Bahir Dar Institute of Technology Facult

BAHIR DAR UNIVERSITY
BAHIR DAR INSTITUTE OF TECHNOLOGY
FACULTY OF ELECTRICAL AND COMPUTER ENGINEERING
SCHOOL OF RESEARCH AND GRADUATE STUDIES
DESIGN AND ANALYSIS OF ADAPTIVE NEURO-FUZZY BASED POWER SYSTEM

STABILIZERS FOR DAMPING OSCILLATIONS IN MULTI-MACHINE POWER
SYSTEMS
BY
Getaneh Mesfin
Advisor:
Dr.-Ing Belachew Bantyirga
A Thesis submitted to Bahir Dar Institute of Technology in the partial Fulfillment of the
Requirements for the DEGREE OF MASTER OF SCIENCE IN ELECTRICAL
ENGINEERING (POWER SYSTEM ENGINEERING)
December, 2017
Bahir Dar, Ethiopia

BAHIR DAR UNIVERSITY
BAHIR DAR INSTITUTE OF TECHNOLOGY
FACULTY OF ELECTRICAL AND COMPUTER ENGINEERING
SCHOOL OF RESEARCH AND GRADUATE STUDIES
DESIGN AND ANALYSIS OF ADAPTIVE NEURO-FUZZY BASED POWER SYSTEM

STABILIZERS FOR DAMPING OSCILLATIONS IN MULTI-MACHINE POWER
SYSTEMS
Getaneh Mesfin
Approval by Board of Examiners
________________________ _____________ ____________
Chairman, Dept. Graduate committee Signature Date
Dr.-Ing. Belachew Bantyirga ______________ ____________
Thesis Advisor Signature Date
_______________ ____________
Internal Examiner Signature Date
________________ ____________
External Examiner Signature Date
ii
DECLARATION
I hereby declare that all information in this thesis document has been obtained and presented in
accordance with academic rules and ethical conduct. I also declare that, as required by these
rules and conduct, I have fully cited and referenced all material and results that are not original
to this work
Getaneh Mesfin _______________ __________
Name Signature date
Place: Bahir Dar
Date of submission: ____________________
This thesis has been submitted for examination with my approval as a university advisor.
Dr.-Ing Belachew Bantyirga ____________ ________
Advisor‟s Name Signature Date
iii
ACKNOWLEDGMENT
First of all my gratitude goes to the Almighty God for his positive will and help without which
the things I made here were value less.
Next, Words are inadequate to express the overwhelming sense of gratitude and humble regards
to my advisor Dr.-Ing .Belachew Bantyirga, chairman of Electrical and computer Engineering
for his unlimited help from the very beginning of selecting the title and to his constant
motivation, support, expert guidance, and constructive suggestion for the submission of my
progress report of thesis work “DESIGN AND ANALYSIS OF ADAPTIVE NEURO FUZZY
BASED POWER SYSTEM STABILIZERS FOR DAMPING OSCILLATIONS IN MULTI-
MACHINE POWER SYSTEMS”.
I would like to thank all whose direct and indirect support helped me completing my thesis in
time. This report would have been impossible if not for the perpetual moral support from my
family members, especially to my beloved mother Shashe Ayal, who has been the inspiration of
my life and my friends. I would like to thank them all.
Last but not the least; my deepest thanks go to my wife, Eden Tefera, for their affection, love,
and encouragement during my study.
iv
ABSTRACT
The Power systems are subjected to low-frequency disturbances that might cause loss of
synchronism and an eventual breakdown of the entire systems. The “low-frequency oscillations”
is one of the operational constraints which limit bulk power transmission through the power
networks, and also cause an eventual breakdown of the entire systems. For this problem power
system stabilizers (PSSs) are used to generate supplementary control signals for the excitation
system to damp the low-frequency power system oscillations and to offer an extra damping for
the synchronous generators. The supplementary power system stabilizer must be capable of
providing appropriate stabilization signals over a wide range of operating conditions, and
disturbances. However, a conventional power system stabilizer (CPSS) provides a positive
damping torque in phase with the speed signal to cancel the effect of the system negative
damping torque, because of the gains of this controller are determined for a particular operating
conditions. The conventional Power System Stabilizer which uses lead-lag compensation, where
the gain settings designed for specific operating conditions, is providing poor performance
under different loading conditions. The constantly changing nature of power system makes the
design of CPSS is a difficult task. Therefore, it is so difficult to design a stabilizer that could
present a good performance in all operating points of electric power systems. This thesis is
devoted to overcoming the drawback of conventional power system stabilizers (CPSSs), by
designing and modeling of Adaptive Neuro-Fuzzy power system stabilizers (ANFPSSs) are
proposed. The new design has a capacity to suppress and damp the oscillations when the
generators are subjected to different disturbances. The thesis deals with the design procedure for
a fuzzy logic based PSS (FLPSS) and a self-learning adaptive neural network based power
system stabilizer (ANFPSS) that improves the dynamic stability and provides supplementary
signals as consequences of which extending the power stability limits. The speed deviation of a
synchronous machine and acceleration are chosen as the input signals to the proposed controller.
The proposed technique has the features of a simple structure, fast response, effective and the
economical approach for attaining stabilization of the power systems. The simulation results of
the comparative study show that low frequencies inter-area power oscillation damping of
ANFIS controller based stabilizer is superior to the conventional power system stabilizer.
Moreover, the damping time provided by ANFIS controller based power system stabilizer is
73% less in comparison to CPSS under three - Phase fault conditions.
v
TABLE OF CONTENTS
ACKNOWLEDGMENT................................................................................................................. iv
ABSTRACT ..................................................................................................................................... v
TABLE OF CONTENTS ................................................................................................................ vi
LIST OF FIGURES ........................................................................................................................ xi
LIST OF TABLES ......................................................................................................................... xv
NOMENCLATURES AND ABBREVIATIONS ........................................................................ xvi
CHAPTER ONE .............................................................................................................................. 1
INTRODUCTION ........................................................................................................................... 1
1.1 Background of Study.............................................................................................................. 1
1.1.2 Classification of stability .................................................................................................2
1.2 Statement of problem ............................................................................................................. 4
1.3 Objectives of the Study .......................................................................................................... 5
1.3.1 General objective .............................................................................................................5
1.3.2 Specific Objectives ..........................................................................................................5
1.4 Scope of the study .................................................................................................................. 6
1.5 Methodology .......................................................................................................................... 6
1.6 Significant of the thesis .......................................................................................................... 7
1.7 Organization of the thesis ....................................................................................................... 7
CHAPTER TWO ............................................................................................................................. 9
LITERATURE REVIEW ................................................................................................................ 9
CHAPTER THREE ....................................................................................................................... 13
EXCITATION SYSTEM AND POWER SYSTEM STABILIZER ............................................. 13
3.1 The Excitation System of the Alternator .............................................................................. 13
3.1.1 Effects of Excitation system ..........................................................................................14
vi
3.1.2 Types of Excitation Systems .........................................................................................15
3.2 Excitation System Model ..................................................................................................... 16
3.2.1 Formulation of Heffron-Phillips Model of Excitation System for Stability Studies .....17
3.2.2 Effect of AVR on synchronizing and damping torque component ...............................19
3.3 Power System Stabilizer (PSS) Model ................................................................................. 20
3.3.1 General Structure of power system stabilizer ................................................................22
3.4 State space model of single machine and multi-machine system ........................................ 25
3.4.1 State - Space Representation of LTI Systems ...............................................................25
3.4.2 The State Equations .......................................................................................................26
3.4.3 State Equation Based Modeling Procedure ...................................................................28
3.5 Small signal model ............................................................................................................... 29
3.5.1 Eigenvalue analysis method ..........................................................................................30
3.5.2 Damping ratio and linear frequency ..............................................................................32
3.5.3 Damping performance criteria .......................................................................................32
3.5.4 Small signal performance measures .............................................................................33
3.5.5 The alternator state space model ...................................................................................33
3.6 Multi-machine power system analysis ................................................................................. 35
3.6.1 State space model of multi-machine system .................................................................36
CHAPTER FOUR .......................................................................................................................... 39
CONVENTIONAL POWER SYSTEM STABILIZER DESIGN ................................................. 39
4.1 Introduction .......................................................................................................................... 39
4.2 Design of Automatic voltage Regulator (AVR) ................................................................... 40
4.2.1 Automatic voltage regulator (AVR) design using root locus method ...........................42
4.2.2 AVR design using frequency response method.............................................................44
4.2.3 AVR design using state-space method ..........................................................................47
4.3 Design of power system stabilizer (PSS) ............................................................................. 49
vii
4.3.1 Power system stabilizer (PSS) design using root locus method ....................................52
4.3.2 Power system stabilizer (PSS) design using frequency response method .....................54
4.3.3 PSS design using state-space method ............................................................................58
4.4 conclusions ........................................................................................................................... 60
CHAPTER FIVE ........................................................................................................................... 62
DESIGN OFADAPTIVE NEURO-FUZZY BASED POWER SYSTEM STABILIZERS .......... 62
5.1 Introduction .......................................................................................................................... 62
5.2 Fuzzy logic Control system .................................................................................................. 63
5. 2.1 Fuzzy sets .....................................................................................................................64
5. 2.2 Membership functions ..................................................................................................64
5.3 Fuzzy Logic Controller ........................................................................................................ 65
5.3.1 Linguistic variable and rules bases ................................................................................67
5.3.2 Scaling factors ...............................................................................................................68
5.4 Artificial Neural Network .................................................................................................... 69
5.4.1 Basic Elements ..............................................................................................................69
5.4.2 Neural control ................................................................................................................70
5.5 Neuro-Fuzzy Systems .......................................................................................................... 71
5.5.1 Adaptive network fuzzy inference systems ...................................................................71
5.5.2 Neuro – fuzzy controller ...............................................................................................71
5.5.3 Adaptive network fuzzy inference systems as an estimator ..........................................72
5.6 Design of Fuzzy Logic Based Power system stabilizer ....................................................... 73
5.6.1 Input/output Variables ...................................................................................................74
5.6.2 Membership function editor ..........................................................................................74
5.6.3 Rule editor .....................................................................................................................77
5.6.4 Rule viewer ....................................................................................................................80
5.6.4 Surface viewer ...............................................................................................................80
viii
5.7 Adaptive Neuro-Fuzzy Controller Design ........................................................................... 81
5.7.1 Structure of the study system ........................................................................................82
5.7.2 Fuzzy logic Based Adaptive Neuro-Fuzzy Interface System PSS ................................83
5.7.3 Neural Network Representation of Fuzzy Logic Based power system stabilizer .........85
5.7.4 Offline adaptation by using ANFIS ...............................................................................86
5.8 Small-signal stability analyses of Multi-machine Systems .................................................. 88
5.8.1 A Two-Area 4-Machine, 10-Bus test Systems ..............................................................90
CHAPTER SIX .............................................................................................................................. 92
SIMULATION RESULTS AND DISCUSSION .......................................................................... 92
6.1 Simulation Result of CPSS for dynamic performance measure (case study one) ............... 92
6.2 Performance measures with Fuzzy logic controller ............................................................. 93
6.3 Performance measures with Neuro-Fuzzy controller ........................................................... 94
6.4 Comparisons and Discussion ............................................................................................... 96
6.5 Performance evaluation with small disturbances (case study two) ...................................... 96
6.5.1 Performance evaluation of Excitation System ..............................................................97
6.5.2 Performance evaluation with Conventional Power System Stabilizer (CPSS) .............99
6.5.3 Performance with Adaptive Neuro-Fuzzy Based Power System Stabilizer ...............101
6.6 Comparison of Conventional PSS and Adaptive Neuro-Fuzzy Based PSS ....................... 103
6.7 The proposed Simulation of Multi-machine power system (case study three) .................. 104
6.7.1 Performance evaluation without conventional power system stabilizer ( CPSS) under
single line to ground fault .........................................................................................105
6.7.2 Performance evaluation with CPSS under single line to ground fault ........................106
6.7.3 Performance evaluation with CPSS under symmetrical three phase fault (the worst
loading condition) .....................................................................................................108
6.7.4 Performance evaluation of Neuro-Fuzzy based PSS (ANFPSS) under single line to
ground fault ..............................................................................................................109
ix
6.7.5 Performance evaluation of Neuro-Fuzzy based pss (ANFPSS) under symmetrical
three phase fault ( the worst loading condition ) ......................................................111
CHAPTER SEVEN ..................................................................................................................... 113
CONCLUSIONS AND RECOMMENDATIONS ...................................................................... 113
7.1 Conclusions ........................................................................................................................ 113
7.2 Recommendation ................................................................................................................ 114
REFERENCES ............................................................................................................................ 115
APPENDICES ............................................................................................................................. 118
Appendix-1 Important results and data .................................................................................... 118
Appendix-2 Optimal synchronous machine and pss parameters ............................................. 120
Appendix-2A Optimal power system stabilizer parameters ................................................121
Appendix-2B System Data ..................................................................................................121
x
LIST OF FIGURES
Figure 1.1 classification of power system stability ........................................................................ 3
Figure 3.1 Schematic of the excitation system ........................................................................... 13
Figure 3.2 Block diagram of a synchronous generator excitation system ................................... 14
Figure 3.3 thyristor excitation systems with AVR....................................................................... 16
Figure 3.4 Haffron- Phillips Model of Excitation System with AVR ......................................... 18
Figure 3.5.haffron- Phillips model of excitation system with pss ............................................... 22
Figure 3.6: General Structure of PSS ........................................................................................... 22
Figure 3.7 block diagram of state-space representation [1] ......................................................... 29
Figure 3.9 single line diagrams of 4 machines and 10 buses System .......................................... 36
Figure 3.10 Block diagram including excitation system and PSS ............................................... 37
Figure 4.2 Simulink diagram of one synchronous machine with AVR simulated in Mat lab ..... 40
Figure 4.3 terminal voltages (VT) without pss ............................................................................ 41
Figure 4.4 Deltas (Load angle) without pss ................................................................................. 41
Figure 4.5 Electrical torques without pss ..................................................................................... 42
Figure 4.6 Root locus of Voltage Regulator forward loop showing dominant poles .................. 43
Figure 4.7 Frequency response of an uncompensated AVR Loop .............................................. 44
Figure 4.8 Comparison of frequency response with and without VR loop ................................. 46
Figure 4.9 Step responses without AVR ...................................................................................... 46
Figure 4.10 Step response of the lag compensated VR ............................................................... 47
Figure 4.11 the frequency response of 1st and 7th order controller for AVR ............................. 49
Figure 4.12 Simulink diagrams of PSS with AVR simulated in Mat lab .................................... 50
Figure 4.13 terminal voltages (VT) with pss ............................................................................... 51
Figure 4.14 Deltas or rotor angle with pss ................................................................................... 51
Figure 4.15 Electrical torques with pss ........................................................................................ 52
xi
Figure 4.16 Root locus of PSS loop showing the dominant complex poles ................................ 53
Figure 4.17 Root-locus of the final PSS loop showing Φp ~180º for dominant poles ................ 54
Frequency response of the Damping Loop ................................................................................. 55
Figure 4.19 Maximum phase addition Φm Vs alpha α ................................................................ 56
Figure 4.20 Root locus plot of the PSS loop showing the dominant poles .................................. 57
Figure 4.21 Step response of the final system with PSS loop ..................................................... 57
Figure 4.22 Root locus plot of the damping (PSS) loop with 5th order PSS implemented ......... 59
Figure 4.23 Frequency response comparisons of 11th and 5th order pss for pure damping….59
Figure 4.24 the frequency response of 1st and 7th order controller for AVR ............................. 60
Figure 4.25 Frequency response comparisons of the 11thorder and 5thorder pss for pure
damping .................................................................................................................. 61
Figure 5.1 the basic configuration of a fuzzy logic control ......................................................... 66
Figure 5.2 Non-linear model of a neuron..................................................................................... 70
Figure 5.3 Neuro-Fuzzy Control scheme ..................................................................................... 72
Figure 5.4: Basic Structure of Fuzzy Logic Controller................................................................ 73
Figure 5.5 FIS editor .................................................................................................................... 75
Figure 5.6 Membership function for speed deviation .................................................................. 76
Figure 5.7 Membership function for acceleration........................................................................ 76
Figure 5.8 Rule editors ................................................................................................................. 77
Figure 5.9 Rule Viewers .............................................................................................................. 80
Figure 5.10 Surface viewers ........................................................................................................ 81
Figure 5.12 Block diagram of ANFIS PSS model ....................................................................... 83
Figure 5.13 Neural Network Models of the ANFIS .................................................................... 86
Figure 5.14 ANFIS editor ............................................................................................................ 87
Figure 5.15 Structures of Sugeno type ANFIS for PSS ............................................................... 88
Figure 5.16 Reference frame transformation ............................................................................... 89
xii
Figure 5.17 Two Area Four-Machine power systems for Stability Analysis .............................. 90
Figure 6.1 Simulink diagram of conventional power system stabilizer....................................... 92
Figure 6.2 Unit step response of conventional power system stabilizer ...................................... 93
Figure 6.3 Simulink diagram of Fuzzy logic controller............................................................... 93
Figure 6.4 Unit step response of FLC for power system stabilizer ............................................. 94
Figure 6.5 Simulink diagram of Neuro-fuzzy controller ............................................................. 94
Figure 6.6 Unit step response of Neuro-Fuzzy controller system ............................................... 95
Figure 6.7 Unit step system responses for all controllers ............................................................ 95
Figure 6.8 Test Systems for Proposed Adaptive Neuro- Fuzzy Based PSS ................................ 97
Figure 6.9 Speed Deviation Responses with excitation system for -Ve K5 ................................ 98
Figure 6.10 speed deviation Responses with excitation system for +Ve K5 ............................... 98
Figure 6.11 Responses with CPSS for -Ve K5 ............................................................................ 99
Figure 6.12 Terminal voltage Responses with CPSS for -Ve K5 .............................................. 100
Figure 6.13 Responses with CPSS for +Ve K5 ......................................................................... 100
Figure 6.14 Terminal voltage Responses with CPSS for +Ve K5 ............................................. 101
Figure 6.15 Responses with ANFPSS for -Ve K5 ..................................................................... 101
Figure 6.16 Responses with ANFPSS for +Ve K5 .................................................................... 102
Figure 6.17 Terminal voltage Responses with ANFPSS for -Ve K5 ........................................ 102
Figure 6.18 Terminal voltage Responses with ANFPSS for +Ve K5........................................ 103
Figure 6.19 Simulation model implemented with CPSS and proposed ANFIS-PSS for Inter area
power oscillation stability Analysis ..................................................................... 104
Figure 6.20 Voltages and Power transfer from Area1 to Area2 ................................................ 105
Figure 6.21 Performance of without PSS for Rotor angle deviation of machine, Rotor speed of
machine (ω), active power of machine (Pa), and terminal voltage of machine when
single line to ground fault occurs in transmission line. ........................................ 106
Figure 6.22 Voltages and Power transfer from Area1 to Area2 ............................................... 106
xiii
Figure 6.23 Performances of CPSS for Rotor angle deviation of machine, Rotor speed of
single line to ground fault occurs in transmission line. ........................................ 107
Figure 6.25 Performance of CPSS for Rotor angle deviation of machine, Rotor speed of
symmetrical three phase fault occurs in transmission line. ................................. 109
Figure 6.27 Performance of Neuro-fuzzy for Rotor angle deviation of machine, Rotor speed of
single line to ground fault occurs in transmission lines ....................................... 110
symmetrical three phase fault occurs in transmission line ................................... 112
xiv
LIST OF TABLES
Table 5.1: Membership functions for fuzzy variables ................................................................. 74
Table 5.2 Decision Table ............................................................................................................. 78
Table 5.3 Rule matrix .................................................................................................................. 84
Table 6.1 Comparison between CPSS, FPSS and NFPSS ........................................................... 96
Table 6.2 Comparison of Settling time, the maximum overshoot of the dynamic responses of
rotor angle deviations of the study system with CPSS, and ANFISPSS .................. 103
Table A.1. Optimal synchronous machine and pss parameters values ..................................... 120
Table A. 2. Optimal power system stabilizer parameters .......................................................... 121
Table B.3. Machines Data .......................................................................................................... 121
Table C.4. the line data of the generators .................................................................................. 122
xv
NOMENCLATURES AND ABBREVIATIONS
Abbreviations Description
AC Alternating Current
ANFIS Adaptive Network Fuzzy Interference Systems
ANFPSS Adaptive Neuro-fuzzy Power System Stabilizer
ANN Artificial Neural Network
AVR Automatic Voltage Regulator
CPSS Conventional Power Stabilizer
DC Direct Current
FIS Fuzzy Interface System
FLC Fuzzy Logic Controller
FLPSS Fuzzy Logic Power System Stabilizer
LFO Low Frequency Oscillation
LG Line to Ground
MISO Multi Input Single Output
PID Proportional Integral Derivative Controller
PSS Power System Stabilizer
SMIB Single Machine Infinite Buses
TCSC Thyristor Controller Series Compositor
COA Center Of Gravity
HVDC High voltage direct current
LIST OF SYMBOLS
H Inertia time constant [s]

Ks Shaft stiffness constant [p.u. /rad]
K Gain factor
u Instantaneous voltage [p.u.]
T Time constant
TD Damping torque
xvi
TS Synchronizing torque [p.u.]
xd Dynamic state vector with n state variables
x0 Initial state vector
A State matrix
B Control matrix
C Output matrix
D Feed forward matrix
u Stabilizing signal
δ Power angle [degree]
λ Eigenvalue of system matrix A
σ Real part of eigenvalue λ
ζ Damping coefficient associated with eigenvalue λ
ᶲi Right eigenvector associated with the i-th eigenvalue
Ψd, Ed‟ Direct axis flux and field respectively
Ψq, E‟q Quadrature axis flux and field respectively
Vterm Terminal voltage
Pe Power delivered to the infinite bus.
Ω Angular frequency in radian/s
σ Leakage quantity
K1 − K6 Heffron-Philips constants
T1 − T4 Classical stabilizer time constants
Vs Correction voltage
Vref Reference voltage
x State vectors
y Output vectors
Vt Generator terminal voltage
Sm Slip
Efd Equivalent excitation voltage (field circuit voltage)
Eq Internal voltage behind transient reactance xd
Ed Internal voltage behind transient reactance xq
xvii
KE, TE AVR gain and time constant, respectively
Tw Washout time constant
Tm, Te Mechanical and electrical torques, respectively
m Mechanical quantity
e Electrical quantity
0 Initial values
d, q direct, quadrature axis component
h Main field quantity
Min, Max Maximum and minimum values
Σ Sum
Right eigenvector associated with the i-th eigenvalue
Left eigenvector associated with the i-th eigenvalue
γ Angular displacement [rad]
ψ Instantaneous flux [p.u.]
t Time [s]
xviii
CHAPTER ONE
INTRODUCTION
1.1 Background of Study
Power system stability was a famous topic which debated by an electrical engineer since
1920th. The ability of power system to recover from faults is becoming more important
nowadays because of the complexity of the system. Modern power systems are highly
sensitive to large disturbances which can propagate through the large network of
interconnection between in the absence of adequate safeguards. The main problems arise from
the increase in power demand. For the country like Ethiopia with people residing mostly in
villages and suburbs, supplying power to them has become the major responsibility for the
leaders of our country. With increasing lengths of transmission lines and the power carried by
them, there is a sharp increase in the probability of power fluctuations and faults. At this time
there may be a complete failure of power grids [1].
The Power transactions are increasing day by day in the restructured power systems. The
restructured power system is, therefore, expected to be operated at a greater variety of
operating Points and closer to their operating constraints. The “low-frequency oscillations” is
one of the operational constraints which limit bulk power transmission through power
network. In such scenario, power system controls play significant role. Power system controls
can contribute either positive or negative damping. Generation control and particularly the
generator voltage regulation can be significant sources of negative damping. High gain in the
generator voltage regulation can lead to poor or negative damping of the oscillation. This
problem has led to the implementation of Power System Stabilizer (PSS) to damp out the
oscillations. In order to have reliable generation and transmission, a power system should be
stable. Most present-day power generators are equipped with power system stabilizer (PSS) for
controlling slowing oscillating type instability. The Power system stabilizers are used to
produce controlling signals for the system and to decrease the oscillation of low-frequency
power system. A variety of methods are proposed to overcome the faults of common power
system stabilizer of which can name fuzzy logic, neural networks, and PID controller. The
thesis regarded to the Neuro-fuzzy controller to control the system based on simulations, it is
defined that the Neuro-fuzzy controller makes better answers to common stabilizer Fuzzy
1
Logic Controller (FLC) is chosen also as a controller for this thesis because it has several
advantages compared to the other classical controller. The advantages of FLC are such as
simplicity of control, low cost and the possibility to design without knowing the exact
mathematical model of the process. Neuro-Fuzzy also used a controller because of its
advantages that a combination of neural networks and fuzzy control enhances the performance
of the controller [2].
1.1.2 Classification of stability
Power system stability may be broadly defined as that property of power system that enables it
to remain in a state of operating equilibrium under the normal operating condition and to
regain an acceptable state of equilibrium after being subjected to disturbance [1].Power system
stability can be classified into three categories.
Dynamic or Small Signal Stability
Dynamic stability is concerns with the response of the synchronous machine to small
perturbations that are oscillating in nature. If these oscillations are of small amplitude, the
system may be considered as small signal stable, but if the amplitude of Oscillations is of
growing nature, with the passage of time the system may lose its stability. Usually, heavy
power flow in transmission lines or interaction of controller with system frequency is
responsible for small signal instabilities [3].
Steady State Stability

Steady-state stability analysis is the study of a power system, and its generators of strictly
steady-state conditions and trying to answer the question of what is the maximum possible
generator load that can be transmitted without loss of synchronism of any one generator.
Transient Stability
Transient stability is involving response of synchronous machine to large disturbances such as
application and clearing of faults, sudden load changes, of transmission lines or generators.
The figure below gives the overall picture of the stability problems. In this thesis focused on
small disturbance stability which is a part of the rotor angle stability [1].
2
Power system stability
The ability to remain in operating equilibrium

Equilibrium between opposite forces
Angle stability Frequency Voltage stability

stability
 Ability to maintain synchronism
 Torque balance of synchronous
Large
Machines
disturbance
voltage stability
Transient stability  Switching event
 Dynamics of ULTC, load
 Coordination of protection
and control
 Large disturbance
 First –swing a periodic drift Small
disturbance
 Study period up to 10 s
voltage stability
Small signal stability
Mid-term Long term
stability stability
Oscillatory instability
Non-
Oscillatory
instability
Insufficient Synchronizing torque Unstable the control action insufficient damping torque
Local Inter-area Control Torsional

plant modes modes modes
modes
Figure 1.1 classification of power system stability
3
1.2 Statement of problem
The increasing magnitude and complexity of interconnected power systems due to competitive
energy markets, economy, and population development have created the need to operate the
power systems close to their capacity limits. This leads to stability problems or poor dynamic
behaviors like power oscillations. These oscillations can cause a reduction of the system
components lifetime, expensive operations of the electrical grids and in the worst case, risks of
partial system collapse. Meanwhile, in the synchronous generator, the damping that the field
and damper windings provide to the rotor oscillations is weakened due to excitation control
system action. The reason for this is that in the rotor circuits appear additional currents induced
by the voltage regulation and those currents oppose to the currents induced by the rotor speed
deviations.
Now a day, the installment of the automatic voltage regulator (AVR) in a generator was
common in power system. Many generators are designed with high gain, fast acting AVR‟s to
enhance large scale stability to hold the generator in synchronism with the power system
during a large transient fault condition. But, the high gain in the generator voltage regulation
can lead to poor or negative damping of the oscillations. This thesis addressed a power system
stabilizer (PSS) to be added to synchronous generators to counteract the effect of the high gain
of AVR‟s.
Power system stabilizers (PSS) must be capable of providing appropriate stabilization signals
over a wide range of operating conditions and disturbances. However, a traditional (existing)
PSS provides a positive damping torque in phase with the speed signal to cancel the effect of
the system negative damping torque. Because the gains of this controller are determined for a
particular operating condition, they can‟t be valid for a wide range of operating conditions, is
giving the high starting overshoot, sensitivity to controller gains and sluggish response due to
sudden disturbances and loading conditions within small time. The constantly changing nature
of power system makes the design of Conventional PSS is a difficult task.
Power systems are highly non-linear, and random disturbances like load changes and time-
varying operating conditions make precise modeling of large power systems in real time very
difficult. This thesis attempt is made to overcome the problem of this nonlinearity and
4
modeling impression using the hybrid of fuzzy logic and neural network control is a powerful
tool in controlling systems which are not precisely characterized.
To overcome the drawback of conventional power system stabilizer (CPSS), designing and
modeling of Adaptive Neuro-Fuzzy power system stabilizer (ANFPSS) is proposed. The new
design has a capacity to suppress and damp the oscillations when the generators are subjected
to different disturbances. The output of Adaptive Neuro-fuzzy power system stabilizer is used
by the excitation system in order to damp the oscillations; hence to maintain the reliable and
sustainable operations of the grid, under both small and large disturbances.
1.3 Objectives of the Study
1.3.1 General objective
The general objective of this thesis is to design and analysis of adaptive Neuro-Fuzzy based
power system stabilizers for damping oscillations in the MATLAB package with control
system toolbox and to improve the small signal stability of the power systems.
1.3.2 Specific Objectives
The specific objectives include:
 Analysis of the nature of the power system stability, excitation system, the automatic
voltage regulator for synchronous generator and power system stabilizer.
 To review different approaches of Power system stabilizer designs, for further adoption
in this thesis.
 To develop the Neuro-fuzzy Power System Stabilizer which makes the system stable
quickly when fault happened in the transmission lines, and to validate the effectiveness
of the proposed controller.
 To reduce the settling time, rise time, peak amplitude and overshoot of the system.
 To operate the power system in a wide range.
 To improve the stability and reliability of the synchronous generators.
 By using the simulation to validate the Adaptive Neuro-fuzzy based power system
stabilizer and compare its performance with conventional power system stabilizer.
5
1.4 Scope of the study
The Adaptive Neuro-fuzzy power system stabilizer to be developed is constrained to an

analysis in a Single machine Infinite Bus (SMIB) model, and the simulation tests for Multi
machine systems. Moreover, the analysis of the PSS performance and the application of the
other control structure are conceived as a theoretical analysis of the results supported by
simulations.
1.5 Methodology
 Develop robust power system stabilizer for a single machine connected to an infinite
bus system using periodic output feedback model. The AVR, exciter and PSS are
modeling in MATLAB/Simulink according to IEEE standards and specifications.
 To perform the CPSS analysis using the SM-IB test model. For this task, MATLAB
Control System Toolbox is used to build an automatic tool to tune the lead-lag filters in
the PSS compensation stages.
 After the MATLAB tool has been built and test, a theoretical analysis of the Adaptive
Neuro-Fuzzy PSS performance to damp electromechanical oscillations is carried out.
 Design and modeling of Adaptive Neuro-fuzzy power system stabilizers for a multi-
machine power system using periodic output feedback is presented. In this method, the
periodic output feedback gain matrix is full, which makes control input as a function of
the output of all machines. Adaptive Neuro-fuzzy power system stabilizers for all
machines can be applied simultaneously to all machines. It is found that the designed
controller provides a good damping enhancement for a multi-machine power system. A
deep study of that control structure will open the possibility to evaluate if it can be
implemented as a controller in the synchronous generators. Simulations results are
performed.
 Finally, all the analysis and simulations results are presented in a final report of the
thesis including detailed description and instructions how to use the Adaptive Neuro-
fuzzy power system stabilizer (ANFPSS).
6
1.6 Significant of the thesis
The purpose of this master thesis is to develop a methodology of the modern Adaptive Neuro-
fuzzy power system stabilizer (ANFPSS) applied to the synchronous generators, Additionally,
the aim of this work is extended to study the application of an alternative control structure to
the generator for damping electromechanical oscillations, which is based on signal estimation
theory.
 Provide large blocks of power and increase the reliability of the system.
 Reduce the number of machines which are required both for operation at peak load and
required as a spinning reserve to take care of a sudden change of load.
 Provide an economical source of power to the consumers.
 All electrical devices are prone to failure when exposed to the power stability
problems.
 Enhance the existing and future power generation and transmission system.
1.7 Organization of the thesis
Chapter 1: gives an introduction to the principles of power system stability, the reasons and
motivation, problem statements, discusses the objectives, scopes and outlines methodologies
of the study. To damp out the low frequency oscillations PSS was introduced to have better
performance and stability.
Chapter 2: Describes previous works which are directly or indirectly related to the idea of this
thesis work are examined. Works reviewed in this section are generally categorized in to three
as: a recently published journal papers, M.Sc. thesis and Ph.D. thesis. Even though this work is
benefited from ample related materials, due to space limitation, only some selected materials
from each category are reviewed.
Chapter 3: Describes theoretical background of Excitation System, power system modeling

including synchronous machine, an automatic voltage regulator, excitation system elaborated
and derived from certain theories. The machine models described briefly and the linkage
between stability and its component also explained. The various components involved are
washout filter, dynamic compensator, torsional filter, limiter etc. Each block has its own
7
advantages and use which is vividly described in the chapter. Then various conventional
design techniques are studied so that better analysis is possible. The methods can be
implemented to study various aspects of PSS. The excitation system of an alternator system is
mathematically modeled in this section. MATLAB Simulink design is also provided there. The
various input and output parameters are selectively chosen.
Chapter 4: To analysis a Design of AVR and PSS Using Conventional Methods basically
deals with methods to implement the designing of PSS. The PSS design methods take in to
consideration are Root Locus Technique, Frequency Response Method, and State Space
Analysis in MATLAB SIMULINK. The design of AVR and PSS are done simultaneously
from which we can know the requirement of either lag or lead or lag-lead compensators.
Chapter 5: Introduces the concept of Artificial Intelligence especially the Artificial Neural
Networks, Fuzzy Logic, and Hybrid Intelligent System especially the Adaptive Neuro-Fuzzy
Inference System (ANFIS) and Deals with the Modern system control design of power system
stabilizer system with and Neuro-Fuzzy controller. The designed controller ensures that, for
any operating condition within a pre-defined domain, the system remains stable when
subjected to disturbances and it also deals the introduction of a wide area monitoring system
for better functioning of the system.
Chapter 6: Shows the simulation results obtained and comparison between different results
beyond to the starting Overshoot, settling time, Rise time and sluggish response.
Chapter 7: provides the conclusions and recommendations so far. The possible limitations in
proceeding research towards this work are discussed.
8
CHAPTER TWO
LITERATURE REVIEW
Numerous works have been done and published on the damping of power system low-
frequency oscillations to provide stability. In this section previous works which are directly or
indirectly related to the idea of this thesis work are examined. Works reviewed in this section
are generally categorized in to three as: a recently published journal papers, M.Sc. thesis and
Ph.D. thesis. Even though this work is benefited from ample related materials, due to space
limitation, only some selected materials from each category are reviewed.
Yagami and Tamura (2009) [4] provide a power system stability improvement technique
with the help of grouping of fault current limiter and thyristor controlled braking resistor. The
fault current limiter functions for restriction of fault currents, improvement of the power
system stability and containment of turbine shaft torsional oscillations. Next, the thyristor
controlled braking resistor functions with the intention of quick by managing generator
disturbances. The success of both devices has been illustrated with the help of Three-lines-to-
ground (3LG) fault in a two-machine infinite bus system. Simulation results represent a better
power system stability improvement and also the damping of turbine shaft torsional
oscillations with permitted level of temperature rise.
Li Zhengguo et al (2007) [5] is the concept of switched controller to examine a single

machine infinite bus power system when a symmetrical 3-phase short circuit fault occurs in
one of the transmission lines. Generally, a linear controller cannot provide a good transient
performance for such a power system with a large fault. The proposed switched controller
handles both temporary and permanent faults. The performance and the efficiency of the
proposed approach have been illustrated using simulation results. It should be noted that this
approach provides a solution to stabilization problem when a fault occurs. The future
extension of this approach is to investigate on achieving a good post-fault performance.
Soon Kiat Yee and Milano Vic (2008) [6] proposed a Fuzzy Logic Controller (FLC) for
decentralized stabilization of multi-machine power systems. The authors presented a unique,
largely analytical technique for design of robust Multi-Input–Single-Output (MISO) FLC for
enhancing damping and stability of an electrical power system without affecting the voltage
9
regulation. The proposed decentralized FLC uses a systematic analytical approach based on a
performance index in order to by-pass the need for prior knowledge about the system. The
proposed FLC tracks speed deviations to zero in order to stabilize the power output of the
generator, while, at the same time, it controls and stabilizes the terminal voltage of the
generator.
A. Dysko, W.E. Leithead and J. O’Reilly (2010) [7] have described a step-by-step
coordinated design procedure for power system stabilizers (PSSs) and automatic voltage
regulators (AVRs) in a strongly coupled system. The proposed coordinated PSS/AVR design
procedure is established within a frequency domain framework.
PDe Mello (1969) [8] has explored the phenomenon of stability of synchronous machines
under small perturbations by examining the case of single machine connected to an infinite bus
through an external reactance. The design of PSS for a single machine connected to an infinite
bus has been described using fast output sampling feedback.
J.H.Chow and J.J.Sanchez-Gasca (1989) [9] examine the four pole-placement for the design
of the power system stabilizer with the emphasis on the frequency characteristics of these
controllers. However in his work were not optimized in a global sense.
The basic PSS design idea in this paper is based on the stabilizer proposed by Demello and
Concordia (1969) [8]. However, the phase characteristics were obtained using a multi-
machine eigenvalue program instead of a single machine model. This work emphasized
enhancement of over-all system stability, and the authors considered simultaneous damping of
inter-area and local modes and discussed the performance of the PSS under different system
conditions. The authors claimed that the frequency response method used to compensate the
lag between the excitation input and the electrical torque was fairly robust.
A PID (Proportional + Integral + Derivative) type fuzzy logic power system stabilizer was
introduced by Hiyama et al (1994) [10]; the information of the integration of the speed
deviation was also used as one input and the origin of the phase plane was moved left ward or
right ward depending on the sign of the integral. Both simulations and experiments were
performed to demonstrate the effectiveness of the modification. Hiyama's heuristic based
approach showed some success in his series of research works. However, in his work, the
10
fuzzy PSS parameters were not optimized in a global sense because he claimed that the
parameters are reasonably insensitive to external conditions.
Son and Park (2000) applied the Linear Quadratic Gaussian technique to the design of TCSC
damping controller. The optimal Hankel norm approximation technique was used for obtaining
a low-order power system model, and a controller was designed based on the reduced-order
model. The authors also discussed the application of the Loop Transfer Recovery technique to
preserve the robustness of the designed damping controller. The 3-machine 9-bus system was
used here to verify the performance of the control strategy. Even though those methods
showed good results on a 3-machine 9-bus system, due to the matrix size problem, they were
not applied to higher order systems.
A hierarchical control strategy for both dynamic and steady state stability enhancement was
proposed by Del Rosso et al (2003) [3]. Control Strategies to mitigate adverse interactions
among the TCSC hierarchical controls were also presented. In this paper, the authors analyzed
and compared various locally measurable input signals qualitatively using the equal area
criterion. However, due to the limitation of the method, the authors only concentrated on
comparing the use of active power and line current as input signals and did not make an effort
to analyze the possibility of using bus voltage and bus frequency as input signals to the
damping controller.
M. F. Othman, M. Mahfouf and D.A. Linkens, have described the design procedure for a
fuzzy logic based power system stabilizer (FLPSS) and Adaptive Neuro-fuzzy inference
system (ANFIS) and investigates their robustness for single machine power system. Speed
deviation of a machine and its derivative are chosen as the input signals to the FLPSS [11].
Vani, M.U, Raju, G.S and Prasad, K.R.L [12] in their paper have presented a step-by-step
design methodology of an Adaptive Neuro-Fuzzy Inference system and optimization methods
based automatic voltage regulator and power system stabilizer.
Chun-Jung Chen [13] has presented an adaptive power system stabilizer (PSS) which
consists of a recurrent neural network controller (RNNC) and a compensator to damp the
oscillations of the power system. The function of RNNC is to supply an adaptive control signal
to the exciter or governor, which can damp most of the power system‟s oscillations.
11
Sumina D. (2008) [14] has presented the usage of neural network based excitation control on
single machine infinite bus. The proposed feed forward neural network integrates a voltage
regulator and a power system stabilizer.
Barton (2004) presented a robust artificially intelligent Adaptive Neuro-Fuzzy Inference

System (ANFIS) based PSS (ANFPSS) design for damping electromechanical modes of
oscillations and enhancing power system synchronous stability. An actual power system was
decomposed into separate subsystems; each subsystem consisted of one machine. The local
ANFPSS was associated with each subsystem. The local feedback controllers were depending
only on information particular to their subsystem. The input signals were the speed, power
angle and real power output. Nonlinear simulations showed the robustness of the ANFPSS.
Hsu and Chen (1991) used a neural network to tune the parameters of a conventional PI
(Proportional + Integral) type PSS. Abdel-Magid et al (2000) wanted to select a single set of
PSS parameters which can make the PSS simultaneously stabilize the power system over a
wide range of operating conditions.
All the above papers are proposed by different researchers are valid for a limited operating
conditions .These are not valid for wide operating conditions due to high starting over shoot,
sensitivity to controller gains and sluggish response during sudden disturbances and heavy
loading conditions. This thesis the Neuro-fuzzy techniques have implemented from the fusion
of Artificial Neural Networks (ANN) and Fuzzy Inference Systems (FIS) which form a
framework for solving real world problems. A Neuro-fuzzy system is based on a fuzzy system
which is trained by a learning algorithm derived from neural network theory. While the
learning capability is an advantage from the viewpoint of FIS, the formation of linguistic rule
base will be advantageous from the viewpoint of ANN. There are several approaches to
integrate ANN and FIS and very often the choice depends on the application. The proposed
Neuro-fuzzy controller‟s structure guides the speed deviation and acceleration error signals
through the fuzzy inference system to get an output suitable for damping the oscillations.
Simulation results validate the superior performance of ANFIS system over the conventional
approach. The Multiple model adaptive controllers are developed for SMIB system and Multi
machine system as intelligent control. The controller performance assessment is designed by
using Fuzzy Logic control and Neural Network control.
12
CHAPTER THREE
EXCITATION SYSTEM AND POWER SYSTEM STABILIZER
3.1 The Excitation System of the Alternator
A modern excitation system contains the components of Automatic Voltage Regulators

(AVR), Power System Stabilizers (PSS), and filters, which help in stabilizing the system and
maintaining almost constant terminal voltage and constant frequency. These components can
be analog or digital depending on the viability, complexity, and operating conditions. The final
aim of the excitation system is to reduce swings due to transient rotor angle instability and to
maintain a constant voltage. To perform this, a step voltage is fed as a reference voltage, which
is taken to be a constant. The AC voltage is first converted into DC voltage by rectifier units
and is fed to the excitation system via its components such as AVRs, PSSs etc. The AVR
sends actuating signal which actuates the excitation by the exciter of the Alternator. This is a
continuous process and has got certain limitations. Meanwhile, there is a so-called Auxiliary
control which may be represented by a PSS in a real-time application. The PSS takes up data
from infinity bus directly. The PSS deals with small voltage fluctuations and does fine
adjustment as required [12].
Synchronous
generator
Transmission line
Auxiliary control
Excitation
AVR
Figure 3.1 Schematic of the excitation system Vref
13
3.1.1 Effects of Excitation system
The main objective of the excitation system is to control the field current of the synchronous
machine. The field current is controlled so as to regulate the terminal voltage of the machine.
As the field circuit time constant is high (of the order of a few seconds), fast control of the
field current requires field forcing. Thus exciter should have a high ceiling voltage which
enables it to operate transiently with voltage levels that are 3 to 4 times the normal. The rate of
change of voltage should also be fast. Because of the high reliability required, unit exciter
scheme is prevalent where each generating unit has its individual exciter. From the power
system viewpoint, the excitation system should contribute to effective control of voltage and
enhancement of system stability. It should be capable of responding rapidly to a disturbance so
as to enhance transient stability [15].
Limiters and
protective circuits
Terminal voltage
transducer and Load
compensator
Exciter Generator
Ref Regulator To power
System
Power system
stabilizer
Figure 3.2 Block diagram of a synchronous generator excitation system
14
Fundamentally, simplest excitation system consists of an exciter only. When the excitation
system is also performs the task of maintaining the terminal voltage of alternator constant,
under varying load conditions. The functional block diagram of a typical control system for a
large synchronous generator is shown in Figure 3.2. The following is a brief description of the
various subsystems identified in above Figure 3.2.
Exciter: to provide dc power to the synchronous machine field winding, constituting the
power angle of the excitation system.
Regulator: processes and amplifies input control signal to a level and form appropriate for
control of the exciter.
Terminal voltage transducer and load compensator: is sensing generator terminal voltage,
rectifies filters it to the dc quantity, and then compares it with a reference voltage which
represents the desired terminal voltage, and also load compensation can be provided.
Power system stabilizer: - Provides an additional input signal to the regulator to damp power
system oscillation. Some commonly used input signals are rotor speed deviation, accelerating
power and frequency deviation.
Limiter and protective circuits: These include a wide array of control and protective
functions which ensure that the capability limits of exciter and generator are not exceeded.
Common functions are field-current limiter, maximum excitation limiter, under excitation
limiter etc.
3.1.2 Types of Excitation Systems
DC excitation system: The system which utilize a direct current generator with a
commentator as the source of excitation system power.
AC excitation system: The system which uses an alternator and either stationary or rotating
rectifiers to produce direct current needed for generator field.
Static excitation system: The system in which excitation power is supplied through
transformers and rectifiers. The first two types of exciters are also called rotating exciters
which are mounted on the same shaft as the generator and driven by the prime mover. The
voltage regulator for DC excitation systems were based on rotating amplifier (amplifying) or
15
magnetic amplifiers. AC and static excitation systems invariably use electronic regulators
which are fast acting and result in the phase control of the controlled rectifiers using thyristor.
Static excitation system offers the ultimate response, which is virtually negligible, and ceiling
voltage which are limited only by generator rotor design considerations. With the help of fast
transient forcing of excitation and the boost of internal machine flux, the electrical output of
the machine may be increased during the first swing compared to the results obtainable with a
slow exciter. The static excitation system utilizes transformers to transform voltage to an
appropriate level. Rectifiers, either controlled or non-controlled, provide the necessary direct
current for generator field [7].
3.2 Excitation System Model
The input control signal to the excitation system is normally the generator terminal voltage E
may be expressed in complex form:
Appling a small perturbation, it may write as:
( ) ( ) ( )
By neglecting higher order terms in above expression, the above equation reduced to
Terminal Voltage transducer +Vref Exciter EFmax
1 V1- KA Efd
𝝨
ET
1 𝑠𝑇𝑅
EFmin
Figure 3.3 thyristor excitation systems with AVR
16
From block of figure.3.3, using perturbed values,
Therefore,
( )
In terms of perturbed values
ΔEt = KA ( )
For the small scale analysis, the thyristor excitation system as shown in fig.3.3 is considered.
The nonlinearity associated with the ceiling on the exciter output voltage represented by EFMAX
and EFMIN, which is ignored for small-disturbance studies.
3.2.1 Formulation of Heffron-Phillips Model of Excitation System for Stability Studies
The basis for the model presented here, which was originally proposed by Heffron and
Phillips, is the “Single Machine Infinite Bus” (SMIB) setup. By introducing a number of new
constants, a very compact notation is achieved. The model is useful to directly implement a
simplified representation of an SMIB system, including the mechanical dynamics, field
winding, and excitation system. This implementation can be used directly for stability studies.
In the present case, a generic simplified representation of the excitation system is used.
Detailed descriptions and common variants of these systems can be found in. All quantities
presented are in per unit. The mechanical system is represented by the system inertia and the
damping constant, where the torque balances ∆Tm − ∆Te is considered as an input and the
incremental torque angle ∆δ as an output. The electrical part of the system consists of three
main parts:
a. The composition of the electrical torque (influenced by ∆δ over constant K1 and the
internal incremental voltage ∆ψfd over constant K2),
b. The effect of the field winding (determined by the field winding constant K3
and influenced by ∆δ over constant K4),
c. The effect of the excitation system (influenced by ∆δ over constant K5 and ∆ψfd over
constant K6).
17
The excitation system itself is modeled by a first-order transfer function including the
amplification factor KA and the time constant TA.
K4
Influence of torque angle on internal voltage
Voltage set mechanical torque change
Point change Exciter +Δvs ΔEfd field circuit + ΔTm mechanical part
𝑲𝟑
𝑲𝑬
Vr
⅀ Δ
𝟏 ψfd𝑺𝑻𝟑
+ ∆ᴪfd
⅀
∆Te ⅀
𝟏 Δ𝝎r 𝝎𝒐 ∆
⅀ 𝟏 𝑺𝑻𝑬 K2 𝟐𝑯𝑺 𝑲𝑫
𝑺
-Δv Excitation system

1 field winding
Influence on internal synchronous speed

Voltage on electrical torque
K1
Influence on torque angle on
K6 Influence on internal voltage Electrical torque
+ On terminal voltage
𝟏 ΔEt +
K5
𝟏 𝑺𝑻𝑹 ⅀
Influence torque angle on terminal
Figure 3.4 Haffron- Phillips Model of Excitation System with AVR
Interpretation of K-constants
The constants K1 to K6 shown in block diagram describe internal influence factors within the
system and can be found by a comparison of coefficients with the equations governing the
synchronous machine dynamics. While K1 and K2 are derived from the computation of the
electric torque, K3 and K4 have their origin in the field voltage equation. K5 and K6 come
from the equation governing the terminal voltage magnitude [1].
K1-Influence of torque angle on electric torque

K2-Influence of internal Voltage on electric torque
18
K3- field winding constant
K4-Influence of torque angle on field voltage
K5-Influence of torque angle on terminal voltage
K6-Influence of internal voltage on terminal voltage
∆Et = K5 ∆δ + K6 ∆ψfd
The coefficient K6 is always positive, whereas K5 can be either positive or negative,

depending on the operating condition and external network impedance RE +jXE. The value of
K5 has a significant bearing on the influence of the AVR on the damping of the system
oscillations. With constant field voltage (∆Efd = 0), the field flux variations are caused only by
feedback of ∆ through coefficient k4. This represents the demagnetizing effect of armature
reaction. The change in air‐gap torque due to field flux variations caused by rotor angle change
is given by
The constants K2, k3 k4, are usually positive. The contribution of ∆ᴪfd to synchronizing and
damping torque components depends on the oscillating frequency.
3.2.2 Effect of AVR on synchronizing and damping torque component
With the automatic voltage regulator action, the field flux variations are cause by the field
voltage variations, in addition to armature reaction. From diagram of figure 3.4 to observe that
∆ψfd 0 ( )1
By grouping terms involving ∆ᴪfd and rearranging,
, ( ) -
( )
The range in air‐gap torque due to change in field flux linkage is
∆Te ∆ᴪfd = K2∆ᴪfd
19
As noted before, the constants K2, k3, k4 and k6 are usually positive; however, k5 may take
either positive or negative values. The effect of the AVR on damping and synchronizing
torque components is therefore primarily influenced by k5 and kE.
For high values of external system reactance and high generator output K5 is negative. In
practice, the situation where K5 is negative is commonly encountered. For such cases, a high
response exciter is beneficial in increasing synchronizing torque. However, in so doing it
introduces negative damping. Thus have conflicting requirements with regard to exciter
response. The one possible recourse is to strike a compromise and set the exciter response so
that it results in sufficient synchronizing and damping components for the expected range of
system operating conditions. This may not always be possible. It may necessary to use a high
response exciter to provide the required synchronizing torque and transient stability
performance. With a very high external system reactance, even with low exciter response the
net damping torque coefficient may be negative. An effective way to meet the conflicting
exciter performance requirements with regard to system stability is to provide a power system
stabilizer.
3.3 Power System Stabilizer (PSS) Model
The basic function of a power system stabilizer is to add damping to the generator rotor
oscillations by controlling its excitation using auxiliary stabilizing signal(s). To provide
damping, the stabilizer must produce a component of electrical torque in phase with the rotor
speed deviations. It is well established that fast acting exciters with high gain AVR can
contribute to oscillatory instability in power systems. This type of instability is characterized
by low frequency (0.2 to 3.0 Hz) oscillations which can persist (or even grow in magnitude)
for no apparent reason [16]. This type of instability can endanger system security and limit
power transfer. The major factors that contribute to the instability are:
Loading of the generator or tie lines

Power transfer capability of transmission lines
Power factor of the generator (leading power factor operation is more problematic than
lagging power factor operation)
20
AVR gain
A cost-efficient and satisfactory solution to the problem of oscillatory instability is to provide

damping for generator rotor oscillations. This is conveniently done by providing Power
System Stabilizers (PSS) which are supplementary controllers in the excitation systems. The
objective of designing PSS is to provide additional damping torque without affecting the
synchronizing torque at critical oscillation frequencies. It can be generally said that need for
PSS will be felt in situations when power has to be transmitted over long distances with weak
AC ties. Even when PSS may not be required under normal operating conditions, they allow
satisfactory operation under unusual or abnormal conditions which may be encountered at
times. Thus, PSS has become a standard option with modern static exciters and it is essential
for power engineers to use these effectively. Retrofitting of existing excitation systems with
PSS may also be required to improve system stability.
The theoretical basis for a PSS may be illustrated with the aid of the block diagram shown in
Figure 3.5. This is an extension of the block diagram shown in Figure 3.4 and includes the
effects of a PSS. Since the purpose of a PSS is to introduce a damping torque component, a
logical signal to use for controlling generator excitation is the speed deviation ∆ωr. If the
exciter transfer function and the generator transfer function between ∆Efd and ∆Te were pure
gains, a direct feedback of ∆ωr would result in a damping torque component. However, in
practice both the generator and the exciter exhibit frequency dependent gain and phase
characteristics. Therefore, the PSS transfer function G(s) should have appropriate phase
compensation circuits to compensate for the phase lag between the exciter input and the
electrical torque. In the ideal case, with the phase characteristic of G(s) being an exact inverse
of the exciter and generator phase characteristics to be compensated, the PSS would result in a
pure damping torque at all oscillating frequencies [17].
21
Figure 3.5.haffron- Phillips model of excitation system with pss
3.3.1 General Structure of power system stabilizer
The block diagram of basic structure of power system stabilizer is depicted in Figure 3.6.It
consists of gain block, washout circuit, dynamic compensator, and limiter.
Signal washout Lead- Lag Limiter

Compensator output
sTW
Input (∆ωr) Kpss v1 v2 (Vs)
1 sTW T(S)
Figure 3.6: General Structure of PSS
22
The functions of each of the components of PSS are given in subsequent sections.
 PSS Gain
Stabilizing gain KPSS determines the amount of damping introduced by PSS. Ideally,
PSS gain is set to get the maximum damping of the oscillatory modes. However, due
to practical considerations, a high gain may not be always the best option and may
cause excessive amplification of stabilizer input signal. In general, the gain value is set
such that it results in satisfactory damping of critical system modes without
compromising the stability limits.
 Washout Circuit
The washout circuit is provided to eliminate steady-state bias in the output of PSS
which is modifying the generator terminal voltage. The PSS is expected to respond
only to transient variations in the input signal, said rotor speed and not to the dc offsets
in the signal. The washout circuit acts essentially as a high pass filter and it must pass
all frequencies that are of interest. If only the local modes are of interest, the time
constant Tw can be chosen in the range of 1 to 2. However, if inter-area modes are
also to be damped, then Tw must be chosen in the range of 10 to 20. The value of Tw
= 10 is necessary to improve damping of the inter-area modes [1]. There is also a
noticeable improvement in the first swing stability when Tw is increased from 1.5 to
10. The higher value of Tw also improved the overall terminal voltage response during
system islanding conditions.
 Lead-Lag Compensator
Lead-Lag compensator block provides the suitable phase lead to compensate for the
phase lag between the exciter input and generator electrical torque. The dynamic
compensator, used in practice, is made up of several multiple stages of lead-lag
compensators depending upon the requirement of phase compensation to be provided.
 PSS Output Limits
Stabilizer output voltage is limited to typical maximum and minimum values to restrict
the level of generator terminal voltage fluctuation during transient conditions. Large
output limits ensure the maximum contribution of stabilizers but generator terminal
23
voltage may face large fluctuation. The main objective in selecting the output limits of
PSS is to allow maximum forcing capability of stabilizer while maintaining the
terminal voltage within desired limits. Most commonly used the value of maximum
limit is between 0.1 and 0.2 p.u. while the minimum limit is taken between -0.05 and -
0.1p.u.
 Input of PSS
Many signals, like rotor speed deviation, frequency deviation, change in load angle,
change in electrical power etc. are possible to use as an input signal to PSS. However,
from a practical point of view, the following three types of input signals are most
commonly used as input to power system stabilizer:
 Rotor Speed Deviation (∆ω)

 Frequency Deviation (∆f)
 Electrical Power Deviation (∆P)
Though in practice, the speed deviation signal is used as input to PSS, it is inherently sensitive
to torsional modes of oscillations in the frequency range of 8 to 20 Hz, which can lead to
negative damping for the torsional mode. Hence, It also advisable to use torsional filter-
typically a low pass filter-for avoiding interaction of PSS with the torsional mode of
oscillations.
Speed as input: - A power system stabilizer utilizing shaft speed as an input must compensate
for the lags in the transfer function to produce a component of torque in phase with speed
changes so as to increase damping of the rotor oscillations.
Power as input: - The use of accelerating power as an input signal to the power system
stabilizer has received considerable attention due to its low level torsional interaction. By
utilizing heavily filtered speed signal the effects of mechanical power changes can be
minimized. The power as input is mostly suitable for closed loop characteristic of electrical
power feedback.
Frequency as input:- The sensitivity of the frequency signal to the rotor input increases in
comparison to speed as input as the external transmission system becomes weaker which tend
24
to offset the reduction in gain from stabilizer output to electrical torque ,that is apparent from
the input signal sensitivity factor concept.
The PSS parameters should be such that the control system results into the following.
 Maximize the damping of the local plant mode as well as inter-area mode oscillations
without compromising stability of other modes.
 Enhance system transient stability
 Not adversely affect system performance during major system upsets which cause
large frequency excursions.
 Minimize the consequences of excitation system malfunction due to component
failure.
3.4 State space model of single machine and multi-machine system
3.4.1 State - Space Representation of LTI Systems
The classical control theory and methods (such as root locus) that have been using in class to
date are based on a simple input-output description of the plant, usually expressed as a transfer
function. These methods do not use any knowledge of the interior structure of the plant and
limit us to single-input-single-output (SISO) systems, and as a researcher has seen allows only
limited control of the closed-loop behavior when feedback control is used.
Modern control theory solves many of the limitations by using a much “richer” description of
the plant dynamics. The so-called state-space description provides the dynamics as a set of
coupled first-order differential equations in a set of internal variables known as state variables,
together with a set of algebraic equations that combine the state variables into physical output
variables. The concept of the state of a dynamic system refers to a minimum set of variables,
known as state variables, which fully describe the system and its response to any given set of
inputs. In particular, a state-determined system model has the characteristic that:
A mathematical description of the system in terms of a minimum set of variables xi (t), i = 1, .

. . , n, together with knowledge of those variables at an initial time t0 and the system inputs for
time t ≥ t0, are sufficient to predict the future system state and outputs for all time t > t 0. This
definition asserts that the dynamic behavior of a state-determined system is completely
25
characterized by the response of the set of n variables xi (t), where the number n is defined to
be the order of the system.
3.4.2 The State Equations
A standard form for the state equations is used throughout system dynamics. In the standard
form the mathematical description of the system is expressed as a set of n coupled first-order
ordinary differential equations, known as the state equations, in which the time derivative of
each state variable is expressed in terms of the state variables x1(t), . . . , xn(t) and the system
inputs u1(t), . . . , ur(t). In the general case the form of the n state equations is:-
ẋ1 = f1 (x, u, t)
ẋ2 = f2 (x, u, t)
. .
.= .
. .
Ẋn = fn (x, u, t) (3-1)
Where ẋi= dxi/dt and each of the functions fi (x, u, t), (i = 1 . . . n) may be a general nonlinear,
time varying function of the state variables, the system inputs, and time. It is common to
express the state equations in a vector form, in which the set of n state variables is written as a
state vector x(t) = [x1(t), x2(t), . . . , xn(t)]T, and the set of r inputs is written as an input vector
u(t) = [u1(t), u2(t), . . . , ur(t)]T. Each state variable is a time varying component of the column
vector x (t).This form of the state equations explicitly represents the basic elements contained
in the definition of a state determined system. Given a set of initial conditions (the values of
the xi at some time t0) and the inputs for, t ≥ t0, the state equations explicitly specify the
derivatives of all state variables. The value of each state variable at some time ∆t later may
then be found by direct integration. The system state at any instant may be interpreted as a
point in an n-dimensional state space, and the dynamic state response x (t) can be interpreted
as a path or trajectory traced out in the state space.
In vector notation the set of n equations in Eqs. (1) May be written:
ẋ = f (x, u, t). (3-2)
26
Where f(x, u,t) is a vector function with n components fi (x, u, t).In this note restrict attention
primarily to a description of systems that are linear and time-invariant (LTI), which is systems
described by linear differential equations with constant coefficients. Then the LTI system of
order n and with r inputs in, Eqs. (1) become a set of n coupled first-order linear differential
equations with constant coefficients and may be written compactly in a matrix form:
ẋ1 = a11x1 + a12x2 + . . . + a1nxn+ b11u1 + . . . + b1rur

ẋ2 = a21x1 + a22x2 + . . . + a2nxn + b21u1 + . . . + b2rur
.
. (3-3)
Ẋn = an1x1 + an2x2 + . . . + annxn + bn1u1 + . . . + bnrur
Where
The coefficients of aij and bij are constants which describe the system. This set of n equations
defines the derivatives of the state variables to be a weighted sum of the state variables and the
system inputs.
ẋ = A. ∆x + B.∆u (3-4)
1 11 1 1 1 11 1
1
1 1
[ ] = [ ][ ] + [ ][ ] (3-5)
1 1
where the state vector x is a column vector of length n, the input vector u is a column vector of
length r, A is an n × n square matrix of the constant coefficients aij, and B is an n × r matrix of
the coefficients bij that weight the inputs.
A system output is defined to be any system variable of interest. A description of a physical

system in terms of a set of state variables does not necessarily include all of the variables of
direct engineering interest. An important property of the linear state equation description is
that all system variables may be represented by a linear combination of the state variables xi
and the system inputs ui. An arbitrary output variable in a system of order n with r inputs may
be written:
27
Y (t) = c1x1 + c2x2 + . . . + cnxn+ d1u1 + . . . + d1ur (3-6)
Where
The ci and di are constants. If a total of m system variables are defined as outputs, the m such
equations may be written as:
y1 = c11x1 + c12x2 + . . . + c1nxn+ d11u1 + . . . + d1rur

y2 = c21x1 + c22x2 + . . . + c2nxn + d21u1 + . . . + d2ru
. . (3-7)
. .
Ym = cm1x1 + cm2x2 + . . . + cmnxn + dm1u1 + . . . + dmrur
In matrix form:
1 11 1 1 1 11 1
1
1 1
= + [ ] (3-8)
[ ] [ 1 ][ ] [ 1 ]
∆y = C. ∆x + D. ∆u (3-9)
Where y is a column vector of the output variables yi(t), C is an m× n matrix of the constant
coefficients cij that weight the state variables, and D is an m × r matrix of the constant
coefficients dij that weight the system inputs. For many physical systems the matrix D is the
null matrix, and the output equation reduces to a simple weighted combination of the state
variables [1].
∆y = C. ∆x. (3-10)
3.4.3 State Equation Based Modeling Procedure
The complete system mod el for a linear time-invariant system consists of the following form:
i) A set of n state equations, defined in terms of the matrices A and B, and

ii) A set of output equations that relate any output variables of interest to the state
variables and inputs, and expressed in terms of the C and D matrices. The task of
28
modeling the system is to derive the elements of the matrices, and to write the
system model in the form:
∆ẋ =A. ∆x + B .∆u (3-11)
∆y = C. ∆x + D .∆u (3-12)
By taking the Laplace transformation of the above equations, this obtained the state equations
in the frequency domain:
∆u B + ∆Ẋ 1 ∆X C + ∆y
𝝨 I 𝝨
S
+
Figure 3.7 block diagram of state-space representation [1]
The matrices A and B are properties of the system and are determined by the system structure
and elements. The output equation matrices C and D are determined by the particular choice of
output variables. The alternator state space model of this thesis the design of the final power
system stabilizer has been modeled taking into consideration sub transient effects.
3.5 Small signal model
The small signal stability is the ability of the interconnected synchronous machines of a power
system to remain in synchronism when the power system is subjected to small disturbances.
These small disturbances always occur. Good examples are variations of load and generation.
These variations are defined as small disturbances because they are sufficiently small for the
linearization of system equations and applying Eigen analysis. Such variations result in power
flow variations in the transmission system along with associated rotor angle variations. These
rotor angle variations need to be sufficiently damped out to prevent the continuation of
29
oscillations. With no damping, such variations can instigate power flow variations where the
reduction of load in a group of machines is accompanied by an increase in load of another
group of machines that are at a considerable electrical distance from the first group of
machines. These variations can escalate into large magnitudes leading to the loss of
synchronism [1].
Small signal instability can occur in two forms:-
i) Increase in rotor angle due to insufficient synchronizing torque,

ii) Rotor oscillation due to insufficient damping torque.
Generally a disturbance in a power system can cause a change in the value of the angle (Δδ)
and speed (Δω) which in turn leads to a change in electrical torque.
𝝎 (3-13)
Where
Ks∆ is known as synchronizing torque component which is in phase with change in rotor
angle, ∆δ.
KD∆ω is called the damping torque component which is in phase with change in speed,
∆ω. Both parts of the electrical torque are involved in power system stability. Insufficient
synchronizing torque leads to aperiodic drift in rotor angle and insufficient damping torque
leads to oscillations.
3.5.1 Eigenvalue analysis method
The dynamic behavior of a power system can be described by a set of first order nonlinear
ordinary differential equations and a group of nonlinear algebraic equations in the following
form of vector matrix notation [1].
( ) (3-14)
( ) (3-15)
( ) (3-16)
In which and Xa are the dynamic and algebraic state vectors respectively, u and y are the
input and output vectors respectively. Although power system is a nonlinear, it can be
30
linearized by small signal stability at a certain operating point. The model that is linearized
around this point is therefore only valid close to the operating point. Deviations in the
operating point are described in the following equation, where the operating point is marked
with the superscript 0.
(3-17)
(3-18)
( 3-19)
(3-20)
[( ) ( )( )] (3-21)
As the deviations are small, a Taylor series expansion can be applied to the nonlinear function
f(x, u). By neglecting the second and higher order powers of ∆x and ∆u to get the following
equation
[( ) ( )( )]
( ) (3-22)
( )
(3-23)
Similar expansion can be applied to the functions g and h so that a linear differential algebraic
matrix equation is derived and finally the linearized a linearized ordinary differential equation
model of the system can be written in the following form:
=A. + B .∆u (3-24)
∆y = C. + D .∆u (3-25)
where
, ∆y , and ∆u express state, output, and input vector, respectively; A, B, C, and D
expresses the state, input, output, and feed forward matrices, respectively.
31
3.5.2 Damping ratio and linear frequency
The eigenvalues of λ of matrices can be obtained by solving the characteristic equation of A

with vector Φ:
(3-26)
( ) (3-27)
Det( )=0 (3-28)
This leads to the complex eigenvalues λ of A in the form
𝝎 (3-29)
The real part represents the damping of the corresponding mode and the imaginary part
represents the frequency (f) of the oscillation given by
𝝎
(3-30)
The damping ratio of this frequency is given by type equation here:
(3-31)
√ 𝝎
The parameters ξ and ω in the above equations are used for the stability prediction in this
study.
 When the eigenvalues have negative real parts, the original system is asymptotically
stable.
 When at least one of the eigenvalues has a positive real part, the original system is
considered unstable.
3.5.3 Damping performance criteria
In order to analyze the dynamic behavior of a power system a set of tools based on eigenvalue
analysis must be applied. Such analysis is called modal analysis, because it identifies
oscillating modes of the power system. In order to interpret the results of modal analysis to
power system requirements on damping of its oscillation modes have to be set. Oscillations in
power systems are required to be damped in order to maintain stability. So far there are no
published rules for the governance of Ethiopian power system to prescribe the damping
32
performance standard. However, based on the experience of other countries the
electromechanical modes of oscillation should have a time constant of about 12 seconds and
this corresponds to a damping constant of about -0.083Neper/s. Another rule of thumb is to
require a damping ratio of 5% which gives a very small safety margin. Often a damping ratio
of 15% is required in large interconnected systems and the same is applied in this thesis [16].
3.5.4 Small signal performance measures
Small signal performance measures provide a means of evaluating the response of the close
loop excitation control system to incremental changes in system conditions. Small signal
performance can be expressed in terms in of performance indices used in feedback control
systems [1].
i) Indices associated with time response

 Settling time ≤ 10 sec
 Over shoot ≤ 15%
 Rise time ≤ 0.5 sec
 Maximum peak (Mp) = 1.1-1.6
ii) Indices associated with frequency response
 Low frequency gain
 Cross over frequency
 Phase and gain margin
Gain margin ≥ 6 DB
Phase margin ≥ 40 0
Over shoot = 5 – 15%
3.5.5 The alternator state space model
The SM-IB system can be considered as a theoretical simple system that allows studying the
electromechanical interaction between a single generator and the power system. It is not useful
for studies of large power systems but it helps to understand the effect of the field, damper
circuits and the excitation system in the dynamic response of a single generator. The SM-IB
system model is also the base to analyze and to tune the PSS controller to enhance the dynamic
stability of the generator through the excitation control system. The linearized model will be a
33
suitable model for PSS tuning while the complete one will allow testing the results reached
from the PSS tuning process and from the application of other control structures to damp
power oscillations in the power system [1]. The model which was used for the design of the
final PSS consists of a single-machine infinite bus.
U
EXCITER GENE
RATO
Pe Transmission Line
- Infinite bus
AVR
Vref+
Figure 3.8 Excitation system control models
The voltage regulator controls the input u to the excitation system which provides the field
voltage so as to maintain the generator terminal voltage Vterm at a desired value Vref and
consider the state-space representation of the figure 3.8 system [18] as follows:-
There are 7 state variables, 1 input variable and 3 output variables y.
Where state variables x= [δ ω Eq‟ ψd E‟d ψ q Vr] T

Output variables y = [Vterm ω Pe] T
Input variable u= Vref
Where, δ= rotor angle in radian.
ω= angular frequency in radian/sec.
Ψd, Ed‟= direct axis flux and field.
Ψq, E‟q= quadrature axis flux and field
Vterm= terminal voltage
Pe= Power delivered to the infinite bus.
34
The state space equations are:-
Δx = A. Δx +B. Δu
Δy = C. Δx
Here, the matrices A, B depends on a wide range of system parameters and operating
conditions [17,16].The complete state-space model, including the PSS, matrix coefficient
expansions for the system shown in the above can be found in Appendix -1.
3.6 Multi-machine power system analysis
Analysis of practical power system involves the simultaneous solution of equations consisting
of synchronous machines and the associated excitation system and prime movers,
interconnecting transmission network, static and dynamic load (motor loads), and other
devices such as HVDC converters, static Var compensators. The dynamics of the machine
rotor circuits, excitation systems, prime mover and other devices are represented by
differential equations. The result is that the complete system model consists of a large number
of ordinary differential and algebraic equations [1].
In a real-time power system, many loads are connected to many generators located in different
regions (areas). This may be assumed as extended power system which can be divided into a
number of load frequency control areas interconnected by means of tie lines.
35
Figure 3.9 single line diagrams of 4 machines and 10 buses System
The system contains ten buses and two areas, connected by a weak tie between bus 9 and 10.
Each area consists of two generators, each having a rating of 900 MVA and 20 kV. Inertia
constant of area-1 generators is 6.5sec and of area-2 generators is 6.175sec.Totally two loads
are applied to the system at bus 9 and 10. For the comparisons of the obtained results with that
available in literature, the frequency of operation is chosen as 60 Hz. In these areas load
changes and abnormal conditions lead to mismatches in frequency and scheduled power
interchanges through tie line between areas. These mismatches have to be corrected by
Governor Control, which is defined as the regulation of the power output of generators within
a prescribed area.
3.6.1 State space model of multi-machine system
The machine equations (for kth machine) are
PÉqk = [-Éqk+ (Xdk - Xʹdk) idk +Efdk]
Pδκ =ωв (Smk -smko),
pSmk = , (Smk -smko) + pmk - pek]
36
This model is assumed for synchronous machines by neglecting some damper windings. The
state space model of a 4-machine 10 buses system as shown in Fig.3.9 can be obtained using
machine data, line data and load flow data as given in [1] as:
ẋ = [A] X + [B] (∆Vref +∆Vs)
y =[C]x
Te
Network
Mechanical
Tm
system
Sm Ẻq id, iq
Vref
AVR
Vs Efd
Field
winding
PSS
Figure 3.10 Block diagram including excitation system and PSS
Where
x = [x1, x2, x3, x4] T, and y = [y1, y2, y3, y4] T
Xk (k=1, 4) denotes the states of kth machine, and YK (k=1, 4) denotes the output of
the kth machine.
37
The elements (sub matrices of 4 4) of matrix are depend on the machine and
network parameters. Thus I have
11 1 1 1 1
1
A= , B =
1
[ 1 ] [ ]
1 1 1
C= , (∆Vref +∆Vs) =
[ ] [ ]
To test the performance of the proposed stabilizer, simulation studies were performed on two
systems: a one-machine-infinite-bus system and a two-area, 4-machine, 10-bus system, and
seven states are used to model each generator [1].
( 𝝎 )
The sampling period used in all cases was 0.01 sec and Power System Toolbox has been used
for this simulation study.
The block diagram including excitation system and power system stabilizers is given in the
above Fig. 3.10
38
CHAPTER FOUR
CONVENTIONAL POWER SYSTEM STABILIZER DESIGN
4.1 Introduction
In this chapter, a conventional power system stabilizer is designed on the basis of the block
diagram representation of the system introduced in chapter two. Here the design procedure is
performed in the frequency response, Root Locus and state spaces approach methods.
Conventional power system stabilizers (CPSSs) are basically designed on the basis of a linear
model for the power system. The power system is first linearized around a specific operating
point of the system. Then, assuming that disturbances are small such that the linear model
remains valid, the CPSS is designed. Therefore, a CPSS is most useful for preserving the
dynamic stability of the power system [2].
The schematic below represents different methods of Power System Stabilizer design Review:
Power system
stabilizer design
Conventional Non-conventional
methods methods
Analog Digital Non-linear Adaptive
Linear Non-linear Analog Digital Analog Digital
Figure 4.1 Methods of PSS design
39
4.2 Design of Automatic voltage Regulator (AVR)
Automatic Voltage Regulator (AVR) is a device which automatically controls and stabilizes
the terminal voltage output of the generator using feedback. It is designed in such a manner to
achieve good damped response even if the generator is disconnected from the network.
However a static AVR that uses constant feedback gains has a destabilizing effect on
electromechanical oscillations. Whenever the terminal voltage tends to fall due to load, the
drop is approximately fed back to increase the excitation and vice-versa such that the required
terminal voltage is maintained in the output terminals. Automatic Voltage Regulator (AVR) is
implemented in feed forward open-loop model itself. Then used the designed AVR model and
implement the PSS in a negative feedback loop. The transfer function of the filters and limiter
are fixed while that of the dynamic compensator varies from method to method. Each method
has its own unique set of algorithms and produces a different result [19].
Figure 4.2 Simulink diagram of one synchronous machine with AVR simulated in Mat lab
40
Figure 4.3 terminal voltages (VT) without pss
Figure 4.4 Deltas (Load angle) without pss
41
Figure 4.5 Electrical torques without pss
The general conventional power system stabilizer designed here adopted has been grouped
under three heads:
 Root-Locus approach (PID controlled Lead/Lag compensator)

 Frequency response approach (Lead-Lead compensator)
 State-Space approach (Observer based Controllers)
4.2.1 Automatic voltage regulator (AVR) design using root locus method
The root locus indicates the achievable closed‐loop pole locations of a system as a parameter
(usually the controller gain) varies from zero to infinity. For a given plant it may or may not be
possible to implement a simple proportional controller (i.e., select again that specifies closed‐
loop pole locations along the root locus) to achieve the specified performance constraints. In
fact, in most cases, it will not be possible. When this occurs, it is the control engineer‟s job to
select a controller structure (again and numbers of poles and zeros of a controller transfer
function) and the respective controller parameters (values for the gain and poles and zeros) to
change the shape of the root locus so that for some values of the controller gain, the dominant
second order closed‐loop poles lie within the performance region. In controller design there are
multiple possible solutions, some better than others. It is possible to have multiple designs that
42
satisfy the given performance constraints, but practical implementation issues and cost could
be prohibitive for some designs [9].
Enter the percent overshoot and settling time constraints in mat lab code. Remember that these
constraints are based on a first-order system step response and for higher order systems are
predicated on the assumption of first order dominance of the closed‐loop system poles.
Therefore, these design constraints are guidelines and you may have to refine your design to
stay further within these constraints to meet the performance specifications. Using the matrices
A, B depends on a wide range of system parameters and operating conditions [17,16]. The
complete state-space model, including the PSS, matrix coefficient expansions for the system
shown in the above can be found in Appendix -1.The given 7th order system has a mixed
proportion of dominant and nominal poles. Controlling the dominant poles stabilizes the
disturbances. In this case, a pair of dominant conjugate pole pairs is taken the AVR is
analyzed. The Root locus plot of the VR loop is made using only Proportional Controller.
Figure 4.6 Root locus of Voltage Regulator forward loop showing dominant poles
43
4.2.2 AVR design using frequency response method
The frequency response design method involves the use of bode-diagrams to measure the
phase and gain margin of the system and compensating the phase by using a lag controller for
AVR and lead controller for PSS. First, to observe bode plot response of the open-loop Power
system. In the frequency response design, plotting the frequency response from u to Vterm and
also shows the gain and phase margin. From this to find that:
 Gain margin Gm. = 35dB

 Phase margin Pm = INF.
 DC gain= -2.69dB (0.78)
Figure 4.7 Frequency response of an uncompensated AVR Loop
The design specification require the DC gain > 200 (=46dB) & phase margin > 90º.
44
Required gain Kc=10^ ((200+0.78)/20) = 273 .New gain crossover frequency= 5rad/s. This
requires a lag-compensator for AVR, having transfer function:
V(s) = Ki 0 1
Where Ki = 0 1 p=
Now, the lag required at 5rad/sec is -18dB. Hence,
20log = -18 i.e. ᵦ = 8
To choose the corner frequency f=0.1, to make the system faster. So, z =0.1.Hence
p=0.1/8 = 0.0125, Kl= 273/8 = 34.125.
Thus the final AVR is:
V(s) = 34.1250 1
Since the purpose of the lead compensator is to move the phase curve upwards in order to
satisfy the phase margin specification, which needs to determine how much positive phase
shift is required. The first step in this determination is to evaluate the phase margin of the
system. The uncompensated phase margin is:
PM uncompensated = 180O+∠G (jωx)
There are no other poles or zeros; the phase shift of G (jω) is −180◦ at all frequencies.
Therefore, the uncompensated phase margin is PM uncompensated = 180◦ + (−180◦) = 0◦.
The frequency response of the uncompensated and the lag compensated system VR are shown
below:
45
Figure 4.8 Comparison of frequency response with and without VR loop
Figure 4.9 Step responses without AVR

46
Figure 4.10 Step response of the lag compensated VR
Rise time = 0.5sec.

Maximum overshoot Mp= 7.4%
Settling time ts is more than 10 sec.
4.2.3 AVR design using state-space method
The state space design involve designing full state observers using pole placement to measure
the states and designing the controller such that the closed loop poles lie in the desired place.
One can obtain the 1 -input 1-output model of SMIB from Vref to Vterm. Hence, one gets A1
(7*7 matrix), B1 (7*1 vector), C1 (1*7 matrix), and D1 (1*1) by using mat lab M-files the
results as given in Appendix-1. Then find the open loop poles of this system.
(-114.33, -35.37, -26.72, -0.48±9.33j, -3.07, -0.1054)
Using Kc=acker (A1, B1, modified poles), to design full-order observer to measure states; and
choose observer dominant pole to be far from the jw axis, hence it decays very fast.it to be
47
taken -8.0+0.0j and leave other poles unchanged. Finally find the state space representation
and the transfer function:
The 7th order observer-controller transfer function.
405.6 s^6 + 7.32e04 s^5 + 3.588e06 s^4 + 6.357e07 s^3 + 4.906e08 s^2 + 4.832e09 s
+ 1.177e10
-------------------------------------------------------------------------------------------------------------
S^7 + 193.6 s^6 + 1.122e04 s^5 + 2.722e05 s^4 + 3.259e06 s^3 + 2.811e07 s^2
+ 1.849e08 s + 3.201e08
Then minimize the order of this controller to 1storder by approximate pole-zero cancellations
using the mat lab M-files.
Poles of observer-controller Zeros of observer-controller
- 114.23 -114.32
-35.87 -35.36
-26.71 -26.71
-13.14
-0.613+9.58j -0.48+9.6j
-0.613+9.58j -0.48 - 9.6j
-2.42 -3.02
Thus, this is left with a single pole -13.14. So, the VR is given by:
V(s) =34.125 0 1
48
Figure 4.11 the frequency response of 1st and 7th order controller for AVR
4.3 Design of power system stabilizer (PSS)
Automatic Voltage Regulator (AVR) is implemented in feed forward open-loop model itself.
Then used the designed AVR model and implement the PSS in a negative feedback loop. The
PSS had initially the torsional filter, washout filter, and limiter. All components are doing the
designing of the dynamic compensator for effective implementation of the system. The
transfer function of the filters and limiter are fixed while that of the dynamic compensator
varies from method to method. Each method has its own unique set of algorithms and
produces a different result. After introducing properly tuned PSS in AVR loop, it can be
visualized from Fig.4.13, Fig.4.14 and Fig.4.15 of terminal voltage, rotor angle deviation and
Electrical torque respectively, that the system oscillations have been sufficiently damped out in
short period of time [20].
49
Figure 4.12 Simulink diagrams of PSS with AVR simulated in Mat lab
50
Figure 4.13 terminal voltages (VT) with pss
Figure 4.14 Deltas or rotor angle with pss
51
Figure 4.15 Electrical torques with pss
4.3.1 Power system stabilizer (PSS) design using root locus method
The root locus indicates the achievable closed‐loop pole locations of a system as a parameter
(usually the controller gain) varies from zero to infinity. For a given plant it may or may not be
possible to implement a simple proportional controller (i.e., select again that specifies closed‐
loop pole locations along the root locus) to achieve the specified performance constraints. In
fact, in most cases, it will not be possible.
The design system is able to follow the step input by the introduction of the AVR; but due to
the PI controller of the AVR, the swing mode (dominant complex poles) becomes unstable and
oscillations are introduced in the output Vterm. Now, to reduce the oscillations, this to
introduce a feedback loop involving the swing in rotor angular speed (∆ω) as input to the
Power system stabilizer (PSS) loop.
52
Figure 4.16 Root locus of PSS loop showing the dominant complex poles
The dominant complex poles are at (-0.48+9.33i, -0.48-9.33i), then to find the angle of
departure (Φp) from the pole using MATLAB.
Φp = 43.2798 deg. Based on this angle of departure to design the lead-lead compensator:
P(s) = K 0 . /1 0 . /1 such that Φp=180º for perfect damping. Hence to add
the angle of 137º this cannot be done using a single lead compensator So using two lead
compensators in series each adding an angle of 68.5º. K is chosen from the root locus plot of
the final PSS loop such that damping ratio ζ > 15%. After the design to get the value:
z= 3.5 p= 24 Kα= 13.74 K= 0.42, then the final lead-lead compensator is given by:
P(s) = 0.42 0 . /1 0 . /1
So that to implement this PSS and close the loop and simulate the response. The root-locus
plot of the final PSS loop and the comparison of responses are given below:
53
Figure 4.17 Root-locus of the final PSS loop showing Φp ~180º for dominant poles
4.3.2 Power system stabilizer (PSS) design using frequency response method
The introduction of the voltage regulator eliminates the steady state error and makes the
system much faster. But it also introduces low frequency oscillations in the system. Hence to
design the Power system stabilizer loop taking input as the perturbation in rotor angular speed
(∆ω). The state-space model from Vref to ω with the regulation loop closed. As given in figure
4.18, to isolate the path Q(s) = effect of speed on electric torque due to machine dynamics and
find Aω matrix from the main matrix A. Thus are A33 (5*5 matrix), a32 (5*1 vector), a23
(1*5 matrix). Using mat lab code programing the matrix results has been placed in an
appendix.
54
Figure 4.18 Frequency response of the Damping Loop
From the figure 4.18 to obtain that Phase at 2rad/sec = -40º while Phase at 20 rad/sec = -102º.
The required increasing phase of the feedback loop which will add pure damping to the
dominant poles.
Hence require a lead compensator of the given by
P(s) = K 0 . /1 0 . /1 where Ka 0 1
The parameters of the transfer function were obtained by the constraints, addition of phase of
40º at 2 rad/sec, 65º at 12 rad/sec and 102º at 20 rad/sec. The Maximum phase addition Φm is
at 20 rad/sec =102º. This is too large for a single lead compensator as shown in figure 4. 18.
55
Figure 4.19 Maximum phase addition Φm Vs alpha α
From the above figure 4.19, to observe that for Φm > 60º, α is too small. Hence using two
identical lead-compensators in series each compensating is 51º.
From the relation:-

1
Sin Φm = , a= 0.1254
1
1
Ka = =8, T= = 0.141, z = =7.08 , P=
𝝎
From the root locus plot, to get K for ζ >12%, K=6.
P(s) = 6 0 . /1 0 . /1
Then to implement this PSS and close the loop and simulate the resultant model, then find the
step response and the rise time and maximum overshoot of the compensated system.
56
Figure 4.20 Root locus plot of the PSS loop showing the dominant poles
Figure 4.21 Step response of the final system with PSS loop
The final model clearly shows that the system with PSS provides very good performance in
terms of steady state error and settling time.
57
4.3.3 PSS design using state-space method
The state space design involve designing full state observers using pole placement to measure
the states and designing the controller such that the closed loop poles lie in the desired place.
Introducing PSS makes the system transfer function to 11th order. Transfer function of the 11th
order controller of the mat lab code result is
-25.76 s^10 - 9696 s^9 - 1.347e06 s^8 - 8.982e07 s^7 - 3.325e09 s^6 - 7.237e10 s^5- 9.002e11
s^4 - 5.494e12 s^3 - 1.087e13 s^2 - 2.063e12 s - 1.03e11
-----------------------------------------------------------------------------------------------------
s^11 + 260 s^10 + 2.587e04 s^9 + 1.387e06 s^8 + 4.585e07 s^7 + 9.924e08 s^6
+ 1.445e10 s^5 + 1.409e11 s^4 + 8.281e11 s^3 + 1.781e12 s^2 + 3.046e11 s + 1.523e10
From the root locus plot, to find the dominant complex pole is (-0.48 ± 9.4j).
For faster response, shift it to: (-1.5 ± 9.4j), leaving all other poles unchanged. Using
MATLAB code, which gets the controller gain matrix Kc=acker ( , modified poles).For
the observer design, choose the poles as (-1.5 ± 9.4j) so that it decays faster.
= place (Ag', Cg', poles_obs)'.Thus to get the 11th order observer- controller as:
Ao = A1- ( *C1) – (B1*Kc) Bo = Co = Kc Do= 0
By using mat lab code to minimize this PSS from 11th to 5th order by approximate pole-zero
cancellations
-20 s^4 - 4120 s^3 - 51580 s^2 - 10440 s – 540

--------------------------------------------------------
S^5 + 39 s^4 + 507 s^3 + 6183 s^2 + 1112 s + 57
Finally, in the state-space method, an exact 11th order controller is derived from a full order
state-observer. This is highly impractical and expensive, and thus to minimize the order of the
system by approximate pole-zero cancellations which make it a viable and alternate solution
leading to precision. From the figure below the 5th order PSS manifests a pure damping at the
dominant pole as the angle of departure is 180º. The gain for ζ=12% is found to be 0.84.
58
Figure 4.22 Root locus plot of the damping (PSS) loop with 5th order PSS implemented
Figure 4.23 Frequency response comparisons of 11th order and 5th order pss for pure damping
59
4.4 conclusions
The design of Automatic Voltage Regulator (AVR) includes a constant voltage source ( here a
step input) implementing the regulation loop in each case and conclude that even when the
Rise Time(tr) , Peak Overshoot( Mp) , Settling time ( ts) and Steady-State Error ess constraints
are satisfied, meanwhile the system is not perfectly damped and there are oscillations in it.
Then used the designed AVR model and implement the PSS in negative feedback loop and
After introducing properly tuned PSS in AVR loop, it can be visualized from Fig.4.13,
Fig.4.14 and Fig.4.15 of terminal voltage, rotor angle deviation and Electrical torque
respectively, that the system oscillations have been sufficiently damped out in short period of
time (4.5 sec, 12 sec, 17 sec) respectively.
The Conventional power system stabilizer (CPSS) damps the low-frequency oscillations in the
shaft speed of a synchronous machine. Since the design is on the basis of a block diagram of
the system derived for a specific operating point, the CPSS has the best response for this
operating point. If the operating point of the system changes, the performance of the CPSS will
degrade. The figure below shows that the comparison of the 1st and 7th order freq. responses‟.
Figure 4.24 the frequency response of 1st and 7th order controller for AVR
60
In this thesis the design is performed by increasing the system order from first order to higher
(11th) order the constraints are to meet the performance specifications.
Gain margin ≥ 6 DB
Phase margin ≥ 40 0
Over shoot = 5 – 15%
Figure 4.25 Frequency response comparisons of 11th order and 5th order pss for pure damping
Last but not the least, in the state-space method; an exact 11th order controller is derived from
a full order state-observer. This is highly expensive and impractical, and so to minimize the
order of the system by approximate pole-zero cancellations which make it a viable and
alternate solution leading to precision.
61
CHAPTER FIVE
DESIGN OFADAPTIVE NEURO-FUZZY BASED POWER SYSTEM

STABILIZERS
5.1 Introduction
The high gain continuously acting automatic voltage regulators help to improve the dynamic
limits of power systems. Moreover, it‟s introduced negative damping, and so makes the system
unstable, especially in large, weakly coupled systems. To overcome this problem, a
supplementary stabilizing signal is introduced in the excitation system. Power System
Stabilizers (PSSs) offer this supplementary stabilizing signal and are widely used to suppress
the generator electromechanical oscillations and enhance the overall stability of power
systems. A conventional PSS is based on the use of a transfer function designed for a linear
model representing the generator at a certain operating point. The Conventional PSSs are
widely used and have improved the stability of power systems. However, because a
conventional PSS is designed for a particular operating point for which the linearized transfer
function model is obtained, it does not offer satisfactory results over a wide range of operating
conditions.
Power systems are highly nonlinear, and random disturbances like load changes and time-
varying operating conditions make precise modeling of large power systems in real time very
difficult. It is this nonlinearity and imprecision in modeling lead to the combination (hybrid) of
fuzzy logic and neural network control which is a powerful tool in controlling systems which
are not precisely characterized.
The design of an adaptive neural-fuzzy inference system [21] based PSS (ANFIS PSS)
combined the physical perspectives of the power system oscillations, system identification,
and the intelligent control theories (more specifically, fuzzy logic and neural network.) It uses
the post-disturbance value of the speed deviation and its derivative (obtained online) as inputs,
and the heuristic fuzzy control rules are based on the understanding of physical aspects of the
power system oscillations. The ANFIS PSS uses a Mamdani‐type fuzzy logic controller whose
membership functions and consequences are also tuned on-line by use of a neural network.
62
5.2 Fuzzy logic Control system
Fuzzy logic system (FLS) is a convenient way to map input space into output space. This
means for the given conditions or sample inputs of the system, the designed fuzzy logic system
would lead to the crisp output, no matter how non‐linear or stochastic in nature. In order to
understand the working properly, it should be noted that FLS is not the only way of solving
problems. If it replaced with the black box, there may be a number of things in that box
solving the problem, such as linear systems, expert system, neural network, differential
equations, interpolating multidimensional look-up tables, etc. Now the equation arises to what
sort of problems fuzzy logic should be applied and to which problem it‟s not beneficial. The
answer is, when the problem is simple enough and a crisp mathematical expression solves the
problem more effectively, fuzzy logic should be avoided. When the problem is very complex,
ambiguous and information available is imprecise, fuzzy reasoning provides a way to
understand system behavior by allowing interpolation between input and output situations.
Fuzzy Logic Controllers are very useful when an exact mathematical model of the plant is not
available; however, experienced human operators are available for providing qualitative rules
to control the system. Fuzzy logic, which is the logic on which fuzzy logic control is based, is
much closer in spirit to human thinking and natural language than the traditional logic systems.
Basically, it provides an effective mean of capturing the approximate, inexact nature of our
knowledge about the real world. Viewed in this perspective, the essential part of the fuzzy
logic controller (FLC) is a set of linguistic control rules related to dual concepts of fuzzy
implication and the compositional rule of inference. In essence, the FLC provides an algorithm
which can convert the linguistic control strategy based on expert knowledge into an automatic
control strategy. The methodology of the FLC appears very useful when the processes are too
complex for analysis by conventional quantitative techniques.
Fuzzy logic is a derivative from classical Boolean logic and implements soft linguistic
variables on a continuous range of truth values to be defined by conventional binary i.e. [0, 1].
It can often be considered a subset of conventional set theory. The fuzzy logic is capable to
handle approximate information in a systematic way and it is suited for controlling nonlinear
systems and for modeling complex systems where an inexact model exists or systems where
ambiguity or vagueness is common. The importance of fuzzy logic derives from the fact that
63
most modes of human reasoning and especially common sense reasoning are approximate in
nature. In doing so, the fuzzy logic approach allows the designer to handle efficiently very
complex closed-loop control problems. There are many artificial intelligence techniques that
have been employed in modern power systems, but fuzzy logic has emerged as the powerful
tool for solving challenging problems. As compared to the conventional PSS, the Fuzzy Logic
Controller (FLC) has some advantages such as [22]:
 A simpler and faster methodology.

 It does not need any exact system mathematical model.
 It can handle nonlinearity of arbitrary complexity.
 It reduced the design development cycle.
 It is based on the linguistic rules with an IF-THEN general structure, which is the basis
of human logic.
 It is more robust than conventional nonlinear controllers.
 It reduces hardware cost.
 Simple to implement
5. 2.1 Fuzzy sets
The Fuzzy set, which the name implies, that a set without a crisp is a boundary. The transition
from belonging to a set to not belong to a set is gradual, and this smooth transition is
characterized by membership functions. The fuzzy set theory is based on fuzzy logic, where a
particular object has a degree of membership in a given set that may be anywhere in the range
of 0 to 1 [23].
5. 2.2 Membership functions
A membership function is a curve that defines how the values of a fuzzy variable in a certain
region are mapped to a membership value µ (or degree of membership) between 0 and 1. The
MF maps each element of X to a membership degree between 0 and 1 (included). Obviously,
the definition of a fuzzy set is a simple extension of the definition of a classical (crisp) set in
which the characteristic function is permitted to have any value between 0 and 1. If the value
of the membership function is restricted to either 0 or 1, then A is reduced to a classical set.
For clarity, the researcher shall also refer to classical sets as ordinary sets, crisp (normalized)
64
sets, non-fuzzy sets, or just sets. Usually, X is referred to as the universe of discourse, or
simply the universe, and it may consist of discrete (ordered or non-ordered) objects or it can be
a continuous space. In practice, when the universe of discourse X is a continuous space,
usually partition it into several fuzzy sets whose MFs cover X in a more or less uniform
manner. These fuzzy sets, which usually carry names that conform to adjectives appearing in
our daily linguistic usage, such as ”large,” ”medium,” or ”small,” are called linguistic values
or linguistic labels. Thus, the universe of discourse X is often called the linguistic variable.
The fuzzy membership not only provides for a meaningful and powerful representation of the
measurement of uncertainties but also provides the meaningful representation of vague
concepts expressed in natural language. If X is a collection of objects denoted generically by x,
then a fuzzy set in X is defined as a set of ordered pairs.
A= *( ( )) + (5.1)
Where
µA(x) is the membership function for the set A. There exist different shapes of membership
functions. The shapes could be triangular, trapezoidal, curved or their variations. In the real-
time implementation, both the triangular membership function and trapezoidal membership
function have been used extensively due to their simple formulas and computational
efficiency. These two membership functions can have symmetrical or unsymmetrical shape. In
this thesis to implement the triangular membership function has been used due to its
computational efficiency.
5.3 Fuzzy Logic Controller
Fuzzy Logic Controller comprises of four principal components. These are:
 a Fuzzification interface
 a knowledge base
 decision making logic
 Defuzzification interface.
65
KNOWLEDGE
BASE
FUZZIFICATION
INTERFACE DEFUZZIFICATION
INTERFACE
DECISION
Fuzzy Xk MAKING LOGIC UK Fuzzy
CONTROLLED
Process output state Xk SYSTEM (PROCESS)
UK Actual control
Non fuzzy Non fuzzy
Figure 5.1 the basic configuration of a fuzzy logic control
Function of Fuzzification interface:
 Measures the values of input variables (Process output).

 Performs a scale mapping that transfers the range of values of input variables into
corresponding universe of discourse.
 Performs the function of Fuzzification that converts data into suitable linguistic values
which may be viewed as labels of fuzzy set.
Function of knowledge base
 The database provides necessary definitions, which are used to define linguistic control
rules and fuzzy data manipulation in FLC.
 The rule base characterizes control goals by means of a set of linguistic control rules.
66
Function of Defuzzification interface
 a scale mapping, which converts the range of value of output variables into
corresponding universes of discourse,
 Defuzzification, which yields a non-fuzzy control action from an inferred fuzzy control
action.
The function of decision making logic
It is the kernel of an Fuzzy logic controller; it has the capability of simulating human decision
making based on fuzzy concepts and of inferring fuzzy control actions employing fuzzy
implication and the rules of inference in fuzzy logic [24].
5.3.1 Linguistic variable and rules bases
Linguistic variables are values defined by fuzzy sets. A linguistic variable such as „High
Speed‟ for example could consist of numbers that are equal to or between 50km/h and 80km/h.
The conditional statements that make up the rules that govern fuzzy logic behavior use these
linguistic variables and have an if-then syntax. These if-then rules are what make up fuzzy rule
bases. A sample if-then rule where the variables A and B represent linguistic variables could
be: If x is A then y is B the statement is understood to have both a premise, if „x is A‟, and a
conclusion, then „y is B‟. The premise is also known as the antecedent returns a single number
between 0 and 1 whereas the conclusion is also known as the consequent assigns the fuzzy set
B to the output variable y. Another way of writing this rule using the symbols of assignment
(=) and equivalence (= =) is: if x = = A then y = B An if-then rule can contain multiple
premises or antecedents. For example, if velocity is high and a road is wet and brakes are poor
then….Similarly, the consequent of a rule may contain multiple parts. If the temperature is
very hot then the fan is on and throughput is reduced interpreting these rules involves a
number of distinct steps: Firstly, the inputs must be fuzzified; to do this all fuzzy statements in
the premise are resolved to a degree of membership between 0 and 1. This can be thought of as
the degree of support for the rule. At a working level, this means that if the antecedent is true
to some degree of membership, then the consequent is also true to that same degree. Secondly,
fuzzy operators are applied for antecedents with multiple parts to yield a single number
between 0 and 1. Again this is the degree of support for the rule. Thirdly, the result is applied
to the consequent, this step is also known as an implication. The degree of support for the
67
entire rule is used to shape the output fuzzy set. The outputs of fuzzy sets from each rule are
aggregated into a single output fuzzy set. This final set is evaluated (or defuzzified) to yield a
single number. Immediate advantages of this approach become apparent. Fuzzy sets can be
combined using fuzzy rules to define system behavior and thus complex non-linear systems
can be expressed linguistically. In fact, as will be shown later, rule tables can represent fuzzy
controllers. The process of fuzzifying a single crisp input, applying fuzzy operators and then
defuzzifying to produce a single crisp output is known as fuzzy inference [23].
5.3.2 Scaling factors
To choose membership functions, first of all, one needs to consider the universe of discourse
for all the linguistic variables, applied to the formulation of the rules. To specify the universe
of discourse, one must first determine the applicable range for a characteristic variable in the
context of the system designed. The range you select should be carefully considered. For
example, if you specify a range which is too small, regularly occurring data will be off the
scale that may impact on an overall system performance. Conversely, if the universe for the
input is too large, a temptation will often be to have wide membership functions on the right or
left to capture the extreme input values. It is usually desirable and often necessary to scale, or
normalize, the universe of discourse of an input/output variable. Normalization means
applying the standard range of [–1, +1] for the universe of discourse both for the inputs and the
outputs. In the case of the normalized universe, an appropriate choice of specific operating
areas requires scaling factors. An input scaling factor transforms a crisp input into a
normalized input in order to keep its value within the universe. An output scaling factor
provides a transformation of the defuzzified crisp output from the normalized universe of the
controller output into an actual physical output. The role of a right choice of input scaling
factors is evidently shown by the fact that if your choice is bad, the actual operating area of the
inputs will be transformed into a very narrow subset of the normalized universe or some values
of the inputs will be saturated. One can see when the output is scaled; the gain factor of the
controller is scaled. The choice of the output scaling factor affects the closed loop gain, which
as any control engineer knows, influences the system stability. The behavior of the system
controlled depends on the choice of the normalized transfer characteristics (control surface) of
the controller. In the case of a predefined rules table, the control surface is determined by the
68
shape and location of the input and output membership functions. To start tuning recommends
the following priority list:-
 The output denormalisation factor has the most influence on stability and oscillation
tendency. Because of its strong impact on stability, this factor is assigned to the first
priority in the design process.
 Input scaling factors have the most influence on basic sensitivity of the controller with
respect to the optimal choice of the operating areas of the input signals. Therefore,
input scaling factors are assigned the second priority.
 The shape and location of input and output membership functions may influence
positively or negatively the behavior of the controlled system in different areas of the
state space provided that the operating areas of the signals are optimally chosen.
Therefore, this aspect is the third priority [21].
5.4 Artificial Neural Network
It is well known that biological systems can perform complex tasks without recourse to
explicit quantitative operations. In particular, biological organisms are capable of learning
gradually over time. This learning capability reflects the ability of biological neurons to learn
through exposure to external stimuli and to generalize. Such properties of nervous systems
make them attractive as computation models that can be designed to process complex data. For
example, the learning capability of biological organisms from examples suggests possibilities
for machine learning [13].
5.4.1 Basic Elements
Neurons are the basis of the neural networks. A neuron is an information-processing unit that
is fundamental to the operation of a neural network. Figure 5.2 shows the model for a neuron.
There are three basic elements of the neuron model, as described below:
A set of synapses or connecting links, each of which is characterized by a weight. A

signal xj at the input of the synapse j connected to neuron k is multiplied by synaptic
weight wkj.
An adder for summing the input signals, weighted by the respective synapses.
69
An activation function for limiting the amplitude of the output of a neuron. This limit
usually is in the unit interval [0, 1] or alternatively [-1, 1].
This model also includes an externally applied threshold θk that has the effect of lowering the
net input of the activation function.
Wk1
X1 UK YK
𝛗
Input
WK2
𝝨 output
Signals X2
. .
. . Threshold
Wkp
Xp
Figure 5.2 Non-linear model of a neuron
5.4.2 Neural control
Neural control refers both to a methodology in which the controller itself is a neural network
and to a methodology in which controllers are designed based on a neural network model of
the plant. These two basically different approaches for implementing neural networks in
control are referred to as direct and indirect design methods which have seen that fuzzy control
is a control method relying on perception-based information expressed in fuzzy logic. This is
the case where the available data is in the form of a collection of linguistic. If . . . then . . . rule.
In other Words, fuzzy control is a mathematical method for implementing control strategies
expressed in a natural language. This situation arises mostly in the control of complex systems,
a situation that human operators handle well and for which natural language is an appropriate
means for describing control strategies. As its name indicates, neural control refers to another
control method when available data are in the form of measurements (observed numerical
data) of the plant‟s behavior. This is the case where information is only in the form of system
behavior, either of the real plant or of its simulated model, expressed as input-output
measurements [25].
70
5.5 Neuro-Fuzzy Systems
Both neural networks and fuzzy system are motivated by imitating human reasoning process
which utilizes human expertise. In fuzzy systems, relationships are represented explicitly in the
form of the if-then rules. In neural networks, the relations are not explicitly given but are
encoded in the networks and parameters designed. The Neuro-fuzzy systems combine
semantic transparency of rule-based fuzzy systems with a learning capability of neural
networks [25].
5.5.1 Adaptive network fuzzy inference systems
To illustrate the use of neural networks for fuzzy inference, to present some successful
adaptive neural network fuzzy inference systems, along with training algorithms known as
ANFIS. These structures, also known as adaptive Neuro-fuzzy inference systems or adaptive
network fuzzy inference systems, were proposed by Jang. It should be noted that similar
structures were also proposed independently by Lin and Lee and Wang and Mendel. These
structures are used for control and for many other applications.
5.5.2 Neuro – fuzzy controller
The neural predictive controller can be extended with Neuro-fuzzy controller, which connected
in parallel Figure (5.3). Neuro-fuzzy systems, which combine neural networks and fuzzy logic,
have recently gained a lot of interest in research and application. A specific approach to
Neuro-fuzzy development is the ANFIS (Adaptive Network-based Fuzzy Inference System).
ANFIS uses a feed forward network to search for fuzzy decision rules that perform well on a
given task. By using a given input-output data set; ANFIS creates a Fuzzy Inference System
for which membership function parameters are adjusted using a combination of a back
propagation and least square method [11].
71
Control Algorithm
Model Objection
function
Yi u y
Reference Optimizer Process
Generator
+ E Fuzzy
controller
-
Figure 5.3 Neuro-Fuzzy Control scheme
5.5.3 Adaptive network fuzzy inference systems as an estimator
Adaptive Network Fuzzy Inference System (ANFIS) can be used for the estimation of some
dependent variables in the chemical process. The designed ANFIS estimator is used to infer
the compositions from the measurable tray temperatures distillation column. In the estimator
design process, different ANFIS structure is constructed and trained to find the architecture
that gives the best performance as an estimator. A first step to design an estimator, training
data sets should be generated to train the estimator networks. These datasets consist of
estimator inputs and the desired output values. They are produced from the process input-
output data. Since ANFIS is a data processing method, it is important that the input-output data
must be within the sufficient operational range of including the maximum and minimum
values for both inputs, and output variables of the system. If this is not provided, estimator
performance cannot be guaranteed and thus the designed estimator will not be accurate.
Having generated the training data, the estimators that have different architectures are trained
with the obtained data sets. The Performances of the trained estimators are evaluated with
model simulations and best estimator architecture is obtained. These simulations are made to
72
verify and to generalize the ANFIS structures. Verification is done to show how good the
estimator structure learned the given training data. This is carried out by simulating the column
models with specific initial process inputs used in obtaining training data sets. The
Generalization capabilities of the estimators are found with other simulations in which the
input process variables are in operational range but not used in training data formation [26].
5.6 Design of Fuzzy Logic Based Power system stabilizer
The basic structure of the fuzzy logic controller is shown in Figure 5.4. Here the inputs to the
fuzzy logic controller are the normalized values of error ‟e‟ and change of error ‟ce‟.
Normalization is done to limit the universe of discourse of the inputs between -1 to 1 such that
the controller can be successfully operated within a wide range of input variables. Here „Ke‟
and „Kce‟ are the normalization factors for error input and change of error input respectively.
For this fuzzy logic controller design, the normalization factors are taken as constants. The
output of the fuzzy logic controller is then multiplied with a gain „Ko‟ to give the appropriate
control signal „U‟. The output gain is also taken as a constant for this fuzzy logic controller.
The fuzzy controller used in power system stabilizer is normally a two input and a single
output component. The two inputs are changed in angular speed and rate of change of angular
speed whereas the output of the fuzzy logic controller is a voltage signal [27].
Ke
K0
1
E Ke
FUZZY
CONTROL U
1
+ Ce
Kce
Kce
Delay
Figure 5.4: Basic Structure of Fuzzy Logic Controller
73
5.6.1 Input/output Variables
The design starts with assigning the mapped variables inputs/output of the fuzzy logic
controller (FLC). The first input variable to the FLC is the generator speed deviation and the
second is acceleration. The output variable to the FLC is the voltage. After choosing proper
variables as input and output of the fuzzy controller, it is required to decide on the linguistic
variables. These variables transform the numerical values of the input of the fuzzy controller to
fuzzy quantities. The number of linguistic variables describing the fuzzy subsets of a variable
varies according to the application. Here seven linguistic variables for each of the input and
output variables are used to describe them. Table 5.1 shows the Membership functions for
fuzzy variables. The membership function maps the crisp values into fuzzy variables. The
triangular membership functions are used to define the degree of membership. Here for each
input variable, seven labels are defined namely, NB, NM, NS, ZE, PS, PM, and PB. Each
subset is associated with a Mamdani membership function to form a set of seven membership
functions for each fuzzy variable.
Table 5.1: Membership functions for fuzzy variables
NB Negative Big
NM Negative Medium
NS Negative Small
ZE Zero
PS Positive Small
PM Positive Medium
PB Positive Big
5.6.2 Membership function editor
The membership function editor shown in Figure (5.5) shares some features with the FIS
editor. The membership function editor is the tool that lets the programmer displays and edits
all of the membership functions associated with all inputs and output variables for entire fuzzy
inference system. To open the Membership Function Editor to work on a fuzzy inference
system that does not already exist in the workspace, there is not yet any membership functions
74
associated with the variables that the researcher have just defined with the FIS Editor. The
step below shows how to open the membership function editor which are shown in the Figure
(5.5):-
 Within the FIS editor windows, select Edit > Membership functions
 Within the FIS editor, double click the blue icon called „e‟
 At the command line, type mf edits.
Figure 5.5 FIS editor
75
Figure 5.6 Membership function for speed deviation
Figure 5.7 Membership function for acceleration

76
5.6.3 Rule editor
Rule editor are call up, by choose edit menu rules edit selected.
Figure 5.8 Rule editors
Based on the description of the input and output variable defined with the FIS Editors, the
Rule Editor allows to construct the rule statements automatically. From GUI as shown in
Figure (5.8):-
 Rules are creating by selecting an item in each input and output variable box and one
connection item and clicking add Rule. Then complete range for the variation of each
of the two controller inputs is represented by a 7×7 decision table, i.e. 49 rules.
 Rules are deleting by selecting the rule and clicking Delete Rule.
 Rules are editing by changing the selection in the variable box and clicking
Change Rule.
77
 To specify the weight of a rule by typing in a desired number between -1 and 1 in
Weight, but do not specify the weight, which is assumed to be unity (1). Each entity
shown in Table 5.2 represents a rule. The antecedent of each rule conjuncts speed
deviation (∆ω) and acceleration (∆ ̇ ) fuzzy set values.
Table 5.2 Decision Table
The variables are normalized by multiplying with respective gains Ke, Kce, K0 so that their
values lie between -1 and +1. The membership function for speed deviation, acceleration and
voltage are shown in Figure 5.6 and Figure 5.7 respectively. Knowledgebase involves defining
the rules represented as IF-THEN rules statements governing the relationship between input
and output variables in terms of membership functions. In this stage, the input variables speed
deviation and acceleration is processed by the inference engine that executes 7 × 7 rules
represented in rule Table 5.2. Each entity shown in Table 5.2 represents a rule. The antecedent
of each rule conjuncts speed deviation (∆ω) and acceleration (∆ẇ) fuzzy set values. The
knowledge required to generate the fuzzy rules can be derived from an offline simulation.
Some knowledge can be based on the understanding of the behavior of the dynamic system
under control. For monotonic systems, asymmetrical rule table is very appropriate, although
sometimes it may need slight adjustment based on the behavior of the specific system. If the
system dynamics are not known or are highly nonlinear, trial and error procedures and
experience play an important role in defining the rules. An example of the rule is: If ∆ω is NS
78
and ∆ẇ is NM then U is NB which means that if the speed deviation is negative small and
acceleration is negative medium then the output of fuzzy controller should be negative big.
The procedure for calculating the crisp output of the Fuzzy Logic Controller (FLC) for some
values of input variables is based on the following three steps.
Step 1: Determination of degree of firing (DOF) of the rules
The DOF of the rule consequent is a scalar value which equals the minimum of two antecedent
membership degrees. For example if ∆ω is PS with a membership degree of 0.6 and ∆ ̇ is PM
with a membership degree of 0.4 then the degree of firing of this rule is 0.4.
Step2: Inference Mechanism
The inference mechanism consists of two processes called fuzzy implication and aggregation.
The degree of the firing of a rule interacts with its consequent to provide the output of the rule,
which is a fuzzy subset. The formulation used to determine how the DOF and the consequent
fuzzy set interact to form the rule output is called a fuzzy implication. In fuzzy logic control
the most commonly used method for inferring the rule output is a triangular method.
Step3: Defuzzification
To obtain a crisp output value from the fuzzy set obtained in the previous step a mechanism
called Defuzzification is used. In this example output, U is defuzzified according to the
membership functions shown in Figure 5.10. Here center of gravity (COA) or centroid method
is used to calculate the final fuzzy value. Defuzzification using COA method means that the
crisp output of U is obtained by using the center of gravity, in which the crisp U variable is
taken to be the geometric center of the output fuzzy variable value µout (U) area, where µout
(U) is formed by taking the union of all the contributions of rules with the degree of fulfillment
greater than zero. Then the COA expression with a discretized universe of discourse can be
written as follows:
∑ ( )
( )
∑ ( )
79
5.6.4 Rule viewer
The Rule Viewer shown in Figure 5.9 displays a roadmap of the whole fuzzy inference
process. It based on the fuzzy inference. The three plots across the top of Figure 5.8 represent
the antecedent and consequent of the first rule. Each rule is a row of plots, and each column is
a variable. The rule numbers are displayed on the left of each row [23].
Figure 5.9 Rule Viewers
5.6.4 Surface viewer
Up on opening the Surface Viewer shown in Figure 5.10, as presented with a two-dimensional
curve that represents the mapping from service quality to tip amount. Since this is a one-input
one-output case that can see the entire mapping in one plot. Two-input one-output systems also
work well, as they generate three-dimensional plots that MATLAB can adaptively manage.
80
When to move beyond three dimensions overall, starting to encounter trouble displaying the
results. Accordingly, the Surface Viewer is equipped with pop-up menus that let you select
any two inputs and anyone output for plotting. Just below the pop-up menus are two text input
fields that let you determine how many x-axis and y-axis gridlines you want to include. This
allows you to keep the calculation time reasonable for complex problems.
Figure 5.10 Surface viewers
5.7 Adaptive Neuro-Fuzzy Controller Design
Adaptive Neuro Fuzzy Interface system (ANFIS) is designed to achieve fast and effective
power oscillation damping in two area four machine power system.
81
5.7.1 Structure of the study system
The proposed controller, whose structure is shown in Figure 5.11, consists of two subsystems,
the identifier for the generator and the ANFIS PSS. The parameters of the identifier are
updated based on the error between the estimated generator speed deviation (∆ὠ) and its actual
value (∆ω), while the parameters of the ANFIS PSS are tuned by back-propagating the error
signal between ∆ɷˆ and its desired value ∆ωd [25].
VT
Power system
+
ANFISPSS
+ U ∆ω ∆𝝎̇
Exciter
- + &Generator ∆ω
Vref
Plant -
1
Ζ
Ζ 1 Plant
identifier ∆ὠ
1
Ζ
1
Ζ
Figure 5.11 structures of the study systems - ∆ɷd
The Plant Identifier

The plant identifier is very important for successful tuning of the PSS. Fast parameter tracking
ability is needed especially for time-varying systems.
82
5.7.2 Fuzzy logic Based Adaptive Neuro-Fuzzy Interface System PSS
A zero-order Sugeno-type fuzzy controller with 49 rules is used for the ANFIS PSS whose
block diagram is given in Fig. 5.12. The inputs to the PSS are speed and acceleration, which
are passed through a washout filter to eliminate any existing dc offset. The first scaling block
maps the real input to the normalized input space in which the membership functions are
defined. The second scaling block is used to map the output of the fuzzy inference system to
the real output needed. The fuzzy inference system consists of the Fuzzification block, rule
table block, and Sugeno Defuzzification block.
𝝎̇ ∆𝝎̇
Washout Fuzzifi Rule Sugeno
Scaling Defuzzi
𝝎 filter ∆𝝎 cation Table & u
fication Scaling
Interfac
e
Neural Network
Figure 5.12 Block diagram of ANFIS PSS model
The fuzzy control rule design is based on the understanding of the role and effect of the
controller, which is obtained by experience. The rules are built using the acceleration ( 𝝎̇) and
speed deviation (∆𝝎 ) as control variables, as shown in Table 5-3. In practice it is difficult to
measure Pm, and in addition, due to the slow governor response (relative to the exciter
response), Pm is assumed constant from one sample to the next.
The control rules are of the form: IF (∆𝝎) is Ai AND ( 𝝎̇ ) is Bi, THEN u = ki. Ai, Bi is the
fuzzy sets representing some linguistic terms, and ki is a crisp value. The membership function
for each linguistic term is of the form:
= ( ( )2 i
j2
) (5.3)
Where xj is the jth input, Aij represents the ith linguistic term related to the jth input and ,
j j
i are the centers and spreads of the membership function related to Ai , which are adjustable
by the neural network block of the ANFIS. The linguistic terms used for the membership
function are LP (large positive), MP (medium positive), SP (small positive), ZE (zero), SN
83
(small negative), MN (medium negative), and LN (large negative). The centers of the
membership functions are initially distributed evenly along the normalized input space.
The firing strength of each rule is calculated using the minimum function, which represents the
AND operation. The rule table is shown in Table 5-3. As seen in this table, an example rule is:
IF ∆ω is ZE and ∆𝝎̇ is MP, THEN u = 0.66, where u is the initial value of the output of each
rule (fuzzy consequence) used by the neural network. Since the fuzzy rules are established by
some experience (i.e., by an experienced control designer), the initial parameter values of the
ANFIS will be reasonable. The speed of adaptation of the ANFIS PSS is therefore dependent
on the values obtained from Table 3, which are functions of the PSS control variables (∆𝝎̇ and
∆ɷ) obtained on line [26].
Table 5.3 Rule matrix
∆𝝎̇
LP MP SP ZE SN MN LN
LP 1.0 1.0 1.0 1.0 0.66 0.33 0.0
MP 1.0 1.0 0.66 0.66 0.33 0.0 -0.33

∆ɷ SP 1.0 0.66 0.33 0.33 0.0 -0.33 -0.66
ZE 0.66 0.66 0.33 0.0 -0.33 -0.66 -0.66
SN 0.66 0.33 0.0 -0.33 -0.33 -0.66 -1.0
MN 0.33 0.0 -0.33 -0.66 -0.66 -1.0 -1.0
LN 0.0 -0.33 -0.66 -1.0 -1.0 -1.0 -1.0
The Sugeno Defuzzification algorithm is used to obtain a crisp (normalized) PSS output
signal. After the firing strength of each rule is obtained, the output of the fuzzy inference
system can be obtained by the operation of the weighted average of the form:
∑
∑
=∑ (5.4)
Where = ∑
(5.5)
Where, is the output of the ith rule, N is the number of rules, and is the normalized firing
strength of the ith rule.
84
5.7.3 Neural Network Representation of Fuzzy Logic Based power system stabilizer
The fuzzy logic-based controller can be made adaptive by being represented in a feed-forward
neural network form, as shown in Fig. 5.13. In this figure, K1 and K2 are inputs of scaling
factors and K3 is output scaling factor. This network consists of four layers, with each layer
representing a specific part of the ANFIS controller. The node functions in the first layer
represent the Gaussian membership function given by equation (5.3). In the second layer, the
firing strength of each rule is calculated. The third layer calculates the normalized firing
strength of each rule given by equation (5.5), and the fourth layer combines the output of all
rules to get the overall output of the controller using the Sugeno Defuzzification technique
given by equation (5.4)
There are two sets of parameters needed to be adjusted for the ANFIS PSS; one set contains
the scaling factors for the input and output of the PSS that are outside the fuzzy inference
system, and the other set is the membership function parameters and consequence parameters.
Experience shows that the scaling factors are important in the performance of the fuzzy logic
controller [28]. In this thesis proposed controller, the output scaling factor K3 is determined by
the output limits of the PSS. The scaling factors K1, K2 are chosen such that the variables x1,
x2 (Fig. 5.13) are the normalized values of and ̇ , respectively (i.e. -1≤ x1, x2 ≥
1).Therefore, to set K1, K2 to be (approximately) the reciprocal of the maximum values of
and ̇ respectively, as observed by simulation. The objective for updating the parameters
inside the fuzzy inference system is to minimize the cost function:
(5.6)
J= ̂ (t+1) 2
𝝎 ( )2
̂ (t +1) is the estimated speed deviation at time step t+1, l is the penalty factor to
Where 𝝎
improve the plant output dynamic characteristics such as overshoot and settling time of the
response curve. The gradient descent method is used to update all the adjustable parameters of
fuzzy inference system.
85
WI
𝝎 K1 X1
U
K3
𝝨
𝝎̇ X2 Bi
K2
Layer1 Layer 2 Layer3 Layer 4
Figure 5.13 Neural Network Models of the ANFIS
5.7.4 Offline adaptation by using ANFIS
Here first generate the input-output data pair of the system using the identifier, and by using
the ANFIS module in MATLAB to generate a fuzzy inference system. Two inputs are used,
namely Δω and Δ ̇ , and a single control output for the feedback. A Sugeno type Fuzzy
interface System model is used.
The membership functions of the inputs are of Triangular distribution type. Use seven
membership functions for each input to cover the full range of the respective inputs. Thus, to
get 49 rules for the output function this is linear relation of the inputs. The output is governed
by the AND function and thus the rules are generated.
86
Figure 5.14 ANFIS editor
The above generated fis file is opened in the ANFIS GUI for training and also import the
training data which was previously generated to the graphical user interface (GUI). The neural
network thus has four layers as shown in figure (5.15).
The first layer represents the input membership functions (MFs) which is Triangular. The
second layer represents the AND function. The third layer represents the normalized firing
strength as given in the Sugeno model and, the fourth layer represents the combination of the
rules and their weighted average to find the final output using Sugeno Defuzzification
technique.
87
Figure 5.15 Structures of Sugeno type ANFIS for PSS
5.8 Small-signal stability analyses of Multi-machine Systems
The dynamics of machine rotor circuits, excitation systems, prime movers and other devices
are represented by differential equations. The result is that the complete system model consists
of a large number of ordinary differentials and algebraic equations. Each machine model is
expressed in its own d-q axis references frame which rotates with its rotor. For the solutions of
interconnecting network equations, all voltages and currents must be expressed in a common
reference frame. Usually, a reference frame rotating at synchronous speed is used as the
common reference. Axis transformation equations are used to transform between the
individual machines (d-q) reference frame and the common (R-I) reference frame. For
convenience in the organization the complete set of algebraic equations, the machine stator
equations are also expressed in the common reference frame. The R-axis of the common
88
reference frame is usually used as the reference for measuring the machine rotor angle. For a
machine represented in detail including dynamics of one or more rotor circuits, the rotor angle
is defined as the angle by which the machine q-axis leads the R-axis as shown figure 5.16.
For a machine represented by a classical model, the rotor angle is the angle by which the
voltage Ѐ leads the R-axis. Under dynamic conditions, the angle changes with rotor speed.
EI
ER R
Figure 5.16 Reference frame transformation
The formulation of the state equations for small signal analysis involves the development of
linearized equations about an operating point and elimination of all variables other than the
state variables. The general procedure is the similar to that used for single-machine infinite bus
system. However, the need to allow for the representation of extensive transmission network,
loads, a variety of excitation systems and prime mover models, HVDC links, and static Var
compensators makes the process very complex. Therefore, the formulation of the state
equations requires a systematic procedure for treating the wide range of devices [1].
89
5.8.1 A Two-Area 4-Machine, 10-Bus test Systems
To test the performance of the proposed stabilizer, simulation studies were performed on two
systems: a one-machine-infinite-bus system and a two-area, 4-machine, 10-bus system. Seven
states are used to model each generator [1].
( 𝝎 )
The sampling period used in all cases was 0.01 sec and Power System Toolbox has been used
for this simulation study.
The enhancement of damping of low frequency oscillations in multi-machine power systems

by the application of supplementary controller has become a matter of great attention. It is
much more significant in current scenario where many large and complex power systems
frequently operate close to their stability limits. However; there is common perception that the
application of PSS is almost a mandatory requirement on all generators, the use of high price
PSS with each and every generator is constrained by economical limits [29].
Figure 5.17 Two Area Four-Machine power systems for Stability Analysis
90
In this thesis presents the effective and fast damping of inter-area power oscillation and
improved dynamic stability of interconnected power systems by designing of an ANFIS
controller for damping low frequency power oscillations in two area four machine system.
Disturbances have received much attention. Simulation model used for study of the CPSS and
proposed ANFIS controller based power system stabilizer (Hybrid-PSS) for Inter area power
oscillation stability.
The multi-machine power system is shown in Fig 5.17 that consists of two fully symmetrical
areas linked together by two 230 kV lines of 220 km length. Each area is equipped with two
identical round rotor synchronous acts as thermal plant generators rated 20kV/900MVA
connected to transformer (T1, T2, T3, and T4). The synchronous machines (M1, M2, M3, and
M4) in all area have identical parameters, except for the inertia which is H = 6.5s for all
generators in Area1 and H = 6.175s for all generators in Area 2.
Thermal generating plants having identical speed regulators and fast static exciters with a 200
gain at all locations. Each generator produces 700 MW. The loads are assumed everywhere as
constant impedance load. The Area1 and Area2 loads are 967 MW (L1) and 1767 MW (L2)
respectively. The load voltage profile was improved by installing 187 MVAr capacitors (C1
and C2) in each area to make closer to unity. Area 1 is exporting to Area 2 through two tie-
lines and a single tie-line with power transfer level 413 MW and 353 MW, respectively [1]
[30]. The complete parameters for this system are given in the Appendix -2A.
91
CHAPTER SIX
SIMULATION RESULTS AND DISCUSSION
6.1 Simulation Result of CPSS for dynamic performance measure (case study one)
To demonstrate the simulation results of a power system stabilizer by using the design of
controllers with computational optimization approach method, by substituting the value of
parameters: KP = 5, KD = 0.48 and KI = 5.
The simulation result clearly shows that the output of the model has large Tr, Ts and steady-
state error in the case of the model without a controller.
Figure 6.1 Simulink diagram of conventional power system stabilizer
92
Figure 6.2 Unit step response of conventional power system stabilizer
6.2 Performance measures with Fuzzy logic controller
The model used in Simulink/Mat lab to analyze the effect of the fuzzy logic controller in
damping small signal oscillations when implemented on single machine infinite bus system is
shown in Figure 6.3. The output responses as seen from the simulation results are crisp and
have good design specifications such as rise time, overshoot and settling time.
Figure 6.3 Simulink diagram of Fuzzy logic controller
93
Figure 6.4 Unit step response of FLC for power system stabilizer
6.3 Performance measures with Neuro-Fuzzy controller
The model used in Simulink/Mat lab to analyze the effect of the hybrid Neuro-fuzzy controller
in damping small signal oscillations when implemented on single machine infinite bus system
is shown in Figure 6.5.
Figure 6.5 Simulink diagram of Neuro-fuzzy controller

94
Figure 6.6 Unit step response of Neuro-Fuzzy controller system
The figure below shows that unit step system response for all controllers CPSS, FLC and
Neuro-fuzzy
Figure 6.7 Unit step system responses for all controllers
95
6.4 Comparisons and Discussion
Finally, to compare the conventional PSS or CPSS with the PSS developed using Adaptive
Neuro-Fuzzy inference system. As seen in Figure 6.7, the ANFPSS has the best output
response; the least overshoot and settling time. Also, it produces the best damping which is
manifested in the plot showing the amplitude. Thus, by proper training algorithms, the
Adaptive Neuro-fuzzy PSS can surpass the performance of the CPSS
Table 6.1 Comparison between CPSS, FPSS and NFPSS
Models Ts (sec.) Over Steady Methods Over shoot Settling time

shoot state Error improvement improvement
CPSS 13 4% 0 CPSS to FPSS 62.50% 21.54%
FPSS 10.2 1.5% 0 FPSS to 46.67% 26.5%
NFPSS
ANFPSS 7.5 0.8% 0 CPSS to 80% 42.31%
NFPSS
Performance is evaluated on the basis of maximum over shoot, settling time and Steady State
Error. The parameters obtained show that NFPSS achieves a better performance than CPSS
and FPSS. The Neuro-fuzzy controller enhances the performance of drive system. Simulation
results show that the error tends to be zero. Successful simulations demonstrate that Neuro-
fuzzy controller can achieve more desire performance than CPSS controller and fuzzy
controller. ANFPSS Stabilizer achieved the settling time by 42.31% and the overshoot 80%
quicker than Conventional Power System Stabilizer.
6.5 Performance evaluation with small disturbances (case study two)
The performance of the proposed model is tested on Single Machine Infinite Bus System
(SMIB) as shown in Figure 6.8. Then the performance of SMIB system has been studied
without excitation system, with excitation system only, with conventional PSS (lead-lag) and
with Adaptive Neuro-fuzzy based PSS by using the K constants. The dynamic models of
synchronous machine, excitation system, governing system and conventional PSS are
described in chapter two.
96
The Small signal performance measures provide a means of evaluating the response of the
close loop excitation control system to incremental changes in system conditions. Small signal
performance can be expressed in terms in of performance indices (settling time, overshoot, and
Rise time) used in feedback control systems.
Figure 6.8 Test Systems for Proposed Adaptive Neuro- Fuzzy Based PSS
6.5.1 Performance evaluation of Excitation System
The time response of the angular speed and angular position with excitation system has shown
in Figure 6.9 and Figure 6.10 for a positive and negative value of a K5 constant. From Figure
6.9 the response characteristic shows that under damped oscillations are resulted. The figure
shows that it has negative damping due to the fact that K5 constant calculated above is
negative, which is true for high values for external system reactance and high generator
outputs and thus the system is unstable.
97
Figure 6.9 Speed Deviation Responses with excitation system for -Ve K5
Figure 6.10 speed deviation Responses with excitation system for +Ve K5
The performance when analyzed with a positive value of K5, the system is stable. Positive K5
is possible for low values of external system reactance and low generator outputs. So for
positive, the damping is positive and thus system is stable.
98
6.5.2 Performance evaluation with Conventional Power System Stabilizer (CPSS)
Figure 6.11 shows the variation of angular speed, angular position and torque variation when
PSS (lead-lag) is applied for the negative value of K5. The result shows that the system is
unstable for the negative value of a K5 constant.
The figure 6.11 and figure 6.12 shows that they have negative damping due to the fact that K5
constant calculated above is negative, which is true for high values for external system
reactance and high generator outputs and thus the system is unstable. Hence the CPSS operate
under a specific condition which is low values of external system reactance and low generator
outputs.
Figure 6.11 Responses with CPSS for -Ve K5
99
Figure 6.12 Terminal voltage Responses with CPSS for -Ve K5
Figure 6.13 Responses with CPSS for +Ve K5
Figure 6.13 it shows that the system is stable for positive value of K5 constant and the higher
angular position is attained with positive K5 and Figure 6.14 shows that the terminal voltage
of the system which is stable for positive value of a k5 constant.
100
Figure 6.14 Terminal voltage Responses with CPSS for +Ve K5
6.5.3 Performance with Adaptive Neuro-Fuzzy Based Power System Stabilizer
The model used in Simulink/Mat lab to analyze the effect of Neuro-fuzzy controller in
damping small signal oscillations when implemented on single machine infinite bus system is
shown in Figure 6.1. The variation of angular position, angular speed and torque variation with
Neuro-fuzzy based PSS for the negative and positive value of a K5 constant is shown in Figure
6.15 and Figure 6.16 respectively. The results show that both are stable, but the higher angular
position is attained with positive K5.
Figure 6.15 Responses with ANFPSS for -Ve K5
101
Figure 6.16 Responses with ANFPSS for +Ve K5
The terminal voltages with Adaptive Neuro-fuzzy based PSS for both negative and positive
value of a K5 constant are shown in Figure 6.17 and Figure 6.18. The result clearly shows that
the ANFPSS is operating under a wide range and it is effective and robustness controller than
the CPSS.
Figure 6.17 Terminal voltage Responses with ANFPSS for -Ve K5
102
Figure 6.18 Terminal voltage Responses with ANFPSS for +Ve K5
6.6 Comparison of Conventional PSS and Adaptive Neuro-Fuzzy Based PSS
Table 6.2 Comparison of Settling time, the maximum overshoot of the dynamic responses of
rotor angle deviations of the study system with CPSS, and ANFISPSS
Models
Parameters CPSS ANFPSS
Peak Ts Over Peak Ts Over
Amplitude (sec.) shoot Ampli. (sec.) shoot
Angular speed with –k5 Worst Worst Worst 0.8 2.5 7%
Angular position with -k5 Worst worst Worst 1 0.08 0%
Torque variation with -k5 Worst Worst Worst 0 0.02 0%
Terminal voltage with –k5 worst worst worst 1.14 3 14%
Angular speed with +k5 0.145 6.8 14.5% 1.5 4.8 13.5%
Angular position with +k5 0.07 7 2% 1 0.08 0%
Torque variation with +k5 0 2 0% 0 0.02 0%
Terminal voltage with +k5 0.045 7.2 4.5% 1.12 5 12%
103
From the table 6.2 can be shown that the settling time of angular position, angular speed and
the terminal voltage for ANFPSS is much faster than CPSS. From the Table 6.2, the red color
indicates unstable when the constant k5 is negative and the yellow one indicates the system
output is stable when a k5 constant is positive for CPSS. The settling time angular position for
CPSS takes 7 sec to reach its final steady state value and the settling time angular speed takes
6.8 sec to reach its final steady-state value. The settling time angular position for ANFPSS
takes 0.08 sec to reach its final steady state value and the settling time angular speed takes 2.5
sec, 4.8 sec to reach its final steady-state value for k5 constant (torque coefficient) both
negative and positive respectively. Therefore it can be inferred that Neuro-fuzzy controller
does not require any complex mathematical support and the response is much impressed than
CPSS.
6.7 The proposed Simulation of Multi-machine power system (case study three)
The performance of proposed ANFIS-PSS was evaluated by applying a large disturbance

caused by three-phase fault applied at the middle of one tie line at 1 sec and cleared after 0.083
sec by opening the breakers.
Figure 6.19 Simulation model implemented with CPSS and proposed ANFIS-PSS for Inter
area power oscillation stability Analysis
104
To investigate the Inter area power oscillation damping performance of proposed ANFIS-PSS
with two-area four machine test system, the three phases to ground fault was considered in the
simulation studies .A three-phase fault of 0.083 sec duration is simulated to line.
6.7.1 Performance evaluation without conventional power system stabilizer ( CPSS)

under single line to ground fault
Figure 6.20 Voltages and Power transfer from Area1 to Area2
105
Figure 6.21 Performance of without PSS for Rotor angle deviation of machine, Rotor speed of
machine (ω), active power of machine (Pa), and terminal voltage of machine when single line
to ground fault occurs in transmission line.
6.7.2 Performance evaluation with CPSS under single line to ground fault
106
Figure 6.23 Performances of CPSS for Rotor angle deviation of machine, Rotor speed of
to ground fault occurs in transmission line.
107
6.7. 3 Performance evaluation with CPSS under symmetrical three phase fault (the worst
loading condition)
108
Figure 6.25 Performance of CPSS for Rotor angle deviation of machine, Rotor speed of
symmetrical three phase fault occurs in transmission line.
6.7. 4 Performance evaluation of Neuro-Fuzzy based PSS (ANFPSS) under single line to
ground fault
109
to ground fault occurs in transmission lines
110
6.7.5 Performance evaluation of Neuro-Fuzzy based pss (ANFPSS) under symmetrical
three phase fault ( the worst loading condition )
111
symmetrical three phase fault occurs in transmission line
Fig. 6.20 shows the plot of active power transfer response of system under without CPSS
condition and the response clearly shows that the generators loss of synchronisms (unstable).
Fig. 6.24 presents the result of the examined active power transfer response under CPSS and
the settling time takes 18.5 sec to reach its final steady state value while Figure 5.29 shows the
response of active power transfer response with proposed ANFIS controller based power
system stabilizer and the settling time takes 5 sec to reach its final steady state value. This
meant ANFIS based Power system Stabilizer achieved the settling time by 72.97% quicker
than Conventional Power System Stabilizer under symmetrical three phase fault. Therefore it
can be inferred that Neuro- fuzzy controller does not require any complex mathematical
support and the response is much impressed than CPSS.
112
CHAPTER SEVEN
CONCLUSIONS AND RECOMMENDATIONS
7.1 Conclusions
The real-time response of the non-linear system considered here reveals that ANFIS based
controller provides better response over the conventional controller. ANFLPSS shows the
better control performance than conventional power system stabilizer in terms of settling time
and damping effect. On concluding this thesis, I infer that Adaptive Neuro-Fuzzy Interface
System controller can be implemented for SMALL SIGNAL stability in power systems for the
quicker control action. The ANFIS based tuning of PSS is developed for a single machine
connected to infinite bus (SMIB) system which is used to in local mode oscillations.
The simulation results from the MATLAB Simulink shows that exceptionally a good damping
of low-frequency electromechanical oscillations for a wide range of operating conditions.
These show that they have negative damping due to the fact that K5 constant calculated from
the above is negative, which is true for high values for external system reactance and high
generator outputs and thus the system is stable in both conditions. However, the CPSS is
operating under specific conditions which are the low values of external system
reactance and low generator outputs. The small signal stability analysis a single machine
connected to infinite bus results show that ANFPSS Stabilizer achieved the settling time by
42.31% and the overshoot 80% quicker than the Conventional Power System Stabilizer.
The ANFIS design has an edge over both conventional and Fuzzy designs. The time taken to
stabilize the power system is the least compared to both the alternate designs. When a system
undergoes oscillations, a large time taken for tuning PSS could prove disastrous. The
Prolonged oscillations may lead to the isolation of generating units or shutdown of generating
station. Hence the ANFIS designed PSS proposed could if used prove to be much safer and
could enable the power system engineers to keep up a higher degree of system security.
The ANFIS is developing in this thesis work has been designed for Multi-machine power
systems. The inter-area modes of oscillation may lead to more oscillatory instability. For such
systems, the machine or groups of machines that give the oscillations should be identified and
the PSS on the identified machines should be tuned. This work forwarded a vigorous way
113
to make fast and efficient damping of inter-area power oscillations, with the development
of the proposed Adaptive Neuro-Fuzzy Interface System controller based PSS, an improved
dynamic stability of the connected power systems has been achieved by higher damping of
active power oscillation in two area four-machine power system. The basic aim was to use the
advantage of the Adaptive Neuro-fuzzy inference based structure. After the successful
implementation of the proposed ANFIS controller based power system stabilizer (ANFIS-
PSS), a complete testing process has been also presented by generating three-phase to ground
fault on the mid of the transmission line. The obtained results show that the inter-area power
oscillation damping ability of proposed ANFIS controller based power system stabilizer
(ANFIS-PSS) is higher than the conventional PSS. In addition to this, the results also indicate
that the proposed controller based power system stabilizer takes only the time 5 sec to
completely damp the inter-area power oscillations since CPSS takes 18.5 sec to control the
power oscillations. This meant Adaptive Neuro-Fuzzy Power System Stabilizer achieved the
settling time by 73% quicker than Conventional Power System Stabilizer.
7.2 Recommendation
 MATLAB simulation for power system stabilizer has been done which will be
implemented in real time to observe actual feasibility of the approach applied in this
thesis.
 This technique can be applied to multiple numbers of power system stabilizers.
 Neuro-Fuzzy Controller can be designed by using a single microchip. This microchip
will act as a brain for the machine. With this, will cut the costs for human resources
and the machine will work more effectively. The motives of the Neuro-Fuzzy Power
System Stabilizer are to save costs, reduce human error and automatically give another
look for the machine.
 The proposed controller can also be used for other dynamic security assessment
technique such as Dynamic breaking, Static Var compensators, and HVDC links.
114
REFERENCES
[1] P.Kundur, Power System Stability and Control. Mc-Graw Hill, 1994.
[2] AK Swagat Ranjan Swain, Ashit Kumar Swain,and Abin, "DESIGN OF POWER
SYSTEM," May 2012.
[3] al Del Rosso et, "a hierarchical control strategy for both dyanamic and steady state
stability enhancement," vol. 3, Jan. 2003.
[4] Yagami M. and Tamura J, "Power System Stabilization by Fault Current Limiter and
Thyristor Controlled Braking Resistor," EEE Energy Conversion Congress and
Exposition (ECCE), vol. 1, p. 335–338, 2009.
[5] L. Zhengguo, Y. Guanghong, and W. Changyun, "Stabilization of Power Systems by

Switched Controllers," Chinese Control Conference(ccc), pp. P756-760, 2007.
[6] Soon Kiat Yee and Milano Vic , "Fuzzy Logic Controller for Decentralized Stabilization
of Multimachine Power Systems," IEEE Transactions on fuzzy systems, vol. 16 , No 4,
pp. 971-981, 2008.
[7] A. Dysko, W. Leithead, and J. O‟Reilly, "Enhanced power system stability by

coordinated pss design," Power Systems, IEEE Transactions on, vol. 25, no. 1, p. 413–
422, Feb. 2010.
[8] Demello and Concordia, " Concepts of synchronous machine stability as affected by
excitation control," Power Apparatus and Systems, IEEE Transactions on, vol. PAS-88,
no. 4, p. 316–329, Apr. 1969.
[9] Sanchez-Gasca and J.H. Chow , "Pole-Placement Designs Of Power System Stabilizers,"
IEEE Transactions on Power Systems, vol. 4, No.1, Feb. 1989.
[10] Hiyama, A PID (Proportional +Integral +Derivative) type fuzzy Logic power system
stabilizer. 1994.
[11] Linkens M. M and M. F.Othman, "Designing power system stabilizer for multi machine
power system using neuro-fuzzy algorithm," in jurnal Teknologi, 2001, p. 55-64
115
[12] M. Vani, G. Raju, and K. Prasad, "Robust supplementary controllers for avr and pss,,"
2009 Annual IEEE, no. in India Conference (INDICON), pp. 1-4, Dec. 2009.
[13] C.-J. Chen, T.-C. Chen, H.-J. Ho, and C.-C. Ou, "Pss design using adaptive recurrent
neural network controller," in Natural Computation, 2009. ICNC ’09, vol. 2, no. Fifth
International Conference on, pp. 277-281, Aug. 2009.
[14] D. Sumina, N. Bulic, and G. Erceg, "Simulation model of neural network based
synchronous generator excitation control," EPE-PEMC 2008 13th, no. in Power
Electronics and Motion Control Conference, 2008, pp. 556-560, Sep. 2008.
[15] D.P. Kothari and I. J. Nagrath, Modern Power System Analysis, 4th ed. Mc-Graw Hill
Publcation, 2011.
[16] Bongiorno M. and Beza M., "Power oscillation damping controller by static
synchronous compensator with energy storage," Energy conversion congress.IEEE, pp.
2977-2984, Sep. 2011.
[17] "IEEE Guide for Synchronous Generator Modeling Practices and Applications in
PowerSystem Stability Analyses," Electric Machinery Committee of the IEEE Power
Engineering Society IEEE Std, R2007.
[18] Joe H.Chow, G.E.Boukarim and A.Murdoch, "Power System Stabilizers as

Undergraduate Control Design projects," IEEE Transactions on Power systems, vol. 19,
Feb. 2004.
[19] Dr Mohammed A.Abido, Ahmed A. and Ba-Muquabel, "Review of conventional PSS

design methods," in GCC Conference IEEE, 2006.
[20] l P. Mitra, S. P. Chowdhury and S. K. Pal eta, "Intelligent AVR and PSS With Hybrid
Learning Algorithm," Power and Energy Society General Meeting - Conversion and
Delivery of Electrical Energy in the 21st Century IEEE, pp. 1-7, 2008.
[21] Jang J. and Shing R, "ANFIS: Adaptive-network based Fuzzy Inference System," IEEE
Transactions on SMC, vol. 23 No.3, May 1993.
116
[22] ARSHAD, MOHD. FUAD BIN HAJI MOHD, "MODEL AND PERFORMANCE
SIMULATION FUZZY LOGIC POWER SYSTEM STABILIZER USING MATLAB
7.0," Nov. 2007/2008.
[23] N. Hosseinzadeh, "A Rule Base Fuzzy Power System Stabilizer Tuned By A Neural
Network," IEEE Transaction on energy Conversion, vol. 14 No. 3, Sep. 1999.
[24] Lu J., Nehrir M. H. and Pierre D. A, "A Fuzzy Logic-based Adaptive Power System
Stabilizer for Multi-machine Systems," IEEE Power Engineering Society Summer
Meeting, Seattle, WA, USA, vol. 1, pp. 111-115, 2000.
[25] Syahputra, Agus Jamal and Ramadoni, "Adaptive Neuro-Fuzzy Approach for the Power
System Stabilizer Model in Multi-machine Power System," International Journal of
Electrical & Computer Sciences IJECS-IJENS, vol. 12 No: 02.
[26] Rout Kamalesh Chandra, "Dynamic Stability Enhancement of Power System Using
Fuzzy Logic Based PowerSystem Stabilizer," pp. 18-45, May 2011.
[27] J.-H. Park, H. T. Kang and S. Kim, "Design of fuzzy power system stabilizer using
adaptive evolutionary algorithm," in Industrial Electronics,ISIE 2000, vol. 1., 2000.
[28] Ruhua, You, CONTROLLER DESIGN FOR PSS AND FACTS DEVICES O ENHANCE
DAMPING OF LOW-FREQUENCY POWER OSCILLATIONS. MONTANA STATE
UNIVERSITY, 2006.
[29] Klein Rogers, Moorty and Kundur, "Analytical investigation of factors influencing PSS
performance," IEEE Trans. on EC, vol. 7, no. 3, pp. 382-390, Sep. 1992.
[30] Devendra P.Parmar, Vihang M.Dholakiya, and Santosh, "Optimal Placements of Power
System stabilizers: Simulation studies on Test System," Presented at 2nd International
conference of Current Trends in Technology (NUiCONE‟11) at Nirma University, no.
Ahmedabad, ISBN No. 978-1-4577-2169-4.
117
APPENDICES
Appendix-1 Important results and data
1-Generalized 7th Order States spaces:
1 1 1 1
1 1 1 1 1
A= 1
1 1
1 1 1
[ 1 1 11 ]
B= , 1 -T
1 1 1 1
C= [ 1 ]
1 1 1
2- The model for GW i.e. effect of speed on electrical torque due to machine dynamics
AW [ ]
The model of the damping loop is:-
ξ' = A33ξ + a32ω
τ = a23ξ
Where, K=0.2462 D=0.1563
1 1 1 1
A33
1 1
[ 1 11 ]
118
a32
[ ]
a23 , 1 1 1 1 -
3. Transfer function of the 7th order observer-controller VR:
a) 405.6 s^6 + 7.32e04 s^5 + 3.588e06 s^4 + 6.357e07 s^3 + 4.906e08 s^2 + 4.832e09 s+
1.177e10
----------------------------------------------------------------------------------------------------------
S^7 + 193.6 s^6 + 1.122e04 s^5 + 2.722e05 s^4 + 3.259e06 s^3 + 2.811e07 s^2
+ 1.849e08 s + 3.201e08
b) Transfer function of 1storder minimized VR
1 1
c) Transfer function of the 11th order observer-controller VR:

-25.76 s^10 - 9696 s^9 - 1.347e06 s^8 - 8.982e07 s^7 - 3.325e09 s^6 - 7.237e10 s^5- 9.002e11
s^4 - 5.494e12 s^3 - 1.087e13 s^2 - 2.063e12 s - 1.03e11
---------------------------------------------------------------------------------------------------------- s^11
+ 260 s^10 + 2.587e04 s^9 + 1.387e06 s^8 + 4.585e07 s^7 + 9.924e08 s^6
+ 1.445e10 s^5 + 1.409e11 s^4 + 8.281e11 s^3 + 1.781e12 s^2 + 3.046e11 s + 1.523e10
d) Transfer function of the 5th order minimized PSS
-20 s^4 - 4120 s^3 - 51580 s^2 - 10440 s – 540

-------------------------------------------------------------
S^5 + 39 s^4 + 507 s^3 + 6183 s^2 + 1112 s + 57
119
Appendix-2 Optimal synchronous machine and pss parameters
Table A.1. Optimal synchronous machine and pss parameters values
Synchronous machine values Synchronous machine values

parameters parameters
p 0.9 Lfd 0.153
Q 0.3 ksd 0.849
Et 1.0 ksq 0.849
f 60 ASAT 0.031
Xd 1.81 ksdI 0.434
xq 1.76 ksqI 0.434
xd1 0.3 H 3.5
XL 0 kd 0
Xe 0.65 TR 0.02
Ra 0.003 Tw 10
ɷ 376.9 ETmag 1.0
Tdo1 8.0 ᴪ1 0.8
Ladu 1.65 BSAT 6.93
120
Appendix-2A Optimal power system stabilizer parameters
Table A. 2. Optimal power system stabilizer parameters
PSS parameters values Pss. Values

parameters
k1 0.7636 Kin1 1.8
k2 0.8644 Kin2 29.58
k3 0.3231 Kout 1.05
k4 1.4189 TR 0.02
k5 -0.1463 T1 0.154
k6 0.4167 T2 0.033
kA 200
Kstab 9.5
Appendix-2B System Data
The system consist of two similar area connected by a weak tie line. Each area consists of two
coupled units, the parameters of system as depicted in below tables.
Table B.3. Machines Data
Gen. MVA
No rating Xd Xd‟ Xd” Tdo‟ Tdo” Xq Xq‟ Xq” Tqo‟ Tqo”
1 900 1.8 0.3 0.25 8.0 0.03 1.7 0.55 0.25 0.4 0.05
2 900 1.8 0.3 0.25 8.0 0.03 1.7 0.55 0.25 0.4 0.05
3 900 1.8 0.3 0.25 8.0 0.03 1.7 0.55 0.25 0.4 0.05
4 900 1.8 0.3 0.25 8.0 0.03 1.7 0.55 0.25 0.4 0.05
121
Inertia constant of Area-1 machines are 6.5 (for G1 & G2), Area-2 machines are 6.175
(for G3 & G4). Each step up transformer has an impedance of 0 + j0.15 p.u. on 900MVA and
20/230 kV base, and has an off-nominal ratio of 1.0.
Line Data:
The Transmission system nominal voltage is 230 kV. The parameters of the lines in
per unit on 100 MVA, 230 kV base are shown in below:
Table C.4. The line data of the generators
From bus To bus Length km R p.u /km XL p.u /km BC p.u/km
5 6 25 0.0001 0.001 0.00175
6 7 10 0.0001 0.001 0.00175
7 8 110 0.0001 0.001 0.00175
8 9 110 0.0001 0.001 0.00175
7 9 220 0.0001 0.001 0.00175
9 10 10 0.0001 0.001 0.00175
Load Data:
The load and reactive power supplied by the shunt capacitors at buses 9 and 10 are as
follows:
Bus No. PL (MW) QL (MVAr) QC (MVAr)
9 967 100 200
10 1767 100 350
Exciter and PSS Data of Two Area Test System
KA = 200 TR = 0.01 KST AB = 20.0 TW = 10.0

T1 = 0.05 T2 = 0.02 T3 = 3.0 T4 = 5.4
122

Bahir Dar Institute of Technology Facult

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Bahir Dar Institute of Technology Facult

Uploaded by

Copyright:

Available Formats

BAHIR DAR UNIVERSITY

BAHIR DAR INSTITUTE OF TECHNOLOGY

FACULTY OF ELECTRICAL AND COMPUTER ENGINEERING

SCHOOL OF RESEARCH AND GRADUATE STUDIES

DESIGN AND ANALYSIS OF ADAPTIVE NEURO-FUZZY BASED POWER SYSTEM

Dr.-Ing Belachew Bantyirga

Bahir Dar, Ethiopia

BAHIR DAR INSTITUTE OF TECHNOLOGY

FACULTY OF ELECTRICAL AND COMPUTER ENGINEERING

SCHOOL OF RESEARCH AND GRADUATE STUDIES

DESIGN AND ANALYSIS OF ADAPTIVE NEURO-FUZZY BASED POWER SYSTEM

Approval by Board of Examiners

________________________ _____________ ____________

Chairman, Dept. Graduate committee Signature Date

Dr.-Ing. Belachew Bantyirga ______________ ____________

Thesis Advisor Signature Date

Internal Examiner Signature Date

External Examiner Signature Date

Getaneh Mesfin _______________ __________

Name Signature date

Place: Bahir Dar

Date of submission: ____________________

Dr.-Ing Belachew Bantyirga ____________ ________

Advisor‟s Name Signature Date

TABLE OF CONTENTS ................................................................................................................ vi

LIST OF FIGURES ........................................................................................................................ xi

LIST OF TABLES ......................................................................................................................... xv

NOMENCLATURES AND ABBREVIATIONS ........................................................................ xvi

CHAPTER ONE .............................................................................................................................. 1

1.1 Background of Study.............................................................................................................. 1

1.1.2 Classification of stability .................................................................................................2

1.2 Statement of problem ............................................................................................................. 4

1.3 Objectives of the Study .......................................................................................................... 5

1.3.1 General objective .............................................................................................................5

1.3.2 Specific Objectives ..........................................................................................................5

1.4 Scope of the study .................................................................................................................. 6

1.5 Methodology .......................................................................................................................... 6

1.6 Significant of the thesis .......................................................................................................... 7

1.7 Organization of the thesis ....................................................................................................... 7

CHAPTER TWO ............................................................................................................................. 9

LITERATURE REVIEW ................................................................................................................ 9

CHAPTER THREE ....................................................................................................................... 13

EXCITATION SYSTEM AND POWER SYSTEM STABILIZER ............................................. 13

3.1 The Excitation System of the Alternator .............................................................................. 13

3.1.1 Effects of Excitation system ..........................................................................................14

3.2 Excitation System Model ..................................................................................................... 16

3.2.2 Effect of AVR on synchronizing and damping torque component ...............................19

3.3 Power System Stabilizer (PSS) Model ................................................................................. 20

3.3.1 General Structure of power system stabilizer ................................................................22

3.4.1 State - Space Representation of LTI Systems ...............................................................25

3.4.2 The State Equations .......................................................................................................26

3.4.3 State Equation Based Modeling Procedure ...................................................................28

3.5 Small signal model ............................................................................................................... 29

3.5.1 Eigenvalue analysis method ..........................................................................................30

3.5.2 Damping ratio and linear frequency ..............................................................................32

3.5.3 Damping performance criteria .......................................................................................32

3.5.4 Small signal performance measures .............................................................................33

3.5.5 The alternator state space model ...................................................................................33

3.6 Multi-machine power system analysis ................................................................................. 35

3.6.1 State space model of multi-machine system .................................................................36

CHAPTER FOUR .......................................................................................................................... 39

CONVENTIONAL POWER SYSTEM STABILIZER DESIGN ................................................. 39

4.1 Introduction .......................................................................................................................... 39

4.2 Design of Automatic voltage Regulator (AVR) ................................................................... 40

__________________ _ ______

Dr.-Ing. Belachew Bantyirga __

Getaneh Mesfin _____

Dr.-Ing Belachew Bantyirga ____