
2
+v
d
2
)
V
d
2
n
(3.25)
From (3.25) it is seen that method is dependent on changing of the machine parameters.
3.3 Design of the Control System
The scheme of the selected control system is shown on figure 3.1.
3.3.1 Design of the q axis controllers
The q axis consists of current loop (inner) and voltage loop (outer), fig. 3.5.
The error between referenced and measured DC link voltage produces reference of the q current.
The error between this reference and measured q axis current generates v
sq
. The decoupling
network component is added to the output q axis voltage. In result q axis voltage at the output of the
PI controller is smaller (fig. 3.5).
Fig. 3.5. Q axis control loop
Considering decoupling network and equation (3.9):
32
v
decq
v
sq
=R
s
(1+sT
sq
)i
sq
(3.26)
where decoupling components are:
v
decq
=o\
fd
o
m
L
sd
i
sd
(3.27)
The decoupling component (3.27) is considered as disturbance which is independent on the q axis
current. As in the [10] and [11] fed drives the current loop controller is designed at the beginning.
Design of the current controller (q axis)
The control scheme of the current loop is presented on figure 3.6
Fig. 3.6. Current (q axis) closed control loop system.
The system control scheme is based on [12], [13] includes several blocks: PI controller,
controlled plant and four delay blocks:
– Control block which causes delay because of digital calculations. The block is simple first order
transfer function with time constant T
s
=1/ f
s
, where f
s
is a sampling frequency.
– Holding block holds calculated signals until next step, therefore the delay is introduced and block
also is presented like first order transfer function where time constant is 0.5T
s
.
– The sampling element provides analog to digital conversion and is presented as first order transfer
function where time constant is 0.5T
s
.
– The PWM inverter also introduces delay into a whole system and is modeled by first order
transfer function where time constant is defined as 0.5T
pwm
where T
pwm
=1/ f
pwm
and
33
f
pwm
is a switching frequency.
The system presented on figure 3.3 can be modified by removing sample block from feedback. Finally
system with unity feedback is obtained as it is shown in figure 3.4
Fig. 3.7. Current (q axis) closed control loop system with unity feedback.
The PI controller for q current is described as:
PI
isq
=Kp1
(1+st
i1
)
( st
i1
)
(3.28)
the open loop transfer function of the system from figure 3.3 is:
G
isq
=Kp1
(1+st
i1
)
( s t
i1
)
1
( sT
s
+1)
1
(s0.5T
s
+1)
1
(s0.5T
s
+1)
1
( s0.5T
pwm
+1)
1
( R
s
( sT
sq
+1))
(3.29)
To determine magnitude of a proportional gain of the PI controller rootlocus based technique was
used. The required data for calculations:
f
s
=4000 Hz and hence: T
s
=0.25ms
f
pwm
=4000 Hz and T
pwm
=0.25ms
R
s
=9.62mD
T
sq
=
L
sq
Rs
=4.9ms
34
T determine maximum magnitude of a proportional gain of the PI controller rootlocus based
technique was used.
The rootlocus technique is performed in Matlab software environment.
The root locus of the open loop transfer function of the q axis current is shown on figure 3.8:
Fig 3.8. Root locus of the open loop transfer function of the q axis current.
In the figure 3.8 it is shown that loop begins to be unstable for proportional gain greater than 0.191.
The figure 3.8 also shows that slowest pole (closest to the (0,0) point) is caused by the plant. Hence the
zero of the PI controller is set to cancel this pole [12]:
t
i1
=T
sq
=4.9 ms (3.30)
To determine proportional gain of the PI controller Optimal Modulus (OM) design criterion is used [11]
Hence the equivalent time constant is defined. Equivalent time constant consist of sum of small time
35
constants:
T
eq
=2T
s
+0.5T
pwm
=0.625ms (3.17)
Using (3.30) and (3.31) the equation (3.29) is written as:
G
isq
=
Kp1
( st
i1
)
1
( R
s
( sT
eq
+1))
(3.32)
In the OM method the damping factor Ç=
.2
2
[11].
The generic open loop second order transfer function with damping factor Ç=
.2
2
and time
constant defined as τ has form as:
G
OM
=
1
(2t(1+s t))
(3.33)
To obtain Kpi equations (3.31) and (3.32) are compared, where
t=T
eq
and also using (3.30):
K
pi1
(T
sq
R
s
)
=
1
( 2T
eq
)
(3.34)
and:
K
pi1
=
(T
sq
R
s
)
(2T
eq
)
(3.35)
Finally:
K
pi1
=0.0378
In the PI controller the proportional and integrator gains are related:
36
Ki=
Kp
t
(3.36)
Hence:
Ki1=
K
p1
t
i1
=7.71 (3.37)
The figure 3.9 shows zeropole map of the designed control loop. From the figure 3.9 it is seen that
designed loop is stable.
Fig.3.9. Zeropole map of the q axis current control loop (closed loop transfer function).
The figure 3.10 shows bode plot of the designed controller system.
37
Figure 3.10 Bode plot of the q axis control loop.
Figure 3.11 shows that q current control loop has obtained phase margin
PM=65
o
.
The step response of the closed loop is shown in figure 3.11.
38
Fig. 3.8. Step response of the closed loop q axis current control system.
The features of the designed control loop are as follows:
1. Overshot 4.32 %
2. Rise time 1.9 ms
3. Settling time 5.27 ms.
Design of the voltage controller (q axis)
The voltage control loop is shown in the figure 3.12
From figure 3.12 it is seen that voltage loop contains current (inner) loop.
Hence from (3.32) and (3.35) the following current closed loop function can be derived [12]:
G
oisq
=
1
( 2T
eq
2
s
2
+2T
eq
s+1)
(3.38)
The current loop transformed from figure 3.7 [12] is shown on figure3.13:
39
Fig. 3.12. Voltage control loop (q axis).
Fig.3.13. Modified current loop (q axis).
To simplify current loop transfer function before applying it to voltage loop it could be assumed that
[12]:
(1+0.5T
s
s)(10.5T
s
s)=10.25T
s
2
s
2
≃11+0.5T
s
s≃
1
(10.5T
s
s)
(3.39)
Therefore:
i
sq
i
sq
ref
=
1
((10.5T
s
s)(2T
eq
2
s
2
+2T
eq
s+1))
(3.40)
Further simplification includes neglecting of the second order terms and using new defined time
constant:
T
eqv
=2T
eq
0.5T
s
=1.1ms (3.41)
Finally the approximated q axis current loop transfer function is written as:
40
i
sq
i
sq
ref
=
1
(1+T
eqv
s)
(3.42)
The voltage control loop from figure 3.9 contains previously derived current control loop (3.42) and
similarly to current loop contains also PI controller and delay blocks [11], [12] :
1. Digital calculations block which is as previously first order transfer function T
s
=1/ f
s
wher f
s
is
a sampling frequency.
2. Digital to analog conversion block which is presented like first order transfer function with time constant
0.5T
s
The DC link voltage is calculated using the relation between currents in DC link:
CdU
dc
dt
=i
dc
i
load
(3.43)
where i
load
is equal to
U
dc
R
load
(3.44)
The voltage control scheme from figure 3.9 was rearranged to obtain unity feedback:
Fig. 3.14. Voltage control loop with unity feedback (q axis).
The voltage control system has two inputs: referenced voltage and load current which is treated as
disturbance.
To overcome this problem, the fact that the whole control system and IPMSG are assumed linear, was
taken into account. Therefore the superposition method is applied and first system is considered when
disturbance is neglected while reference voltage is set at the input then in second configuration input
voltage reference is equal to zero and disturbance current is present [11] [12].
41
In the first situation where disturbance signal is set to zero only a proportional controller is enough to
fulfill system requirements [11] [12]. Hence only P component of the PI controller is studied.
The PI controller of the voltage loop:
PI
v
=Kp1
(1+s t
v
)
( st
v
)
(3.45)
When the disturbance input is set to zero the open loop transfer function is:
G
v
=Kp3
(1+st
v
)
( st
v
)
1
( sT
s
+1)
1
( s0.5T
s
+1)
1
( sT
eqv
+1)
1
( sC)
(3.46)
where C is the DC link capacitor value.
New equivalent time constant is defined:
T
sv
=1.5T
s
+T
eqv
=1.5ms (3.47)
And rearranged open loop transfer function is:
G
v
=Kp3
(1+st
v
)
( st
v
)
1
( sT
sv
+1)
1
( sC)
(3.48)
For loot locus analysis only proportional gain is considered:
G
v
=Kp3
1
( sT
sv
+1)
1
( sC)
(3.49)
The figure 3.12 shows root locus of the voltage controller open loop transfer function (3.49).
42
Fig. 3.15. Root locus of the voltage open loop transfer function.
The figure 3.15 shows that voltage loop is stable for any controller proportional gain.
The OM criterion is used to calculate the proper proportional voltage controller gain value [11]. As it was it was
before the damping factor is set up to Ç=
.2
2
. Following the procedure the (3.) is compared to generic
second order transfer function with preset damping factor (3.33) where t=T
sv
.
Kp3
1
C
=
1
( 2T
sv
)
(3.50)
Kp3=
C
( 2T
sv
)
=6.8
(3.51)
To calculate time constant of the voltage loop PI controller the input reference voltage is set to zero
and disturbance signal which is a load current in this case is present.
The analysied set up is presented in figure 3.15
43
Fig. 3.16. Voltage control loop when reference voltage is set to zero.
To maintain this analysis some precalculations have to be done:
The closed loop transfer function is:
G
ov
=
1
(2T
sv
2
s
2
+2T
sv
s+1)
(3.52)
The voltage control loop is shown in figure 3.17:
Fig. 3.17. Voltage control loop.
As it was in previous case to simplify this second order transfer function the following assumption is
put into a consideration:
(1+0.5T
s
s)(10.5T
s
s)=10.25T
s
2
s
2
≃11+0.5T
s
s≃
1
(10.5T
s
s)
(3.53)
And:
U
dc
U
dc
ref
=
1
((10.5T
s
s)( 2T
sv
2
s
2
+2T
sv
s+1))
(3.54)
The second order term is neglected and new time constant is defined:
T
esv
=2T
sv
0.5T
s
=2.9ms (3.55)
44
Therefore (3.54) becomes:
U
dc
U
dc
ref
=
1
(1+T
esv
s)
(3.56)
The figure 3.13 is redrawn to new configuration as it is presented in figure 3.18:
Fig. 3.18. Modified voltage control loop.
Transfer function which is relating dc voltage and load dc current ,based on figure 3.18 is:
U
dc
i
dis
=
(
1
sC
)
(1(
1
sC
)(
1
(1+sT
esv
)
) Kp3
(1+s t
v
)
( st
v
)
)
(3.57)
Then, putting some transformations [12]:
U
dc
i
dis
=
((
1
C
)2t
v
T
esv
s (T
esv
s+1))
(2T
esv
2
t
v
3
+2T
esv
2
t
v
2
+t
v
s+1)
(3.58)
To calculate PI controller time constant the symmetry optimum (SM) was chosen [11].
Therefore:
t
v
=4T
esv
(3.59)
This method provides optimal disturbance rejection by the controller [11]
Further (3.58) is:
45
U
dc
i
dis
=
(8T
esv
s
2
(T
esv
s+1))
(C8T
esv
3
t
v
3
+8T
esv
2
t
v
2
+T
esv
s+1)
(3.60)
The zero pole chart of the (3.60) is presented in figure 3.16
Figure 3.19 Zero pole chart of the (3.60) transfer function.
The figure 3.20 shows step response of the (3.60).
46
Fig. 3. 20 Step response of the (3.60).
When the system reach steady state the DC link voltage is not affected by disturbances.
3.3.2 Design of the daxis controllers
Design of the current controller (d axis)
The d axis controller is based on a current loop figure 3.21.
As it regarding to ZDAC method the d axis reference current is set to zero.
The error between this reference and measured d axis current generates v
sd
. The decoupling
network component is added to the output d axis voltage. In result d axis voltage at the output of the
PI controller is set to zero.
47
Fig. 3.21. Current control loop (d axis).
Considering decoupling network and equation (3.8) :
v
decd
v
sd
=R
s
(1+sT
sd
)i
sd
(3.61)
where decoupling components is:
v
decd
=o
m
L
sq
i
sq
(3.62)
The decoupling component (3.27) is considered as disturbance which is independent on the d axis
current.
The control scheme of the d current loop is presented on figure 3.22
Fig. 3.19. Current control loop (d axis).
The system presented on figure 3.22 can be modified by removing sample block from feedback.
Finally system with unity feedback is obtained as it is shown in figure 3.23
48
Fig. 3.23. Current control loop (d axis) with unity feedback.
The PI controller for q current is described as:
PI
isd
=Kp2
(1+st
i2
)
(s t
i2
)
(3.63)
the open loop transfer function of the system from figure 3.19 is:
G
isd
=Kp2
(1+st
i2
)
( st
i2
)
1
( sT
s
+1)
1
( s0.5T
s
+1)
1
( s0.5T
s
+1)
1
( s0.5T
pwm
+1)
1
( R
s
( sT
sd
+1))
(3.64)
To determine magnitude of a proportional gain of the PI controller rootlocus based technique was
used. The required data for calculations:
f
s
=4000 Hz and hence: T
s
=0.25ms
f
pwm
=4000 Hz and T
pwm
=0.25ms
R
s
=9.62mD
T
sd
=
L
sd
Rs
=3.0ms
To determine maximum magnitude of a proportional gain of the PI controller rootlocus based
technique was used.
The rootlocus technique is performed in Matlab software environment.
49
The root locus of the open loop transfer function of the q axis current is shown on figure 3.24:
Fig. 3.24. Root locus of the open loop transfer function of the d axis current.
In the figure 3.24 it is shown that loop begins to be unstable for proportional gain greater than 0.127.
The figure 3.24 also shows that slowest pole (closest to the (0,0) point) is caused by the plant. Hence
the zero of the PI controller is set to cancel this pole [12]:
t
i1
=T
sq
=3.0ms (3.65)
To determine proportional gain of the PI controller Optimal Modulus (OM) design criterion is used [11]
Hence the equivalent time constant is defined. Equivalent time constant consist of sum of small time
constants:
T
eq
=2T
s
+0.5T
pwm
=0.625ms (3.66)
Using (3.30) and (3.31) the equation (3.29) is written as:
50
G
isd
=
Kp2
( s t
i2
)
1
( R
s
( sT
eq
+1))
(3.67)
In the OM method the damping factor Ç=
.2
2
[11].
The generic open loop second order transfer function with damping factor Ç=
.2
2
and time
constant defined as τ has form as:
G
OM
=
1
(2t(1+s t))
(3.68)
To obtain Kpi equations (3.67) and (3.68) are compared, where t=T
eq
and also using (3.65):
K
pi1
(T
sd
R
s
)
=
1
( 2T
eq
)
(3.69)
and:
K
pi1
=
(T
sd
R
s
)
( 2T
eq
)
(3.70)
Finally:
K
pi1
=0.023 (3.71)
In the PI controller the proportional and integrator gains are related:
Ki=
Kp
t
(3.72)
Hence:
Ki2=
K
p2
t
i2
=7.69 (3.73)
The figure 3.25 shows zeropole map of the designed control loop. From the figure 3.25 it is seen that
designed loop is stable.
51
Fig.3.26. Zeropole map of the d axis current control loop (closed loop transfer function).
The figure 3.26. shows bode plot of the designed controller system.
52
Fig. 3.26. Bode plot of the qaxis control loop.
Figure 3.26 shows that q current control loop has obtained phase margin
PM=65
o
.
The step response of the closed loop is shown in figure 3.27.
53
Fig. 3.27. Closed loop step response (d axis).
The features of the designed control loop are as follows:
1. Overshot 4.32 %
2. Rise time 1.9 ms
3. Settling time 5.27 ms.
3.4 Summary
The selected FOC method was briefly explained in this chapter. The flux weakening algorithm was
introduced also. In the chapter the controller design was also. The stability of the controller loops was
proven.
54
55
Chapter 4. Simulations Results and Implementation
In this chapter simulation results are presented and commented. Hardware
implementation is presented and described. The laboratory setup is
presented and briefly described also.
4.1 Simulations
4.1.1 Method of Testing
The complete generator system including the machine, SVM inverter and the controller has been
investigated and tested using computer simulation. For simulation and analysis purposes the Matlab/
Simulink software was selected.
It is assumed that switches in voltage source converter (VSC) are ideal and no switching
losses are conducted. The IPM machine is modelled in dq reference frame as it was explained
in Chapter 2.
The parameters used are enclosed in Appendix.
The whole system model has been built in continuous environment.
Step load is applied to the DC link voltage to test performance of the simulated generator system.
In the evaluated system, load is presented by 9 brushless permanent magnet motors. Therefore in
simulations the nine step variable load is applied. However the most important test is under the worst
case load conditions (all 9 motors are started in the same moment) and system will be tested under
this conditions. It is assumed that generator is already running at no load mode and then load to the
DC bus is applied. The DC link voltage waveform is crucial in these tests because the bus voltage
should be regulated to remain within voltage band specified due to requirements given by norms.
The system is also tested at high speed condition to verify flux weakening mode controller and the
56
system ability to give constant power regardless input generator speed (within assumed bandwidth).
4.1.2 Obtained Results
System behavior at base speed
Fig. 4.1. DC link bus voltage response for full step load (from 0 to 2kW).
The generator is running at speed 2200 rpm and then at time 0.35 s the full load is applied. This test
simulates the worst condition situation when it is no load at DC bus and then all 9 motors are started
simulatenously.
It is seen on figure 4.1 that DC link voltage drop due to applied load is significant and could not be
accepted because such voltage drop decreases the operation quality of the load motors. However the
reason of such behavior of the system lies behind the DC link capacitance value. Therefore the value
of DC link capacitor was changed from 18.3mF to 500mF. Obtained results are shown in figure 4.2.
The change of DC link capacitor influenced voltage control loop and in result voltage controller
parameters have been changed: kp from 1.9 to 30.
57
It can be observed that the shape of the DC link voltage waveform in transient mode has changed as
well. With new parameters the acceptable margin of DC voltage drop ( around 10%) has been reached
but the transient last longer (figure 4.2.). The shape of this DC link waveform probably could be still
improved by fine tuning of the voltage controller.
Fig. 4.2. DC link bus voltage response for full step load (from 0 to 2kW after ) DC link capacitance increased.
On the figure 4.3. the phase current is plotted and on figure 4.4 phase voltage is plotted. One can
recognize voltage drop on figure 4.4. It is due to applied step load.
58
Fig. 4.3. Phase current response to load change.
Fig. 4.4. Phase voltage plot.
In the figure 4.5 it is seen that daxis current follows command current. However during the transient
stage daxis current increases significantly. This sudden current rise is caused by voltage drop in DC
bus (figure 4.1 and 4.2)
59
Fig. 4.5. Daxis current command and response for step load from 0 to 2kW.
In the figure 4.6 qaxis current command and response are presented. It is seen that response closely
fits the commanded current.
Fig. 4.6. Qaxis current command and response for step load from 0 to 2kW.
60
4.1.3 Model verification
The accuracy of the controller design methods could be evaluated here. There were three controllers
designed; DC link voltage controller, qaxis current controller and daxis current controller.
In this chapter system responses to a step load are observed. These results could be compare with
results obtained in Chapter 3. In Chapter 3 controllers were designed using common methods like
symmetry Criterion and Optimum Modulus. In Chapter 3 theoretical responses of the system based
on derived transfer function were calculated. In figures 3.8, 3.17 and 3.24. the system responses to step
signal are presented. These waveforms could be compared with equivalent responses from model
built in Simulink :4.1, 4.5, 4.6.
The plots 3.8 and 4.6 show behavior of the qaxis controller. It can be observed that these plots are
close to each other. Therefore it means that controller parameters were designed in correct way and
simplified transfer function based model has some reflection in the Simulink based system model.
On the figures 3.17 and 4.1 DC link voltage controller behavior is shown. It can be observed that these
waveforms are similar in shape, however this in figure 4.1 seems to be more stiff after transient. It
could be caused by some external factors which were not taken into account during controller design.
Though also it is important to write that voltage controller parameters were changed from values
directly calculated in chapter 3 to obtain better DC link voltage response.
The daxis current response can be observed in figures 4.5 and 3.24. It can be easily seen that Simulink
model response (figure 4.5) is not so close to the theoretically calculated response in Chapter 3.
It is seen in figure 4.5 that step load introduces quite large overshot. However this
misalingment of these two plots is mainly caused by voltage controller parameters change. In
the result different voltage response forces daxis current to changed behavior.
4.1.4 Summary
Stability of the control system was tested in continuous domain. The reliability of the control system
also was tested and it was proven that control is accurate. The model comparison was done and
verification process of the models was also performed.
4.2 Implementation
4.2.1 Digital Signal Processor
The tested control algorithm for generator was implemented in TMS320F2812 Digital Signal Processor
61
(DSP).
The features of this DSP are:
32bit, fixedpoint C28x™ DSP core
150 MIPS operation
1.8/1.9V core and 3.3V peripherals
12 bit ADC, 16 Channels
Motor control peripheralstwo event managers (EVA, EVB)
6 PWM outputs
The TMS320C2812 DSP scheme block is shown in figure 4.7.
Fig. 4.7. Block diagram of the TMS320C28x.
For building and implementing control scheme the Matlab/Simulink TI C2000 DSP toolbox was
chosen. DSP is placed on eZdsp board which is connected with computer via printer port.
62
The control system is built with help of TI C2000 DSP toolbox using specially designed blocks from
DSP library together with generic Simulink blocks. The Simulink builder coverts blocks system to C
code and then it is transferred , with help of the Code Composer Studio program, to the DSP board.
TI C2000 toolbox speeds up implementation process because it has library which contains commonly
used functions in electric drives applications such Clarke and Parke transformations or SVM blocks.
Therefore user is not supposed to writing all control scheme in C code or in Assembler, which is time
consuming process.
For floating point operations the IQmath library was used.
For storing floating point numbers the QN format is used. The native length of the word for 28xx
processor is 32 bit (processor is 32 bit type). The Q format offers compromise between the range and
the precision ( the higher precision the narrower range). The first bit from the left is used as sign
representation. After sign there are integer bits and then the fractional bits. The N number determines
how many bits are used to represent the fractional part of the whole number. Therefore number of the
bits fixed to integer part is determined as 321N [21].
The [20] reported important feature of the PWM timer. During timer underflow the analog to digital
conversion is performed and also rewriting values from shadow registers to working registers is done.
Unfortunately TI C2000 Real Time Workshop tool does not control the shadow register work in
ordered way. Therefore values are written from shadow register to working register both on
underflow and overflow of the PWM timer. It occurs on unpredictable states of PWM generation of
random reset signals.
To avoid such situation the rewriting process has to be handled only during underflow event. To set
up this process two CLD bits; COMCONA and COMCONB has to be set to 00. It is impossible to do it
in TI C2000 so such changes are made in Code Composer Studio after built code in Matlab and after
project is rebuilt again [20].
4.2.2 Signal Sensing
To gathering signals from controlled object the ADC module is used.
Analog inputs of the ADC are within range 0 to 3 V . However measured sinusoidal signals ( i.e. phase
current) are bipolar and the 1.5 V offset has to be added to the input signal to fulfill 03 V requirement
of the ADC input. The signal sensing circuit is shown in figure 4.8. This circuit is put on DSP board
interface.
63
Fig. 4.8. Signal sensing circuit.
4.2.3 Laboratory Setup Layout
The laboratory set up is shown in figure 4.9.
Signals are measured by means of measure box. Measure box is described deeply in [17] , it contains
LEM modules to measure phase currents and DC link voltage. Picture of the measure box is presented
in figure 4.10. Signals gathered by LEMs are send to the analog inputs of the eZdsp. Then signals are
scaled and after analog to digital conversion, processed in DSP. To get position information encoder is
used. The signals from encoder are sent to the eZdsp and via encoder interface further to the DSP. The
control scheme is built in TI C2000 Simulink Toolbox,then compiled in Code Compose Studio and sent
to the eZdsp via parallel port. The control algorithm is processed in DSP and according to measured
input signals and referenced values, the PWM signals are generated and sent to the AC/DC converter.
To transfer PWM signals from eZdsp to the converter electrically isolated optical fibres. Inverter used
in project which includes also DC link capacitors is shown in figure 4.11.
64
Fig. 4.9. Laboratory setup layout.
65
Fig. 4.10 Measurement box designed and constructed by authors of [17].
Fig. 4.11. AC/DC inverter used in project
4.3 . Summary
In this chapter the simulations results have been presented and described.
In the second part of this chapter the hardware used in laboratory work, was described.
The used DSP was briefly described also. Interfaces and measurements devices and other hardware
66
were presented as well.
67
68
Chapter 5. Conclusions and Future Work
5.1 Conlcusions
The presented project work concerns (FOC) of the IPMG which delivers power to the DC link bus.
Regarding this the different AC/DC converters topologies were presented and described. Also several
FOC methods comparison was done. The PWM rectifier topology was selected because it seems to be
most promising and also can handle sophisticated control strategies like FOC scheme.
Mathematical models of the system were derived and presented.
The selected FOC strategy was tested and verified in Matlab/Simulink software . Stable control loop
control method was obtained. It was shown that designed control method fulfill requirements for
robust control scheme for IPM generator.
The selected control strategy was implemented in DSP. The TI C2000 Matlab/Simulink Toolbox was
used to implement control scheme in DSP. The TI C2000 Matlab/Simulink Toolbox decreased time
needed for implementation process.
The laboratory setup was also presented. DSP used as implementation platform is also described in
the project.
5.2 Future Work
The presented control strategy seems to be a promising option. Therefore further investigating in this
topic could bring beneficial improvements to the considered system. There are several issues in the
project worth to work on:
● Sensorless operation of the generator could be introduced. It will reduce the number of sensors
used in the system. This will lead to the higher reliability and lower cost of the whole system.
● The system energy consumption could be estimated and then the energy management strategy
could be introduced. It means that energy storage devices can be used and in the result the
system reliability will increase and also power level of the diesel engine/generator system can
be reduced.
● The investigation for proper design of the generator could be considered . Together with well
69
designed robust controller it extract full performance capability of the system.
● The starting/alternator mode could be also considered because it leads to reduce number of
parts in the propulsion system
70
71
Acronyms
Acronym Description
IPMSM Interior Permanent Magnet Synchronous Machine
IPMSG Interior Permanent Magnet Synchronous Generator
SPMSM Surface Permanent Magnet Synchronous Machine
DSP Digital Signal Processor
FOC Field Oriented Control
VSI Voltage Source Inverter
ESU Energy Storage Unit
PI Proportional Integrating Controller
SO Symmetry Optimum
OM Optimum Modulus
PWM Pulse Width Modulation
SVM Space Vector Modulation
AC Alternating Current
DC Direct Current
EMI Electro Magnetic Interference
CTAC Constant Torque Angle Control
UPF Unity Power Factor
BPI Battery Powered Inverter
QEP Quadratic Encoder Pulses
ADC Analog to Digital Conversion
72
BLDC Brushless DC motor
BEMF Back Electromotive Force
73
74
Nomenclature List
Paramete
r
Description
R
s
Stator resistance
L
sd
Stator inductance in the daxis
L
sq
Stator inductance in the qaxis
i
sd
Stator daxis current
i
sq
Stator qaxis current
\
fd
Permanent magnet flux
o
m
Electrical speed
o
mech
Mechanical speed
T
em
Electromagnetic torque
T
mech
Mechanical torque
T
L
Load Torque
T
sd
Electrical time constant for daxis
T
sq
Electrical time constant for qaxis
V
s
Stator voltage
I
s
Stator current
V
dc
DC link voltage
i
dc
DC link current
\
m
Air gap flux
ß Factor which determines angle of the current in field weakening mode
75
M Modulation index
v
decq
Decoupling voltage component in qaxis
v
decd
Decoupling voltage component in daxis
T
s
Sample time
f
PWM
Switching frequency
76
77
Bibliography
[1] M. P. Kazmierkowski, R. Krishnan, and F. Blaabjerg, Control in Power Electronics. Academic Press,
2002. ISBN 0124027725. Chapter 11
[2] N. Mohan, Advanced Electric Drives. MNPERE, 2001. ISBN 971529205.
[3] B. Bose, Modern Power Electronics and AC Drives. Prentice Hall, 2001. ISBN0130167436.
[4] Application report SPRA588. Implementation of a Speed Field Oriented Control
of 3phase PMSM Motor using TMS320F240.Digital Signal Processing Solutions. September 1999.
[5] O. Koerner, J Brand, K. Rechenberg, Energy Efficient Drive System for Diesel Electric Shunting
Locomotive. EPE 2005 Dresden.
[6] M. J. Khan, M. T. Iqbal, SIMPLIFIED MODELING OF RECTIFIERCOUPLED BRUSHLESS DC
GENERATORS.4th International Conference on Electrical and Computer Engineering ICECE 2006,
1921 December 2006, Dhaka, Bangladesh.
[7] Z. Wang, L. Chang, PWM AC/DC BOOST CONVERTER SYSTEM FOR INDUCTION GENERATOR
IN VARIABLESPEED WIND TURBINES.CCECE/CCGEI, Saskatoon, May 2005.
[8] D. M. Whaley, G. Ertasgin, W. L. Soong, N. Ertugrul, J. Darbyshire,
H. Dehbonei, Ch. V. Nayar, Investigation of a LowCost GridConnected Inverter for SmallScale Wind
Turbines Based on a ConstantCurrent Source PM Generator. IEEE 2006.
[9] I. Meny, P. Enrici, J. J. Huselstein, D. Matt, Simulation and testing low power wind system. IEEE 2004.
[10] N. Mohan, First Course Power Electronics and Drives. MNPERE 2003. ISBN 0971529221.
[11] M. P. Kazmierkowski, H. Tunia. Automatic Control of ConverterFed Drives. ELSEVIER, 1994.
[12] M. Valentini, A. Raducu, T. Ofeigsson, Control of a variable speed variablepitch wind turbine
with full power converter. Aalborg University December 2007.
[13] G. F. Franklin. Feedback Control of Dynamic Systems. Prentice Hall, 2006
[14] F. Jov, A. D. Hansen, P. Soersen, F. Blaabjerg. Wind Turbine Blockset in Matlab/Simulink. General
78
Overview and Description of the Models. Aalborg University, March 2004. ISBN 8789179463.
[15] P. Perera, D. Chandana, Sensorless Control of Permanent Magnet Synchronous Motor Drives. Institute
of Energy Technology, Aalborg University 2002.
[16] J. Wai, T. M. Jahns, A New Control Technique for Achieving Wide Constant Power Speed Operation with
an Interior PM Alternator Machine. IEEE 2001.
[17] T.N. Matzen, E. Schaltz. Sensorless control of an IPMSM for hydraulic pump application. Aalborg
University, 2005. Master Project in Power Electronics and Drives.
[19] T. M. Jahns, FluxWeakening Regime Operation of an Interior PermanentMagnet Synchronous Motor
Drive. Industrial Application Society Meeting, Denver, CO, September 28October 3. IEEE 1987.
[18] M. P. Kazmierkowski, R. Krishnan, and F. Blaabjerg, Control in Power Electronics. Academic
Press, 2002. ISBN 0124027725. Chapter 7,pages 225242.
[20] R. Ciszewski, Sz. Beczkowski, Evaluation of threelevel voltage source PWM inverter topologies.
Aalborg University June 2006.
[21] IQ Math Library A Virtual Floating Point Engine, Module user's guide C28x Foundation Software
[22] P. Vas, Sensorless vector and direct torque control. Oxford Science Publications, 1998. ISBN
0198564651.
[23] eZdsp F2812 Technical Reference. Digital Development Systems. Spectrum Digital inc.
5062650001 Rev. F. September 2003.
[24] P. Krause, O. Wasynczuk, and S. Sudhoff, Analysis of Electric Machinery and Drive Systems.
J.Wiley and Sons, 2002. ISBN 047114326.
[25] Field oriented controf of 3phase acmotors, tech. rep., Texas Instruments Europe, February1998.
Literature Number: BPRA073.
79
80
Appendix A
Interior Permanent Magnet Synchronous Generator Parameters
Interior permanent magnet machines are mechanically robust and therefore suitable for high speed
applications. This type of machine has small effective air gap and in result the armature reaction effect
is strong. This feature enables deep field weakening mode operation and in result provides wide
speed range up to high speeds. Because in these machines Lq is larger than Ld, the reluctance torque
can be utilized [22].
To overcome parameters variables dependency, machine equations were transferred to dq
synchronous rotating reference frame. The parameters of the machine were measured in [17]
The parameters of the machine used in this project are summarized in table A.1:
Table A.1. Parameters of the generator used in the project
Parameter Value Description
P
n
2.2 kW Nominal power
V
dc
24 V Nominal DC link voltage
R
s
Ω 9.62 m Stator winding resistance
L
d
28.7 uH Stator inductance in daxis
L
q
47.2 uH Stator inductance in qaxis
\
fq
9.71 mWb Permanent magnet flux
p 12 Number of the poles
J 0.0182 kg*kg/m Inertia of the machine
81
82
Appendix B
Reference Frame Theory
The general reference frame transformations theory is presented here. The transformations are based on
the Clarke and Parke approaches.
The goal of these transformations is to get rid of position varying inductances. In the result all the
calculations are simpler [24].
Definition of the space vector in the abc reference frame.
The abc reference frame variables
f
a
(t )
,
f
b
(t )
,
f
c
(t )
which are time dependent can be represented
as a space vector
¯
f
abc
:
¯
f
abc
=
2
3
( f
a
(t )+af
b
(t )+a
2
f
c
(t ))
(B.1)
here:
a=e
j2n
3 (B.2)
Fig. B.1. Space vector definition in the abc reference frame [12]
83
The
f
a
(t )
,
f
b
(t )
,
f
c
(t )
variables can be expressed as:
f
a
(t )=Fcos (ot +µ)
(B.3)
f
b
(t )=Fcos(ot +µ
2 n
3
)
(B.4)
f
c
(t )=Fcos (ot +µ
4 n
3
)
(B.5)
Therefore space vector
¯
f
abc
is expressed (figure B.1) :
¯
f
abc
=Fe
j ot+µ
(B.2)
Definition of the space vector in the stationary αβ reference frame.
Clarke transformation is used to express space vector in stationary reference frame (αβ).
The real axis α is aligned with a (figure C.2).
Fig. B.2. Stationary αβ reference frame [12].
Real and imaginary components of
¯
f
oß
can be calculated from following equations:
84
f
o
=
2
3
( f
a
(t )
1
2
f
b
(t )
1
2
f
c
(t ))
(B.3)
f
ß
=
2
3
(0
.3
2
f
b
(t )
.3
2
f
c
(t )) (B.4)
Definition of the space vector in the stationary dq reference frame.
The dq frame is fixed to the rotor and the values are independent from position changes and act like DC
signals.
The Park approach provides transformation from αβ stationary reference frame to dq rotating reference
frame. The real daxis of the rotating reference frame is phase shifted by
0
dq angle from aaxis (figure
B.3).
Fig. B.3.Rotating dq reference frame [12]
Components of the
¯
f
dq
are as follows [25]:
f
d
= f
o
cos 0
dq
+ f
ß
sin 0
dq
(B.5)
f
d
= f
o
sin 0
dq
+f
ß
cos 0
dq
(B.6)
In some cases it is necessary to go directly from
¯
f
abc to
¯
f
dq . Therefore special transformation matrix
85
is used to maintain transition [11], [24].
¯
f
dq
=T
dq
⋅
¯
f
abc
(B.7)
T
dq
=
2
3

cos 0
dq
cos(0
dq

2 n
3
) cos (0
dq
+
2 n
3
)
sin 0
dq
sin (0
dq

2 n
3
) sin (0
dq
+
2 n
3
)
0.5 0.5 0.5
¦
(B.8)
The inverse transformation
¯
f
dq
to
¯
f
abc
is maintained by inverse transformation matrix:
¯
f
abc
=T
dq
1
⋅
¯
f
dq (B.9)
T
dq
1
=
2
3

cos 0
dq
sin 0
dq
1
cos (0
dq

2 n
3
) sin (0
dq

2 n
3
) 1
cos (0
dq
+
2 n
3
) sin (0
dq
+
2 n
3
) 1
¦
(B.10)
86
87
Appendix C
Hardware Setup
eZdsp F2812 Board
The eZdsp F2812 is specially designed board which allows to examine and evaluate the
TMS320C2812DSP [23].
The eZDsp provides fast code verification. C2000 Tools Code Composer driver reduces complexity of
the developed code. Optionally board has JTAG connector which acts like interface for other
debuggers to provide high level language debug [23].
Features of the eZdsp:
● TMS320F2812 Digital Signal Processor
● 150 MIPS operating speed
● 18K words onchip RAM
● 128K words onchip Flash memory
● 64K words offchip SRAM memory
● 30 MHz. Clock
● 2 Expansion Connectors (analog, I/O)
● Onboard IEEE 1149.1 JTAG Controller
● 5volt only operation with supplied AC adapter
● TI F28xx Code Composer Studio tools driver
● On board IEEE 1149.1 JTAG emulation connector
88
Functional overview of the eZdsp F2812
The block diagram of the eZdsp F2812 is presented on figure C.1:
Fig. C.1. Block Diagram of the eZdsp F2812 [23].
The JTAG interface and expansion interface are the major interfaces of the board.
The eZdsp PCB outline is shown in figure C.2. The connectors on the board are labeled.
Fig. C.2. PCB outline of the eZdsp F2812 [23].
89
The connectors description is made in table C.1:
Table C.1. Description of the eZdsp F2812 connectors [23].
The major logic blocks of the eZdsp are [23]:
● Analog interface connector
● I/O interface connector
● JTAG interface
● Parallel port JTAG controller
Power Connector
The eZdsp F2812 board is powered by 5 Volt only power supply delivered with eZdsp board.
The power is supplied via connector P6 (figure C.2).
Memory of the eZdsp F2812
The eZDsp includes the following onchip memory [23]:
● 128K x 16 Flash
● 2 blocks of 4K x 16 single access RAM (SARAM)
● 1 block of 8K x 16 SARAM
● 2 blocks of 1K x 16 SARAM
90
91
Appendix D
Simulation and Implementation Models
In this appendix model built in Matlab/Simulink and used for simulations is presented. Model is
shown in figure D.1.
The system model consist of several components. All these components are described in Chapter 2.
The selected control strategy was implemented in DSP. For implementation process Ti C2000
Simulink Toolbox was used. The figure D.2. Shows model used for implementation in DSP platform:
92
Fig. D.1.. Simulation model built in Simulink.
93
Fig. D.2. Implementation model built in TI C2000 Toolbox
94
Institute of Energy Technology 2008 Pontoppidanstræde 101
Aalborg University
PED10 Spring
9220 Aalborg
TITLE:
DC Link Voltage Control
THEME: PED 10th Master Project PROJECT PERIOD: 4. February 4. June 2008 PROJECT GROUP: PED101013a
Author:
Stefan Partyka
Supervisors: Stig MunkNielsen, Associate Professor, AAU Mads Krogsgaard, SauerDanfoss
No. of copies:3
No. of pages:94 Completed: 4. June 2008
The project purpose is to control an Interior Permanent Magnet Synchronous Generator in the way that DC link voltage is kept constant within +/ 10% limits, regardless of the applied load and rotor speed. The converter topology selection was performed and the PWM rectifier topology was selected for further investigation. Then several methods of the field oriented control methods for generator were presented and at the end Constant Torque Angle Control (CTAC) method was selected for further studies. The simulations of the generator system were made. The selected control method was evaluated and verified. The reliability of the control method is proven also. The selected control strategy was implemented in DSP then. Implementation platform is also described. However due to time constraints in the project work, implementation is done in period between project submission deadline – June 4th and final exam date, June 19th.
.
Acronyms used in the report may be found in the list of acronyms. I would like to acknowledge Associate Professor. Partyka .Preface The presented report is the documentation of my project work in field of the Electrical Engineering. I would like also acknowledge Mads Krogsgaard from Sauer Danfoss for given important clues regarding my project. A list of the nomenclature used in the report can be found in the list of nomenclature. a CDROM is attached. I would give special thanks to Szymon Beczkowski for crucial hints during the laboratory work and and supportive advices for writing of the report. valuable advices and patience during my project work. Stig MunkNielsen for supervising my project. In the back of this report. Aalborg University. Institute of Energy Technology. Finally I would like give also special thanks to Torben Matzen for help in laboratory work. The references to the literature used in the project are in [ ] brackets. 4 June 2008 Stefan E. specialization Power Electronics and Drives performed at the Aalborg University.
...11 1..................................................................3................................................................................1..........................1............................ The Method Selection..................... Objective....................................... Field Oriented Control for Base Speed......................................26 3......6......................... Control System..................3............................................................................1......12 Chapter 2.................1.............................2.................7.........................7......................................................1............................................2...............................................1.................................................................................................................................................................7................................................................................................................................................1 Resources.........1............1 1...3 1.....................2 1...............24 Chapter 3.............27 3..................... Constant Torque Angle Control (CTAC).15 2............................................3....2..............1.................................... Resources and Tools......3..........................1...................... Permanent Magnet Machine Model......................................................................................................................................................2............................................. Flux weakening mode...................................................................................................................7 1...................... Control Method.........................2....................................2................................................... Introduction........................9 1.. Generator with boost type chopper converter.........................8...................................................................................................................................................................................................................12 1. Initial problem...9...............................................................................................................4...................................1 1......27 3.............20 2..........2............................................................................15 2............... Preliminaries..........................................................................................................................11 1.............. Modelling.10 1...................................................................2................................................................................................................. Unity Power Factor Control............................1.............................2....1 1.................................. Modulation Scheme.................1 Modelling........... Problem Analysis........8 1...4.........................5........................................................................ Generator with sixpulse diode rectifier......... The Structure of the Project.....4.................................. Field Oriented Control.........................................................................2..................................................2........................................................................................1......................... DC Link Model......4.......................................2............................4 1............................26 3...........18 2..................... Generator with boost PWM rectifier................. Field Oriented Control of the Generator......12 1.........................Table of Contents Chapter 1...................3........................2........10 1.................................................. Scope on the Project..... Converter Model...... Analysis .....................9 1......3......................................... Maximum Torque per Ampere Control.................... Tools...........Comparison and Selection of the Described Control Methods.....28 ........................................................15 2..................................................................................................2 1...................................................2.........................................................................10 1..................................6 1..... Summary.....................3.............................................................18 2.......................................................................
............................69 5.....................47 3...............................92 ........ Conlcusions......................................... Conclusions and Future Work....................61 4.......56 4..........................................................69 Acronyms....................................................................................................................................1....................................3................... Summary.........3.........2...................................................3...........................................64 4...................................................................................................................................................1....61 4..............................32 3.......88 Appendix D........2.3........... Obtained Results...... Summary............................................................83 Appendix C................2........................... Design of the daxis controllers..................2.....................................56 4.............................. Design of the Control System..........54 Chapter 4.................. Laboratory Setup Layout..............................................................................................................................................61 4.32 3........3.........................................................................................................................................................................................1...........................................................................................................................................................................1.......3................................................2........................78 Appendix A........................................... Summary.............................................................81 Appendix B.................4........................................................................................66 Chapter 5.......................................................................................................................................................................... Model verification.........................72 Nomenclature List.....................................75 Bibliography...............................................................................................................63 4.................................................................................................1........................................2...........................................................1.................................2........................................ Implementation.......................................... Future Work................................................................. Simulations...................................................................................................4........ Simulations Results and Implementation........................................................................................................................... Digital Signal Processor.........................................................................................1.... Design of the q axis controllers ...........................................................3......................................69 5.........................1.............................................................................2....................................1....................................................61 4........... Signal Sensing.......................... Method of Testing..................................................................................................................................................57 4.....................................................56 4................................................
.
.
Then the initial problem will be stated. The application of permanent magnet synchronous generator increases energy efficiency of the whole system. In the automotive application where speed of the prime mover is variable. A traditional. Therefore IPM generator is introduced because this type of machine is mechanically robust and provides control in field weakening region up to high speed . To maintain wide speed range operation the field weakening mode is needed.2 Problem Analysis As it was stated above the main goal of the project is to control power flow from generator to DC load by means of Field Oriented Control (FOC) method. the constant bus voltage irrespective of speed and load is essential. So looking at the presented electrical system description it is seen that robust control design for PM generator is crucial [5]. 1. 1 .Chapter 1. These IPM features and proper design of the machine allows to achieve constant power operation of the machine in wide speed range. electrically excited synchronous generator has lower efficiency because of excitation losses. However the fully successful system control depends on the appropriate control algorithm that can fully utilize the performance of the machine [16] Introducing Energy Storage Unit (ESU) increase reliability of the system. 1. [22]. Preliminaries In presented chapter the introduction to the project will be performed. Additionally presence of the ESU enables to use smaller diesel engine. [16]. Moreover PM generator has more compact mechanical structure and lower mass in compare with conventional synchronous generator [5].1 Introduction The recent development of Permanent Magnet PM machines provides new solutions for industrial and vehicle applications. Bus DC voltage has to be constant under variable load conditions. Also used scientific method and use resources used in the project are given.
Generator used in considered system is Interior mounted Permanent Magnet Synchronous Generator (IPMSG). Output DC bus voltage is controlled by AC/DC power electronics converter installed between generator and DC link. The main schemes of threephase PM generator and AC/DC converters system are:
● ● ●
Generator with classic sixpulse diode rectifier Generator with boost type chopper converter Generator with boost PWM rectifier
1.2.1 Generator with sixpulse diode rectifier
The diode rectifier is very common device in power electronics systems. It has simple topology and does not need sophisticated control devices. However diode rectifier generates harmonic currents to the ac side and also allows only one side power flow [3]. Moreover DC link voltage control in system where diode rectifier is attached to PM generator (so excitation current is constant in linear region of work) is only possible by means of varying generator speed. Therefore when generator is running in variable speed mode it is problematic to keep constant DC link bus voltage in this solution.
Fig 1.1.Uncontrolled diode rectifier coupled generator (BLDC in that case) [6].
1.2.2 Generator with boost type chopper converter
To overcome disadvantages of diode rectifier many techniques have been proposed. The converter presented below consists of diode rectifier and PWM boost chopper transistor [3]. This converter gives maximum output DC voltage (opposite to diode rectifier where output voltage is always lower then supply voltage) [3].
2
Figure1.2.Boost chopper rectifier topology connected to onephase grid inverter [8].
The boost chopper is responsible for : 1. Controlling the ac side current to be sinusoidal with unity power factor. 2. Controlling of the DC link capacitor voltage Vd which is always higher than than the peak ac line voltage. Also in this topology reversed power flow from load to generator is not possible. However advantages of this topology are that because of the less DC link voltage ripple the PWM characteristics are improved in compare to diode rectifier. Also control of of the DC link voltage is possible without changing generator speed [3]. The main disadvantages of this topology besides one direction power flow are high stresses of the components Also high peak currents carried by diodes causes EMI problems [3].
1.2.3 Generator with boost PWM rectifier
Fig.1.3.PWM Rectifier topology [7]
This topology is simple reversed PWM inverter so power from ac side is converted to DC side.
3
Therefore bidirectional power flow is allowed with this topology. The DC voltage is higher than peak ac voltage. The control strategy of PWM rectifier provides sinusoidal ac current and unity power factor [3]. The main disadvantages of this converter topology are: requirements for sophisticated control devices and high switching frequency losses [1]. To compare all of these briefly described topologies, their main features were put together in table1.1:
Table1.1.Features of Threephase Rectifiers [1].
Diode rectifier Regulation of D output voltage C Lo harmonic distortion of ac current w N sinusoidal current waveforms ear Power factor correction Bidirectional power flow — — 
Boost chopper + + 
P M rectifier W + + + + +
From table1.1 it is seen that PWM rectifier is the most promising topology for the project purpose. Moreover the PWM rectifier allows to use sophisticated field oriented control methods oriented in rotating dq reference frame. Also fact that PWM rectifier topology provides bidirectional power flow is promising feature for automotive applications where starter/alternator units are demanded (however bidirectional power flow is not considered in this project) [16]. Hence the PWM rectifier topology it is chosen for further studies.
1.3 Control Method
To obtain high dynamic control, method based on classic Field Oriented Control for Permanent Magnet Synchronous Motor (PMSM) was taken for consideration and will be further investigated. The basic FOC scheme for PMSM is shown on figure 1.1:
4
5 real system scheme is shown.4. So speed reference was replaced by DC link voltage reference.5. The FOC method is performing realtime control of torque variations and control of angular speed and phase currents in transient and steady state. In figure 1.1. To achieve robust control threephase variables are transformed into dq rotating reference frame. Therefore these variables are in DC domain and hence control of system which consists of PI controllers is more reliable. 5 . The real system had to be adopted for laboratory conditions. It consists of generator run by diesel engine. Generator provides constant DC link voltage in steady state. Energy Storage Unit (ESU) is added to keep constant DC voltage under dynamic system conditions. Figure.Figure 1. Field Oriented Control scheme for PMSM [4]. Real system layout. Therefore some modifications were done. In this project FOC was adopted to generator mode of PM machine.
On the figure 1. Figure.75 where: i q =i s (1.1. The π/2 angle is maintained by forcing i d =0 . Modified system layout.1 Constant Torque Angle Control (CTAC) In this method the torque angle is kept constant (α=π/2). in the dq rotating reference frame.2) i d =0 (1. The PMSM torque is expressed: p i 2 fd s (1. The torque angle is between flux linkage oriented in daxis and the current phasor.1) T e =0.3. As a prime mover the PMSM motor was adopted. The variable resistor acts like load fan motors. [1].6 The modified system is shown. The advantage of this method is that it is easy to implement because the torque and current relationship is linearized [15].3) 6 . There are several methods of field oriented control of PMSM: ● ● ● Constant Torque Angle Control (CTAC) Maximum Torque per Ampere Control Unity Power Factor Control 1.6.
[1] 1.5) Because the i d =0 this method is common for Surface Permanent Magnet Synchronous Machines (SPMSM) . 7 .3. Zero daxis current also occurs in the higher air gap flux linkage and higher back emf in result [15].2 Maximum Torque per Ampere Control In this method the stator current is minimized for given torque.75 fd 2 The air gap flux linkages can be described as: m = fd L d i d 2 2 2 0. For IPMSM the reluctance torque cannot be utilized in this method. Hence for IPM with high saliency ratio this is not recommended to use this method [15]. The i q and i d currents are optimized by predefined functions derived from maximum torque per stator current curve (figure 1.5 (1. Therefore copper losses are minimized.4) 0.7).The i s current for a given torque T e can be calculated: i s= Te p (1.
3 Unity Power Factor Control Unity power factor can be achieved by maintaining following equation: L d i d L q i q fd i d =0 (1. T em and can be derived from the following equation: 8 .7.6) and putting into torque equation.1.3.Fig. differentiating and and equating it to zero the expression for I dm is obtained [1] The I dm is the daxis current for the maximum torque in the system.6) 2 2 This control strategy provides maximum possible torque in the system [1]. Maximum torque/amp trajectory (motoring and generating mode) [3]. 1. Deriving i q from (1.
4 Comparison and Selection of the Described Control Methods 1. Inserting these two currents into torque equation yields maximum torque for UPFC strategy [1].4. I 2 I dm=0 (1.10) fd Putting I dm into (1. So machine operates near the base speed and above. In the [18] there are waveform 9 .1 Analysis It is important to notice that generator operates in constant power region (figure 1.8.7) yields I qm at the maximum torque point. Beyond the base speed flux weakening is necessary. It is shown in [1] that these three methods described above most vary in below the base speed operating region. 1.8).8) =3 L d L q fd 2 fd L d (1.9) =2 (1. 1.7) dm where: =4 L d L q L d (1.Torquespeed curve [3] possible value. The base speed is defined as a speed at which the backemf reaches the maximum Fig.
However as it was written before it depends on the machine saliency ratio. EMC problems. 10 .Objective The objective of this project was to study . FOC method which utilizes PWM rectifier will be studied during a project work.2 uH = =1. power factor. influence of the environmental factors on the system.6 . The project does not deal with mechanical noise of the machine.plots shown. Therefore it seems that this method is not suitable for this application. In these waveforms it is seen that the important factors such as torque. However the machine used in the project is interior permanent magnet type. 1. Therefore for the studied generator system purposes the leading selection criterion is the easy way to implement the control algorithm in DSP.4. back emf and current are different for each method mainly below the base speed. Beyond the base speed the waveforms of each methods are close to each other.5 Scope on the Project The PWM rectifier converter topology and FOC control method were selected in this project. The FOC method also will be implemented in Digital Signal Processor (DSP) by means of TI C2000 Simulink toolbox and Code Composer Studio software to control the generator in scaleddown laboratory setup.2 The Method Selection Taking into account conclusions written above the most promising method to use is CTAC. 1. this method was selected for further studies.7uH So the saliency ratio is not that high (some machines has saliency ratio around five) Therefore regarding to the low saliency factor of the used machine and fact that CTAC method is easy to implement . 1. The saliency ratio of the machine used in the project is: Lq 47.optimize and implement selected method which is reliable for DC Link Voltage robust control.64 (1.11) Ld 28. thermal problems.
Magnetic saturation problems are also neglected despite fact that these problems are significant phenomena during operating the IPM machine [16]. To implement a realtime control and measure drive system performance. It has a rated maximum continuous power at 7kW and its nominal supply voltage is 48V. It is supplied by threephase 380 voltage and it has rated torque Tn=12 Nm and rated speed n=4500 rpm.in higher speeds region this effect has not that sufficient influence on working IPM characteristics (in this project generator is running close to base speed and above) [18]. The motor is controlled by Siemens 6SE7022 inverter. the TMS320F2812 Digital 11 . but is only supplied by 24V due to the voltage rating of the generator.Resources and Tools 1. The modification is due to a mounted interface board. It is 12 pole machine.7. At the shaft of the IPM generator an encoder is mounted.1 Resources The IPM generator is a three phase prototype designed by SauerDanfoss. 1. The reason of neglecting magnetic saturation is described in [18].7 . The available BPI is also from SauerDanfoss and its target applications are battery powered mobile vehicles. The encoder gives 10:000 Quadric Encoded Pulses (QEP) per revolution. The BPI is a modified type BPI 5435.2kW. so the rotor position is obtained. The BPI has a built in dead time generator to provide protection against shoot through. which accepts optical PWM signals from an external source. Th machine used in project as prime mover is a threephase brushless servo permanent magnet motor made by Siemens. designed to be supplied by a 24V power supply and it has rated power 2.
The Simulink model is enclosed in the attached CDROM . It has 32 bit fixed point core.Signal Processor (DSP) from Texas Instruments is used. The Simulink model has been performed to understand and predict behavior of the controller .9 The Structure of the Project This report is a documentation of performed project work. and 6 PWM outputs. 12 . appendices and enclosed CDROM. Also some of FOC methods were presented and also one was chosen . effects such temperature variations.2 Tools The model of FOC control of IPM generator was built and tested in Simulink software environment. 12 bit ADC converters (16 channel). In order to be able to control the IPM generator a motor model based on mathematical equations was performed. Therefore the initial problem could be written as: Design of a DC Link Voltage control strategy for IPMSG. saturation or nonsinusoidal winding distribution are neglected. In chapter 2 the principles and theory background of the models used in project are presented. Description of the project goals and available resources are written. The review of the available converter topologies are described and one was selected for further investigation. The selected DSP is designed for industrial control applications.8 Initial problem As it was written before the design of the proper control method for IPM generator is the goal of this project. 1. The report contains a main report. 1.. In chapter 1 necessary introduction to project is presented. All models used in simulations are briefly described.7. To reduce complexity of the model. The controller model then has to be discretized and the new model is built in Simulink toolbox TI C2000 DSP (where control system tested in Simulink is used and the whole system using TI C2000 DSP blocks is built) . 1.
The FOC control method is tested in laboratory. hardware setup and the implementation of the controller in DSP is described. Then performance of the method is evaluated.In chapter 3 the selected FOC method is explained. This chapter contains also description of PI controllers design. In chapter 4 the simulations results. 13 . The conclusions and future work tasks are written in chapter 5.
14 .
AC/DC converter . 15 . However for simulations purposes concentrated on dynamic behavior of the whole system under variable load conditions simulation is focused on generator model and DC link side.1. It is seen that the daxis is aligned with aaxis.1 Permanent Magnet Synchronous Machine is presented. In the real system as prime mover the diesel engine is used. DC link and load have been developed and built. The SVM principles are also described here. However for generator sign conversion has to be taken into account. In the figure 2.1 Permanent Magnet Machine Model In the simulated system there are two permanent magnet machines used one as prime mover and one as generator. The load fan motors are BLDC type controlled by simple sensorless scheme and are supplied by ac/dc converters. Therefore the same model could be used for both machines.Chapter 2.1 Modelling This project deals with IPMSG which provides electric power in DC voltage fan drive system. For modeling purposes the diesel engine is replaced by Permanent Magnet Synchronous Motor (PMSM) controlled by means of Field Oriented Control. Modelling In this chapter mathematical models of the system parts such as generator. The load fan motors are replaced by simple variable resistor. . 2. That for motor is surface magnet mounted type and that for generator is interior magnet mounted type. 2.
2) The real and imaginary parts of the (2.1) numb and (2.Stator abc and dq equivalent windings [2].Figure 2. are expressed as follows: v sdq = Rs i sdq + where: dλ sdq dt + jω m λ sdq (2.1) λsdq = Lsd i sd + λ fd + jLsq i sq (2.1 .2) are: v sd = Rs i sd + and dλ sd − ω m λ sq dt (2.3) v sq = Rs i sq + dλ sq dt + ω m λ sd (2.5) 16 .4) λsd = Lsd i sd + λ fd and (2. The machine model voltage equations in dq rotating reference frame where daxis is fixed to the permanent magnet rotor flux direction.
Therefore to keep order in signs conversion generator model has to be adopted and its equations were changed: 17 .9) m= The expression of electromagnetic torque is written by equation: p3 i sq i sd (2.5) to (2.6) into (2.5) and (2.12) Where T L is the load torque. In the generating mode power flows in reversed direction.10) the torque equation becomes: p3 i L sd 2 2 fd sq T em = L sq i sq i sd (2.λsq = Lsq i sq (2.4) we can obtain: v ds = R s i ds + Lsd di sd − ω m Lsq i sq dt (2.6) Putting (2.3) and (2.7) v sq=R s i sqL sq di sq m L sd i sd m fd dt (2. The acceleration of the machine is expressed by mechanical equation: 1 mech d =T em T L J dt (2.11) It is seen that IPMSM contains torque component because d and q inductances are not equal.10) 2 2 sd sq T em = Putting (2.6) to (2.8) where the electrical rotor speed m is related with mechanical rotor speed mech by expression: p 2 mech (2.
2 DC Link Model DC link is modeled by relation: CU dc =i dc i L (2. U dc where R L is load resistance.v sd = R s i sd L sd di sd m L sq i sq dt (2.16) dt where i dc is DC link current and i L = On figure 2.15) d J dt where T mech is torque produced by prime mover and T em in this case is produced by electrical load of the generator.13) di sq m L sd i sd m fd dt (2.1. 2.2 DC link model is shown.1.2 DC link model. 18 .14) v sq= R s i sq L sq The mechanical equation for generating mode is: 1 mech =T mech T em (2. 2.3 Converter Model The generator is controlled by means of AC/DC converter. Therefore DC link is related with machine via converter. RL Figure 2.
17) 3 V DC s A2s B s C (2. Fort he star type connection the generator voltages are calculated using following equations [14]: V A0 =V AN V 0N =s A V DC V 0N= V DC 2s A s B sC (2. The generator windings are star connected. V BN . model of the VSC is necessary because DC side variables and AC side variables are controlled. Generator is connected to VSC.20) 3 3 The relations between generator voltages are shown in figure 2. Opposite to motoring mode in generation mode.The converter is a voltage source converter (VSC) working as a rectifier in generation mode where power flows to the DC link.18) 3 V DC s A s B 2s C (2. the average VSC model is used without disturbing the final result. Because of fact that system includes mechanical components which time constants are significantly bigger than converter switching times.3 19 . V CN ) divided by 3: V 0N = V AN V BN V CN V DC = s As Bs C (2.19) 3 V B0=V BN V 0N=s B V DC V 0N= V C0=V CN V 0N=s C V DC V 0N= Where V 0N is sum of line to neutral voltages ( V AN .
23) 2 2 20 . The stationary αβ reference frame is used in SVM method and reference space stator voltage vector is written as: 2 2 V = v aref av bref a v cref (2.21) The model of VSC is obtained by putting (2.21) into a matrix form [14]. 2. 2.Fig.4 Modulation scheme The proper modulation technique is needed to control the inverter/rectifier.22) ref 3 Where: a= 1 3 j (2. The relation between DC current and AC currents is: i DC= s A i As B i Bs C i C (2. Voltage Source Converter (VSC) attached to star connected generator [14].17)(2. Therefore SVM method is chosen for further investigation. In [2] the comparison of the PWM (Pulse Width Modulation) method and SVM (Space Vector Modulation) is performed.1. This comparison shows that that SVM method better utilizes the DC bus voltage and provides 15 percent higher voltage factor (compared to PWM method) [2].19) and (2.3.
Therefore to create desired voltage vector in preset sector within the reference frame (fig 2. v bref and v cref are reference voltages in 3 phase abc reference frame.4. 2. Basic voltage vectors [12]. Fig.And v aref . In figure 2. The figure 2.4 shows reference voltage vector projection. two zero states).4). In the figure 2.4 the adjacent voltages are modulated by 100 110 specified time factors to obtain desired voltage vector. As it is seen in figure 2. it is essential to combine adjacent vectors of the reference voltage and to modulate them in proper time period [15].5 the cycle for sector 1 is shown: 21 . It is seen that there are eight possible combinations of the switches position (six active states.4 reference voltage vector is located in sector 1 and adjacent state voltage vectors are V . V .
Fig. The reference voltage vector is expressed : V =V ref 100 t1 T PWM 2 V 110 t2 T PWM 2 t 0t 7 ⋅0 T PWM (2.26) 2 It can be written: 2 V = V DC s Aa V DC⋅s Ba 2 V DC⋅s C (2.27) ref 3 Hence: 22 .24) 2 where (as in figure 2.25) 2 and: t1 T PWM 2 t2 T PWM 2 t 0t 7 =1 T PWM (2. Waveforms in sector 1 [12].5) : T PWM =t 1t 2t 7t 0 (2. 2.5.
28) 2 2 1 2 3 (2. sA= t 1t 2t 7 T PWM (2.36) 2 s A= 23 .31) ref 3 T PWM 3 T PWM 3 2t 2 2 ℑ[V ]=V ref sin = V DC sin (2.32) the time durations for active states in sector 1 can be calculated: t 1= 3 V ref T PWM o sin 60 (2.24) can be expressed in complex form: V =V ref cos sin (2.33) V DC 2 t 2= 3 V ref T PWM sin (2.35) 2 t 2t 7 T PWM (2.32) ref 3 T PWM 3 From (2.34) V DC 2 From waveforms in the figure 2. the duty cycles for control converter can be calculated.30) ref Further it is possible to distinguish real and imaginary part: 2t 1 2t 2 2 2 ℜ[V ]=V ref cos = V DC V DC cos (2.2 2 2 V = V DC⋅1aV DC⋅0a V DC⋅0= V 100 3 3 DC (2.29) V = V DC⋅1aV DC⋅1a V DC⋅0 = V DC j 100 3 3 2 2 The equation (2.5 .31) and (2.
37) 2 The procedure to obtain these times for each sector is similar [1].s A= t7 T PWM (2.2 Summary In this chapter mathematical models of the system components were derived and presented. 2. The system models are realized in continuous domain and are used as connected together to simulate behavior of the whole generator system. 24 . Also SVM scheme for AC/DC converter was described.
25 .
1 Field Oriented Control of the Generator A constant speed IPM generator is controlled via PWM rectifier.1. 3. 3. The vector oriented control is applied to the rectifier. The voltage 26 . IPMSG control system layout (without flux weakening control block).1: Fig. Control System In this chapter the selected field oriented control method is explained. In the second part of the chapter the machine controller design is performed. Also the flux weakening algorithm is described.Chapter 3. The selected vector control method is presented on figure 3. The method is based on a controllers which operates in dq synchronous rotating frame.
4) v sq = Rs i sq L sq si sq m L sd i sd m fd (3.controller gives i sq as a reference to the rectifier controller.6) Rs 27 .5) Defining stator electrical time constants: T sd = L sd (3.2) into sdomain (replacing d/dt by Laplace's operator s) .3) Putting equations (3.1 Field Oriented Control for Base Speed The Field Oriented Control (FOC) method provides robust control of the PMSM. The reason of development dq model is that the FOC method is oriented in dq synchronous rotating reference frame. In result the phase current is in a phase with back emf (BEMF) [9].1). Because CTAC method was applied the i sd is equal to zero.2) Electromagnetic torque is expressed: p3 i L sd 2 2 fd sq T em = L sq i sq i sd (3. 3. In Chapter 2 the dq model of PMSM was presented.1) v sq= R s i sq L sq di sq m L sd i sd m fd dt (3. (3.2. To eliminate the coupling terms between d and q. the relations for i sd and for i sq can be derived: v sd = Rs isd L sd si sd m Lsq isq (3. the decoupling network has been introduced.2 Field Oriented Control 3. The electrical equations for PMSM dq model (generator convection) are as follows: v sd = R s i sd L sd di sd m L sq i sq dt (3.
12) into (3. (3.5) could be written: m Lsq i sq v sd = Rs 1sT sd i sd (3.15).15) (Resistance drop is neglected) Substituting equation (3.13) d q where: v q = e Ld i d fd (3.14) and (3.13) [3]: 28 .10) i sq = v sq m L sd i sd w fd Rs 1sT sq (3.7) Rs Equations (3.2 Flux weakening mode The maximum voltage which can be delivered to the machine is limited by the DC link voltage: V s=2Vdc / (3.4) and (3.T sq = Lsq (3.8) m fd v sq m L sd i sd =R s 1sT sq i sq (3.9) And currents are expressed as: i sd = v sd m L sq i sq R s 1sT sd (3.2.14) v d = e Lq i q (3.12) where: V s= v 2 v 2 (3.11) 3.
2.17) It is seen that (3.16) can be rearranged [3]: 2 2 e 2 2 q q 4V2 d 2 (3.17) is ellipse equation [3]: i d C 2 A2 where: i2 q =1 (3. Voltage limit ellipse is shown on figure 3.18) B A= 2Vd e Ld 2V d e Lq fd Ld (3.16) fd 2 i d Ld 2Vd 2 e Ld iq 2V d 2 e Lq 2 =1 (3.21) is the offset of the ellipse center on the daxis [3]. 29 .20) is the length of the semi major axis C= (3.19) is the length of the semi major axis B= (3.e Ld i d e fd L i = (3.
18) and from figure 3. Voltage limit ellipse (motoring mode) [3]. The implemented algorithm keeps Is current within voltage ellipse by means of reducing the angle of 30 . [19].2. the Is current should be kept within voltage ellipse Hence the field weakening strategy control was introduced to provide wide constant power region. 3. So in order to maintain proper control of the machine. that when the speed increases the ellipse shrinks. The current is limited by the circle (not shown on figure 3.) according to : I s= i 2 i 2 (3.23) s d q Therefore if the demanded torque requires value of current which is placed beyond the voltage ellipse the controller will saturate and reliability of the system drastically falls down. It is seen from (3.2.2.22) d q which can be expressed as circle equation where Is is a circle radius: I 2=i 2 i 2 (3.Fig.
3. Fig.4: Fig. The controller scheme is shown in figure 3. Flux weakening scheme [16].3.4.the commanded current (figure 3. The control scheme from figure 3.4 was implemented in another set up also [16]. so the max torque/amp method was used there but in this project zero d axis strategy was used. 31 . 3. Flux weakening action by means of reducing angle (generating mode) [16].3) [16].
Fig. 3. In result q axis voltage at the output of the PI controller is smaller (fig. 3. The decoupling network component is added to the output q axis voltage.To get current angle dq currents are transformed into the polar coordinates where the angle is multiplied by β which is : 01 (3.5.3. fig. The error between referenced and measured DC link voltage produces reference of the q current. 3. In the project instead integrator .3. is obtained.5). The error between this reference and measured q axis current generates v sq . 3.1 Design of the q axis controllers The q axis consists of current loop (inner) and voltage loop (outer). Current angle is changed by multiplying it by β and the result seen on figure 3.5.25) it is seen that method is dependent on changing of the machine parameters.9): 32 .3 Design of the Control System The scheme of the selected control system is shown on figure 3.24) β is a result of the error between calculated and referenced modulation index (which is around 1). Modulation Index Calculator calculates modulation index [16]: v v M= 2 q 2 d Vd 2 (3. 3. Q axis control loop Considering decoupling network and equation (3.25) From (3.1.PI controller with antiwind up was used.
where f s is a sampling frequency.27) The decoupling component (3.6. The block is simple first order transfer function with time constant T s=1/ f s . The PWM inverter also introduces delay into a whole system and is modeled by first order transfer function where time constant is defined as 0.26) where decoupling components are: v decq= fd m L sd i sd (3. Design of the current controller (q axis) The control scheme of the current loop is presented on figure 3. The sampling element provides analog to digital conversion and is presented as first order transfer function where time constant is 0.5T pwm where T pwm=1/ f pwm and – – – 33 .6 Fig. 3.5T s . controlled plant and four delay blocks: – Control block which causes delay because of digital calculations.v decq v sq =R s 1sT sq i sq (3.5T s . [13] includes several blocks: PI controller. Current (q axis) closed control loop system. Holding block holds calculated signals until next step.27) is considered as disturbance which is independent on the q axis current. As in the [10] and [11] fed drives the current loop controller is designed at the beginning. therefore the delay is introduced and block also is presented like first order transfer function where time constant is 0. The system control scheme is based on [12].
28) s i1 the open loop transfer function of the system from figure 3.62 m Lsq =4. Finally system with unity feedback is obtained as it is shown in figure 3.5T s 1 s0.5T pwm 1 Rs sT sq 1 To determine magnitude of a proportional gain of the PI controller rootlocus based technique was used.7. Current (q axis) closed control loop system with unity feedback.9 ms Rs 34 T sq = . 3. The required data for calculations: f s =4000 Hz and hence: T s=0.3 is: Gisq =Kp1 1s i1 1 1 1 1 1 (3.29) s i1 sT s1 s0.25 ms f pwm =4000 Hz and T pwm=0.f pwm is a switching frequency. The PI controller for q current is described as: PI isq =Kp1 1s i1 (3.25 ms R s=9. The system presented on figure 3.4 Fig.5T s 1 s0.3 can be modified by removing sample block from feedback.
8.8 it is shown that loop begins to be unstable for proportional gain greater than 0.191. Equivalent time constant consist of sum of small time 35 . The rootlocus technique is performed in Matlab software environment.0) point) is caused by the plant.8 also shows that slowest pole (closest to the (0.T determine maximum magnitude of a proportional gain of the PI controller rootlocus based technique was used.9 ms (3.30) To determine proportional gain of the PI controller Optimal Modulus (OM) design criterion is used [11] Hence the equivalent time constant is defined. The figure 3. Root locus of the open loop transfer function of the q axis current. The root locus of the open loop transfer function of the q axis current is shown on figure 3.8: Fig 3. Hence the zero of the PI controller is set to cancel this pole [12]: i1 =T sq =4. In the figure 3.
5T pwm =0.constants: T eq =2T s0. The generic open loop second order transfer function with damping factor constant defined as τ has form as: 2 = and time 2 GOM = 1 2 1s (3.0378 In the PI controller the proportional and integrator gains are related: 36 . where =T eq and also using (3.31) the equation (3.31) and (3.35) K pi1 =0.29) is written as: Gisq = Kp1 1 s i1 R s sT eq 1 (3.625 ms (3.30): K pi1 1 = T sq R s 2Teq and: (3.32) are compared.34) K pi1 = Finally: T sq R s 2Teq (3.33) To obtain Kpi equations (3.17) Using (3.32) In the OM method the damping factor 2 = 2 [11].30) and (3.
37 .9. The figure 3.37) i1 The figure 3. Fig.9 shows zeropole map of the designed control loop. Zeropole map of the q axis current control loop (closed loop transfer function).71 (3.9 it is seen that designed loop is stable.3. From the figure 3.Ki= Kp (3.10 shows bode plot of the designed controller system.36) Hence: Ki1= K p1 =7.
11 shows that q current control loop has obtained phase margin PM =65 o .Figure 3. 38 .10 Bode plot of the q axis control loop. Figure 3.11. The step response of the closed loop is shown in figure 3.
The features of the designed control loop are as follows: 1.12 it is seen that voltage loop contains current (inner) loop.32) and (3.27 ms.Fig.12 From figure 3.8.13: 39 .32 % 2.38) The current loop transformed from figure 3. Settling time 5.35) the following current closed loop function can be derived [12]: Goisq = 1 2T s 2T eq s1 2 eq 2 (3. Step response of the closed loop q axis current control system. Design of the voltage controller (q axis) The voltage control loop is shown in the figure 3.7 [12] is shown on figure3. Rise time 1. Overshot 4. Hence from (3. 3.9 ms 3.
5T s s 2T2 s 22Teq s1 eq (3. Voltage control loop (q axis).40) Further simplification includes neglecting of the second order terms and using new defined time constant: T eqv =2Teq 0.5T s s (3.5T s s≃ s Therefore: 1 1 0.Fig. To simplify current loop transfer function before applying it to voltage loop it could be assumed that [12]: 10.41) Finally the approximated q axis current loop transfer function is written as: 40 . 3. Modified current loop (q axis).25T2 s 2≃110.5Ts s=1 0.5T s=1.13.1 ms (3.3.39) i sq i ref sq = 1 1 0.5T s s 10. Fig.12.
9 contains previously derived current control loop (3.5T s The DC link voltage is calculated using the relation between currents in DC link: CdU dc =i dc i load (3. The voltage control system has two inputs: referenced voltage and load current which is treated as disturbance. was taken into account. Digital to analog conversion block which is presented like first order transfer function with time constant 0. Voltage control loop with unity feedback (q axis).43) dt U dc Rload where i load is equal to (3. To overcome this problem.14. Digital calculations block which is as previously first order transfer function T s=1/ f a sampling frequency. 41 .42) and similarly to current loop contains also PI controller and delay blocks [11]. 3.i sq i ref sq = 1 1T eqv s (3.9 was rearranged to obtain unity feedback: Fig.42) The voltage control loop from figure 3. s wher f s is 2. [12] : 1. the fact that the whole control system and IPMSG are assumed linear. Therefore the superposition method is applied and first system is considered when disturbance is neglected while reference voltage is set at the input then in second configuration input voltage reference is equal to zero and disturbance current is present [11] [12].44) The voltage control scheme from figure 3.
New equivalent time constant is defined: T sv=1.12 shows root locus of the voltage controller open loop transfer function (3.49).In the first situation where disturbance signal is set to zero only a proportional controller is enough to fulfill system requirements [11] [12].45) When the disturbance input is set to zero the open loop transfer function is: G v =Kp3 1s v 1 1 1 1 s v sT s1 s0.49) The figure 3. 42 .48) For loot locus analysis only proportional gain is considered: G v =Kp3 1 1 sT sv 1 sC (3.5T sT eqv =1. The PI controller of the voltage loop: PI v = Kp1 1s v s v (3. Hence only P component of the PI controller is studied.46) where C is the DC link capacitor value.47) And rearranged open loop transfer function is: G v =Kp3 1s v 1 1 s v sT sv1 sC (3.5 ms (3.5T s1 sT eqv 1 sC (3.
) is compared to generic 2 second order transfer function with preset damping factor (3.51) 2Tsv Kp3 Kp3= To calculate time constant of the voltage loop PI controller the input reference voltage is set to zero and disturbance signal which is a load current in this case is present. The analysied set up is presented in figure 3.33) where =T sv .15 shows that voltage loop is stable for any controller proportional gain. Root locus of the voltage open loop transfer function.8 (3. 3.Fig. The figure 3.50) C 2Tsv C =6. As it was it was before the damping factor is set up to 2 = . 1 1 = (3. The OM criterion is used to calculate the proper proportional voltage controller gain value [11]. Following the procedure the (3.15.15 43 .
55) 44 .5T s s≃ s And: 1 1 0. 3. Voltage control loop.54) 1 0.5T s=2. Voltage control loop when reference voltage is set to zero.Fig.17.5Ts s=1 0.5T s s (3. 3.5T s s 10.25T2 s 2≃110.52) The voltage control loop is shown in figure 3.5Ts s 2T2 s 22T sv s1 sv The second order term is neglected and new time constant is defined: T esv =2Tsv 0.17: Fig. As it was in previous case to simplify this second order transfer function the following assumption is put into a consideration: 10. To maintain this analysis some precalculations have to be done: The closed loop transfer function is: Gov = 1 2T s 2T sv s1 2 2 sv (3.16.53) U dc U ref dc = 1 (3.9 ms (3.
putting some transformations [12]: 1 2 v T esv s T esv s1 U dc C = i dis 2T 2 32T2 2 v s1 esv v esv v (3. 3.Therefore (3.59) This method provides optimal disturbance rejection by the controller [11] Further (3. Transfer function which is relating dc voltage and load dc current .54) becomes: U dc U ref dc = 1 1T esv s (3.18.18: Fig.based on figure 3.13 is redrawn to new configuration as it is presented in figure 3. Modified voltage control loop.18 is: U dc = i dis 1 sC 1s v 1 1 1 Kp3 sC 1sT esv s v (3.58) To calculate PI controller time constant the symmetry optimum (SM) was chosen [11].58) is: 45 .56) The figure 3.57) Then. Therefore: v =4Tesv (3.
60).8Tesv s2 T esv s1 U dc = i dis C8T 3 38T2 2T esv s1 esv v esv v (3.20 shows step response of the (3. 46 .19 Zero pole chart of the (3.60) The zero pole chart of the (3. The figure 3.16 Figure 3.60) transfer function.60) is presented in figure 3.
Fig.60). When the system reach steady state the DC link voltage is not affected by disturbances. In result d axis voltage at the output of the PI controller is set to zero. 3.3. 3. 47 . 20 Step response of the (3. The decoupling network component is added to the output d axis voltage. The error between this reference and measured d axis current generates v sd .2 Design of the daxis controllers Design of the current controller (d axis) The d axis controller is based on a current loop figure 3.21. As it regarding to ZDAC method the d axis reference current is set to zero.
22 Fig.23 48 .62) The decoupling component (3. Finally system with unity feedback is obtained as it is shown in figure 3.19.27) is considered as disturbance which is independent on the d axis current. 3. Current control loop (d axis).21.61) where decoupling components is: v decd = m L sq i sq (3. Considering decoupling network and equation (3.22 can be modified by removing sample block from feedback. The system presented on figure 3. 3.Fig.8) : v decd v sd =R s 1sT sd i sd (3. The control scheme of the d current loop is presented on figure 3. Current control loop (d axis).
5T s1 s0. The required data for calculations: f s =4000 Hz and hence: T s=0.25 ms f pwm =4000 Hz and T pwm=0.63) the open loop transfer function of the system from figure 3.0 ms Rs T sd = To determine maximum magnitude of a proportional gain of the PI controller rootlocus based technique was used.5T s1 s0. Current control loop (d axis) with unity feedback.62 m Lsd =3. 49 . The rootlocus technique is performed in Matlab software environment. 3.5T pwm1 R s sT sd 1 To determine magnitude of a proportional gain of the PI controller rootlocus based technique was used.19 is: Gisd =Kp2 1s i2 1 1 1 1 1 (3.25 ms R s=9. The PI controller for q current is described as: PI isd =Kp2 1s i2 s i2 (3.64) s i2 sT s1 s0.23.Fig.
29) is written as: 50 . Root locus of the open loop transfer function of the d axis current. The figure 3.65) To determine proportional gain of the PI controller Optimal Modulus (OM) design criterion is used [11] Hence the equivalent time constant is defined.The root locus of the open loop transfer function of the q axis current is shown on figure 3.24.66) Using (3.24 also shows that slowest pole (closest to the (0. Hence the zero of the PI controller is set to cancel this pole [12]: i1 =T sq =3.31) the equation (3. Equivalent time constant consist of sum of small time constants: T eq =2T s0.0) point) is caused by the plant.30) and (3.625 ms (3.127.24: Fig.0 ms (3.24 it is shown that loop begins to be unstable for proportional gain greater than 0. In the figure 3.5T pwm =0. 3.
68) are compared.69) T sd R s 2Teq (3.Gisd = Kp2 1 s i2 R s sT eq 1 (3.70) K pi1 =0.65): K pi1 1 = T sd R s 2Teq and: K pi1 = Finally: (3.67) and (3.68) To obtain Kpi equations (3.73) i2 The figure 3. where =T eq and also using (3. The generic open loop second order transfer function with damping factor constant defined as τ has form as: = 2 2 and time GOM = 1 2 1s (3.25 shows zeropole map of the designed control loop.71) In the PI controller the proportional and integrator gains are related: Ki= Kp (3. 51 .69 (3.25 it is seen that designed loop is stable. From the figure 3.72) Hence: Ki2= K p2 =7.67) In the OM method the damping factor 2 = 2 [11].023 (3.
3.Fig.26. The figure 3.26. 52 . shows bode plot of the designed controller system. Zeropole map of the d axis current control loop (closed loop transfer function).
27.26. 53 . The step response of the closed loop is shown in figure 3.Fig.26 shows that q current control loop has obtained phase margin PM =65 o . 3. Bode plot of the qaxis control loop. Figure 3.
The flux weakening algorithm was introduced also. In the chapter the controller design was also. The stability of the controller loops was proven. 3. The features of the designed control loop are as follows: 1.Fig. Overshot 4.27.4 Summary The selected FOC method was briefly explained in this chapter.32 % 2.27 ms. Rise time 1.9 ms 3. 54 . 3. Closed loop step response (d axis). Settling time 5.
55 .
load is presented by 9 brushless permanent magnet motors. In the evaluated system. SVM inverter and the controller has been investigated and tested using computer simulation. The DC link voltage waveform is crucial in these tests because the bus voltage should be regulated to remain within voltage band specified due to requirements given by norms.Chapter 4. Step load is applied to the DC link voltage to test performance of the simulated generator system. The laboratory setup is presented and briefly described also.1. 4. The whole system model has been built in continuous environment. Hardware implementation is presented and described. The parameters used are enclosed in Appendix. The IPM machine is modelled in dq reference frame as it was explained in Chapter 2. Therefore in simulations the nine step variable load is applied. It is assumed that switches in voltage source converter (VSC) are ideal and no switching losses are conducted. However the most important test is under the worst case load conditions (all 9 motors are started in the same moment) and system will be tested under this conditions. It is assumed that generator is already running at no load mode and then load to the DC bus is applied.1 Method of Testing The complete generator system including the machine. For simulation and analysis purposes the Matlab/ Simulink software was selected. Simulations Results and Implementation In this chapter simulation results are presented and commented.1 Simulations 4. The system is also tested at high speed condition to verify flux weakening mode controller and the 56 .
2. It is seen on figure 4. DC link bus voltage response for full step load (from 0 to 2kW).system ability to give constant power regardless input generator speed (within assumed bandwidth). 4. However the reason of such behavior of the system lies behind the DC link capacitance value.2 Obtained Results System behavior at base speed Fig.9 to 30.3mF to 500mF. Obtained results are shown in figure 4.1 that DC link voltage drop due to applied load is significant and could not be accepted because such voltage drop decreases the operation quality of the load motors. 4. The generator is running at speed 2200 rpm and then at time 0. This test simulates the worst condition situation when it is no load at DC bus and then all 9 motors are started simulatenously.1. The change of DC link capacitor influenced voltage control loop and in result voltage controller parameters have been changed: kp from 1. 57 .35 s the full load is applied.1. Therefore the value of DC link capacitor was changed from 18.
58 .). DC link bus voltage response for full step load (from 0 to 2kW after ) DC link capacitance increased. It is due to applied step load. With new parameters the acceptable margin of DC voltage drop ( around 10%) has been reached but the transient last longer (figure 4.It can be observed that the shape of the DC link voltage waveform in transient mode has changed as well. Fig. One can recognize voltage drop on figure 4. 4.3. On the figure 4.4 phase voltage is plotted. the phase current is plotted and on figure 4.4.2. The shape of this DC link waveform probably could be still improved by fine tuning of the voltage controller.2.
Phase current response to load change.1 and 4. However during the transient stage daxis current increases significantly.4.Fig. Fig. This sudden current rise is caused by voltage drop in DC bus (figure 4.3. 4. Phase voltage plot. 4.2) 59 . In the figure 4.5 it is seen that daxis current follows command current.
6.Fig. 4. Fig. Daxis current command and response for step load from 0 to 2kW. Qaxis current command and response for step load from 0 to 2kW. 60 . In the figure 4.6 qaxis current command and response are presented. It is seen that response closely fits the commanded current. 4.5.
However this misalingment of these two plots is mainly caused by voltage controller parameters change.8 and 4. Though also it is important to write that voltage controller parameters were changed from values directly calculated in chapter 3 to obtain better DC link voltage response. On the figures 3.8.5) is not so close to the theoretically calculated response in Chapter 3.17 and 4.1. 4.5 and 3. In Chapter 3 theoretical responses of the system based on derived transfer function were calculated.5 that step load introduces quite large overshot.17 and 3. It is seen in figure 4. The model comparison was done and verification process of the models was also performed. In the result different voltage response forces daxis current to changed behavior. In Chapter 3 controllers were designed using common methods like symmetry Criterion and Optimum Modulus. There were three controllers designed. These waveforms could be compared with equivalent responses from model built in Simulink :4. 4.2 Implementation 4.5. 4. however this in figure 4. qaxis current controller and daxis current controller. DC link voltage controller. Therefore it means that controller parameters were designed in correct way and simplified transfer function based model has some reflection in the Simulink based system model.24. The daxis current response can be observed in figures 4.6 show behavior of the qaxis controller. 4.1 Digital Signal Processor The tested control algorithm for generator was implemented in TMS320F2812 Digital Signal Processor 61 . These results could be compare with results obtained in Chapter 3. In figures 3.1 DC link voltage controller behavior is shown.4 Summary Stability of the control system was tested in continuous domain. The reliability of the control system also was tested and it was proven that control is accurate. It can be observed that these plots are close to each other.1. It could be caused by some external factors which were not taken into account during controller design.3 Model verification The accuracy of the controller design methods could be evaluated here.4.1 seems to be more stiff after transient. 3.1.2. The plots 3. It can be observed that these waveforms are similar in shape.24.6. In this chapter system responses to a step load are observed. It can be easily seen that Simulink model response (figure 4. the system responses to step signal are presented.
The features of this DSP are: 32bit. fixedpoint C28x™ DSP core 150 MIPS operation 1.9V core and 3. Block diagram of the TMS320C28x. DSP is placed on eZdsp board which is connected with computer via printer port.7. 16 Channels Motor control peripheralstwo event managers (EVA. 62 . EVB) 6 PWM outputs The TMS320C2812 DSP scheme block is shown in figure 4.8/1. 4.3V peripherals 12 bit ADC.(DSP).7. Fig. For building and implementing control scheme the Matlab/Simulink TI C2000 DSP toolbox was chosen.
The first bit from the left is used as sign representation. phase current) are bipolar and the 1. It is impossible to do it in TI C2000 so such changes are made in Code Composer Studio after built code in Matlab and after project is rebuilt again [20]. The signal sensing circuit is shown in figure 4. To avoid such situation the rewriting process has to be handled only during underflow event. 4. Therefore number of the bits fixed to integer part is determined as 321N [21].8. During timer underflow the analog to digital conversion is performed and also rewriting values from shadow registers to working registers is done. TI C2000 toolbox speeds up implementation process because it has library which contains commonly used functions in electric drives applications such Clarke and Parke transformations or SVM blocks. For floating point operations the IQmath library was used. However measured sinusoidal signals ( i.5 V offset has to be added to the input signal to fulfill 03 V requirement of the ADC input.e. The Simulink builder coverts blocks system to C code and then it is transferred . The N number determines how many bits are used to represent the fractional part of the whole number. This circuit is put on DSP board interface. Therefore values are written from shadow register to working register both on underflow and overflow of the PWM timer. For storing floating point numbers the QN format is used. 63 . After sign there are integer bits and then the fractional bits. which is time consuming process. Analog inputs of the ADC are within range 0 to 3 V .2 Signal Sensing To gathering signals from controlled object the ADC module is used. to the DSP board. Unfortunately TI C2000 Real Time Workshop tool does not control the shadow register work in ordered way. The native length of the word for 28xx processor is 32 bit (processor is 32 bit type). The [20] reported important feature of the PWM timer.The control system is built with help of TI C2000 DSP toolbox using specially designed blocks from DSP library together with generic Simulink blocks. The Q format offers compromise between the range and the precision ( the higher precision the narrower range). COMCONA and COMCONB has to be set to 00. Therefore user is not supposed to writing all control scheme in C code or in Assembler.2.generation of random reset signals. To set up this process two CLD bits. with help of the Code Composer Studio program. It occurs on unpredictable states of PWM.
To transfer PWM signals from eZdsp to the converter electrically isolated optical fibres. it contains LEM modules to measure phase currents and DC link voltage. Signals are measured by means of measure box.8. processed in DSP. Picture of the measure box is presented in figure 4. Signals gathered by LEMs are send to the analog inputs of the eZdsp. Inverter used in project which includes also DC link capacitors is shown in figure 4. Measure box is described deeply in [17] . Signal sensing circuit. 4.10. Then signals are scaled and after analog to digital conversion. the PWM signals are generated and sent to the AC/DC converter. The control scheme is built in TI C2000 Simulink Toolbox.9. 64 .11.Fig.then compiled in Code Compose Studio and sent to the eZdsp via parallel port.2. To get position information encoder is used. The signals from encoder are sent to the eZdsp and via encoder interface further to the DSP. The control algorithm is processed in DSP and according to measured input signals and referenced values. 4.3 Laboratory Setup Layout The laboratory set up is shown in figure 4.
Laboratory setup layout. 4. 65 .9.Fig.
11. AC/DC inverter used in project 4. Fig. The used DSP was briefly described also. Summary In this chapter the simulations results have been presented and described.10 Measurement box designed and constructed by authors of [17].3 . was described. In the second part of this chapter the hardware used in laboratory work.Fig. 4. 4. Interfaces and measurements devices and other hardware 66 .
were presented as well. 67 .
68
Chapter 5. Conclusions and Future Work
5.1 Conlcusions
The presented project work concerns (FOC) of the IPMG which delivers power to the DC link bus. Regarding this the different AC/DC converters topologies were presented and described. Also several FOC methods comparison was done. The PWM rectifier topology was selected because it seems to be most promising and also can handle sophisticated control strategies like FOC scheme. Mathematical models of the system were derived and presented. The selected FOC strategy was tested and verified in Matlab/Simulink software . Stable control loop control method was obtained. It was shown that designed control method fulfill requirements for robust control scheme for IPM generator. The selected control strategy was implemented in DSP. The TI C2000 Matlab/Simulink Toolbox was used to implement control scheme in DSP. The TI C2000 Matlab/Simulink Toolbox decreased time needed for implementation process. The laboratory setup was also presented. DSP used as implementation platform is also described in the project.
5.2 Future Work
The presented control strategy seems to be a promising option. Therefore further investigating in this topic could bring beneficial improvements to the considered system. There are several issues in the project worth to work on:
●
Sensorless operation of the generator could be introduced. It will reduce the number of sensors used in the system. This will lead to the higher reliability and lower cost of the whole system. The system energy consumption could be estimated and then the energy management strategy could be introduced. It means that energy storage devices can be used and in the result the system reliability will increase and also power level of the diesel engine/generator system can be reduced. The investigation for proper design of the generator could be considered . Together with well
●
●
69
designed robust controller it extract full performance capability of the system.
●
The starting/alternator mode could be also considered because it leads to reduce number of parts in the propulsion system
70
71 .
Acronyms Acronym IPMSM IPMSG SPMSM DSP FOC VSI ESU PI SO OM PWM SVM AC DC EMI CTAC UPF BPI QEP ADC Description Interior Permanent Magnet Synchronous Machine Interior Permanent Magnet Synchronous Generator Surface Permanent Magnet Synchronous Machine Digital Signal Processor Field Oriented Control Voltage Source Inverter Energy Storage Unit Proportional Integrating Controller Symmetry Optimum Optimum Modulus Pulse Width Modulation Space Vector Modulation Alternating Current Direct Current Electro Magnetic Interference Constant Torque Angle Control Unity Power Factor Battery Powered Inverter Quadratic Encoder Pulses Analog to Digital Conversion 72 .
BLDC BEMF Brushless DC motor Back Electromotive Force 73 .
74 .
Nomenclature List Paramete Description r Rs L sd Stator resistance Stator inductance in the daxis Stator inductance in the qaxis Stator daxis current Stator qaxis current Permanent magnet flux Electrical speed Mechanical speed Electromagnetic torque Mechanical torque Load Torque Electrical time constant for daxis Electrical time constant for qaxis Stator voltage Stator current DC link voltage DC link current Air gap flux Factor which determines angle of the current in field weakening mode L sq i sd i sq fd m mech T em T mech TL T sd T sq Vs Is V dc i dc m 75 .
M Modulation index Decoupling voltage component in qaxis Decoupling voltage component in daxis Sample time Switching frequency v decq v decd Ts f PWM 76 .
77 .
Khan. [6] M. F.4th International Conference on Electrical and Computer Engineering ICECE 2006. P. 2002. [4] Application report SPRA588. ELSEVIER. P. V. Nayar. [7] Z. IEEE 2004. Chang. M. Rechenberg. 1994. Implementation of a Speed Field Oriented Control of 3phase PMSM Motor using TMS320F240. Hansen.Digital Signal Processing Solutions. T. Dhaka. P. K. Energy Efficient Drive System for Diesel Electric Shunting Locomotive. ISBN. Wind Turbine Blockset in Matlab/Simulink. G. May 2005.CCECE/CCGEI. Chapter 11 [2] N. Ch. N. W. Meny. [3] B. ISBN 0124027725. Feedback Control of Dynamic Systems. Blaabjerg. J. Ofeigsson. Academic Press. Prentice Hall. and F. [10] N. 1921 December 2006. H. [9] I. [11] M. Blaabjerg. A. Raducu. ISBN0130167436. Soersen. General 78 . P. Tunia. Ertasgin. Kazmierkowski. D. A. Jov. 2006 [14] F. SIMPLIFIED MODELING OF RECTIFIERCOUPLED BRUSHLESS DC GENERATORS. Control in Power Electronics. J. IEEE 2006. September 1999. PWM AC/DC BOOST CONVERTER SYSTEM FOR INDUCTION GENERATOR IN VARIABLESPEED WIND TURBINES. Bose.0971529221. Prentice Hall. Control of a variable speed variablepitch wind turbine with full power converter. Bangladesh. 2001. Investigation of a LowCost GridConnected Inverter for SmallScale Wind Turbines Based on a ConstantCurrent Source PM Generator. Advanced Electric Drives. Enrici. Modern Power Electronics and AC Drives. L. Kazmierkowski. Darbyshire. Dehbonei. Koerner. 2001. EPE 2005 Dresden. [8] D. Valentini. Wang. L. Soong. MNPERE. Whaley. [13] G. Aalborg University December 2007. Saskatoon. ISBN 971529205. T. Matt. Franklin. F. Huselstein. Krishnan. J. D. Mohan. Automatic Control of ConverterFed Drives. R. [5] O. Ertugrul. J Brand. Simulation and testing low power wind system. First Course Power Electronics and Drives. M. J. [12] M. MNPERE 2003. Mohan.Bibliography [1] M. Iqbal. H.
Sensorless Control of Permanent Magnet Synchronous Motor Drives. Jahns.pages 225242. Aalborg University 2002. [23] eZdsp F2812 Technical Reference. 1998. 2005. [20] R. [21] IQ Math Library A Virtual Floating Point Engine. Master Project in Power Electronics and Drives. Sudhoff. [17] T. Oxford Science Publications. F. 5062650001 Rev. Krishnan. Vas. Sz. 79 . P. E. Krause. ISBN 0198564651. O. Control in Power Electronics. IEEE 1987. IEEE 2001. Denver. M. Sensorless control of an IPMSM for hydraulic pump application. Beczkowski. Ciszewski. Jahns. Module user's guide C28x Foundation Software [22] P.N. Chapter 7. and F. FluxWeakening Regime Operation of an Interior PermanentMagnet Synchronous Motor Drive. Schaltz. ISBN 047114326. [25] Field oriented controf of 3phase acmotors. D.Overview and Description of the Models. February1998. 2002. Matzen. Aalborg University. Literature Number: BPRA073. ISBN 0124027725. T. Perera. Kazmierkowski. M. J. September 2003. Wasynczuk.. Evaluation of threelevel voltage source PWM inverter topologies. tech. and S. Spectrum Digital inc. Blaabjerg. Wai. A New Control Technique for Achieving Wide Constant Power Speed Operation with an Interior PM Alternator Machine. Texas Instruments Europe. Digital Development Systems. [19] T. Chandana. R. Academic Press. Analysis of Electric Machinery and Drive Systems. 2002. [15] P. Institute of Energy Technology. September 28October 3.Wiley and Sons. ISBN 8789179463. [24] P. [16] J. Sensorless vector and direct torque control. Aalborg University June 2006. CO. March 2004. Industrial Application Society Meeting. Aalborg University. rep. [18] M.
80 .
62 m 28. Because in these machines Lq is larger than Ld. Parameters of the generator used in the project Parameter Value Pn V dc Rs Ld Lq fq p 2. This type of machine has small effective air gap and in result the armature reaction effect is strong.71 mWb 12 0.2 uH 9.1: Table A. To overcome parameters variables dependency.7 uH 47.1. The parameters of the machine were measured in [17] The parameters of the machine used in this project are summarized in table A.2 kW 24 V 9. machine equations were transferred to dq synchronous rotating reference frame.Appendix A Interior Permanent Magnet Synchronous Generator Parameters Interior permanent magnet machines are mechanically robust and therefore suitable for high speed applications. the reluctance torque can be utilized [22]. This feature enables deep field weakening mode operation and in result provides wide speed range up to high speeds.0182 kg*kg/m Description Nominal power Nominal DC link voltage Stator winding resistance Stator inductance in daxis Stator inductance in qaxis Permanent magnet flux Number of the poles Inertia of the machine J 81 .
82 .
B.2) Fig. In the result all the calculations are simpler [24].1) abc 3 here: j2 3 a=e (B.Appendix B Reference Frame Theory The general reference frame transformations theory is presented here.1. f c t which are time dependent can be represented as a space vector f : abc 2 f = f a t af b t a 2 f c t (B. Definition of the space vector in the abc reference frame. The goal of these transformations is to get rid of position varying inductances. The abc reference frame variables f a t . The transformations are based on the Clarke and Parke approaches. Space vector definition in the abc reference frame [12] 83 . f b t .
2). Fig. The real axis α is aligned with a (figure C.2.1) : abc f =Fe j t (B.3) 2 (B.The f a t . f c t variables can be expressed as: f a t =Fcos t (B. Stationary αβ reference frame [12].2) abc Definition of the space vector in the stationary αβ reference frame. B.4) 3 4 (B. Real and imaginary components of f can be calculated from following equations: 84 .5) 3 f b t = Fcos t f c t =Fcos t Therefore space vector f is expressed (figure B. Clarke transformation is used to express space vector in stationary reference frame (αβ). f b t .
4) Definition of the space vector in the stationary dq reference frame.3) 2 c 2 f = 0 3 3 2 f b t 3 2 f c t (B.2 f = f a t 3 1 f t 2 b 1 f t (B. B. Fig.Rotating dq reference frame [12] Components of the fdq are as follows [25]: f d = f cos dq f sin dq (B.3.6) In some cases it is necessary to go directly from f to fdq .3).5) f d= f sin dq f cos dq (B. The Park approach provides transformation from αβ stationary reference frame to dq rotating reference frame. The dq frame is fixed to the rotor and the values are independent from position changes and act like DC signals. Therefore special transformation matrix abc 85 . The real daxis of the rotating reference frame is phase shifted by dq angle from aaxis (figure B.
8) The inverse transformation fdq to f is maintained by inverse transformation matrix: abc 1 f =T dq⋅ f (B.is used to maintain transition [11].10) 86 .5 cosdq sin dq 0. [24]. f =T dq f (B.9) abc dq cos dq 2 1 2 cos dq T dq = 3 3 2 cos dq 3 [ sin dq 1 2 sin dq 1 3 2 sin dq 1 3 ] (B.5 ] (B.7) ⋅ abc dq T dq= 2 3 [ cos dq sin dq 0.5 2 3 2 3 cos dq 2 3 2 sin dq 3 0.
87 .
C2000 Tools Code Composer driver reduces complexity of the developed code. Optionally board has JTAG connector which acts like interface for other debuggers to provide high level language debug [23].1 JTAG Controller 5volt only operation with supplied AC adapter TI F28xx Code Composer Studio tools driver On board IEEE 1149. I/O) Onboard IEEE 1149. The eZDsp provides fast code verification. Features of the eZdsp: ● ● ● ● ● ● ● ● ● ● ● TMS320F2812 Digital Signal Processor 150 MIPS operating speed 18K words onchip RAM 128K words onchip Flash memory 64K words offchip SRAM memory 30 MHz.1 JTAG emulation connector 88 . Clock 2 Expansion Connectors (analog.Appendix C Hardware Setup eZdsp F2812 Board The eZdsp F2812 is specially designed board which allows to examine and evaluate the TMS320C2812DSP [23].
Block Diagram of the eZdsp F2812 [23].1: Fig.2.2. The connectors on the board are labeled.Functional overview of the eZdsp F2812 The block diagram of the eZdsp F2812 is presented on figure C. C. PCB outline of the eZdsp F2812 [23]. The eZdsp PCB outline is shown in figure C. 89 .1. The JTAG interface and expansion interface are the major interfaces of the board. Fig. C.
Description of the eZdsp F2812 connectors [23].2). The power is supplied via connector P6 (figure C.The connectors description is made in table C. The major logic blocks of the eZdsp are [23]: ● ● ● ● Analog interface connector I/O interface connector JTAG interface Parallel port JTAG controller Power Connector The eZdsp F2812 board is powered by 5 Volt only power supply delivered with eZdsp board.1: Table C.1. Memory of the eZdsp F2812 The eZDsp includes the following onchip memory [23]: ● ● ● ● 128K x 16 Flash 2 blocks of 4K x 16 single access RAM (SARAM) 1 block of 8K x 16 SARAM 2 blocks of 1K x 16 SARAM 90 .
91 .
The figure D. Shows model used for implementation in DSP platform: 92 .1. The selected control strategy was implemented in DSP.Appendix D Simulation and Implementation Models In this appendix model built in Matlab/Simulink and used for simulations is presented. All these components are described in Chapter 2. For implementation process Ti C2000 Simulink Toolbox was used. The system model consist of several components.2. Model is shown in figure D.
93 .. Simulation model built in Simulink.Fig.1. D.
Implementation model built in TI C2000 Toolbox 94 .Fig. D.2.
This action might not be possible to undo. Are you sure you want to continue?