Tuning of Power System Stabilizers

(PSSs) using Phenomenological

Models and Optimization Techniques
Sintonizaci on de Estabilizadores de Sistemas de
Potencia (PSSs) empleando Modelos Fenomenologicos
y Tecnicas de Optimizaci on
Diana Carolina Hernandez Gonzalez
Universidad Nacional de Colombia
Facultad de Ingeniera y Arquitectura
Departamento de Ingeniera Electrica, Electronica y Computacion
Manizalez, Colombia
2012
Tuning of Power System Stabilizers
(PSSs) using Phenomenological
Models and Optimization Techniques
Sintonizaci on de Estabilizadores de Sistemas de
Potencia (PSSs) empleando Modelos Fenomenologicos
y Tecnicas de Optimizaci on
Diana Carolina Hernandez Gonzalez
Tesis de grado presentado como requisito parcial para optar al ttulo de:
Magister en Ingeniera-Automatizacion Industrial
Directora:
Ttulo (Ph.D.,MSc) Rosa Elvira Correa Gutierrez
Codirector:
Ttulo (Ph.D.,MSc) Juan Manuel Ramrez Arredondo
Universidad Nacional de Colombia
Facultad de Ingeniera y Arquitectura
Departamento de Ingeniera Electrica, Electronica y Computacion
Manizalez, Colombia
2012
A mi Mama y a mi Pap a.
So I said, Wisdom is better than strength.
But the wisdom of the poor man is despised and
his words are not heeded. The words of the wise
heard in quietness are better than the shouting
of a ruler among fools. Wisdom is better than
weapons of war.
Ecclesiastes 9.
Agradecimientos
A Dios por amarme y porque nunca me abandonas.
A mis padres Eliecer Hernandez y Miriam Gonzalez por su amor incondicional, por su apoyo
y por creer en mi. Un gran ejemplo a seguir. Gracias por estar a mi lado.
A mis hermanos Jennifer y Rafael por estar siempre conmigo. Los amo!
A la pastora Luz Marina Castro por su guia y apoyo espititual.
A mi directora Rosa Elvira Correa y a mi codirector Juan Manuel Ramrez por su guia, gran
apoyo y conocimientos compartidos.
A Diana Paola Montoya y Juan Diego Sanchez por abrirme las puertas de su casa, por su
apoyo, por el conocimiento compartido y por creer en mi.
A mis amigos de la Universidad Nacional y del CINVESTAV: Luis, Javier, Dario, Xiomara,
Guillermo, Nestor, Fredy, Adrian, Jose Luis y Johnny por los espacios de discusi on, por su
apoyo y por hacer muy grata mi estancia en Mexico.
A Oscar Jose Arango por creer en mi, en mi tesis y por su apoyo.
A la Universidad Nacional de Colombia y al CINVESTAV por permitir hacer mi estancia en
Mexico.
A X.M por su apoyo economico.
ix
Resumen
En esta investigacion, se propone una metodologa para coordinar estabilizadores de poten-
cia (PSSs) con el objetivo de amortiguar las oscilaciones. Esta se basa en la minimizacion
de una funcion objetivo que incluye restricciones de igualdad y estudios transitorios. La
minimizacion es llevada a cabo por los algoritmos geneticos, que requieren de simulaciones
en el dominio del tiempo. La efectividad y robustez del procedimiento se muestran a traves
de simulaciones fuera de lnea y simulaciones en tiempo real para varios sistemas de potencia,
teniendo en cuenta diferentes condiciones de operacion y analisis de las fallas en diferentes
lugares
Palabras clave: Sistema Eletrico de potencia, estabilidad transitoria, estabilizadores
de potencia (PSSs), algoritmos geneticos y dSPACE.
Abstract
In this research, a methodology to coordinate power system stabilizers (PSSs) in order to
damp out oscillations is proposed. It is based on the minimization of an objective function
including equality constraints and transient studies. The minimization is carried out by
Genetic Algorithms, requiring time domain simulations. The eectiveness and robustness
of the procedure are demonstrated through digital simulations and real time simulations for
several power system, taken dierent operating conditions into account and analyzing faults
at dierent locations Keywords: Electrical power system, transient stability, power sys-
tem stabilizers (PSSs), Genetic Algorithms(GA), dSPACE.
Contents
Agradecimientos vii
Resumen ix
1 Introduction 2
1.1 Motivation and background . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Problem Statement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 A Brief review of previous work . . . . . . . . . . . . . . . . . . . . . 4
1.2.2 Thesis Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.2.3 Reseach Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.4 Outline of the contents . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2 Preliminaries 8
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Synchronous machines modelling . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3 Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.1 Classical formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.4 A small signal formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4.1 Eigenvalue sensitivities . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5 Evolutionary techniques. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.6 Emulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.6.1 Basics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6.2 dSPACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
3 Optimal Tuning of Power System Stabilizers 24
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2 Proposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3 Single Machine Innite Bus Power System (SMIB). . . . . . . . . . . . . . . 28
3.3.1 Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.3.2 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.4 Multimachine Test System - New England power system. . . . . . . . . . . . 32
3.4.1 Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.4.2 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Contents xi
3.5 Mexican Interconnected Power System (MIPS). . . . . . . . . . . . . . . . . 43
3.5.1 Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.5.2 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4 PSSs Testing on a Real Time Environment 50
4.1 dSPACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.2 Experimental Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
5 Conclusions and future work 58
Bibliography 60
List of Tables
2-1 Stabilizers parameters used as typical. . . . . . . . . . . . . . . . . . . . . . 18
3-1 Operating conditions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3-2 Estimated parameters for the conventional PSSs. . . . . . . . . . . . . . . . 34
3-3 Operating Cases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3-4 Estimated parameters for the conventional PSSs. . . . . . . . . . . . . . . . 45
4-1 Operating Cases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
List of Figures
1-1 Flowchart for the proposed coordination. . . . . . . . . . . . . . . . . . . . . 4
2-1 GA owchart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
3-1 Flowchart for the proposed coordination. . . . . . . . . . . . . . . . . . . . . 27
3-2 Single-machine innite-bus power system. . . . . . . . . . . . . . . . . . . . . 28
3-3 Angular speed-comparing with and without PSS. . . . . . . . . . . . . . . . 29
3-4 Active power ow: comparison with and without PSS. . . . . . . . . . . . . 30
3-5 Active power ow for each case. . . . . . . . . . . . . . . . . . . . . . . . . . 30
3-6 Angular speed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3-7 Dierence between the angular speed and reference speed at time t. . . . . . 31
3-8 New England Power System. . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3-9 Case 70%, fault in the bus 14 . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3-10 Case 70%, fault in the bus 27 . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3-11 Case 70%, fault in the bus 37 . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3-12 Case 100%, fault in the bus 13 . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3-13 Case 100%, fault in the bus 19 . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3-14 Case 100%, fault in the bus 28 . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3-15 Case 130%, fault in the bus 11 . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3-16 Case 130%, fault in the bus 22 . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3-17 Case 130%, fault in the bus 39 . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3-18 Mexican Interconnected Power System-MIPS . . . . . . . . . . . . . . . . . . 44
3-19 MIPS-Case 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3-20 MIPS-Case 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
3-21 MIPS-Case 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4-1 Experiment with dSPACE. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4-2 Case 70%, fault in the bus 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4-3 Case 70%, fault in the bus 20 . . . . . . . . . . . . . . . . . . . . . . . . . . 53
4-4 Case 70%, fault in the bus 25 . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4-5 Case 100%, fault in the bus 13 . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4-6 Case 100%, fault in the bus 16 . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4-7 Case 100%, fault in the bus 28 . . . . . . . . . . . . . . . . . . . . . . . . . . 55
List of Figures 1
4-8 Case 130%, fault in the bus 11 . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4-9 Case 130%, fault in the bus 36 . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4-10 Case 130%, fault in the bus 39 . . . . . . . . . . . . . . . . . . . . . . . . . . 57
1 Introduction
1.1 Motivation and background
Two of the most important design criteria for multimachine power systems are transient
stability and damping of electromechanical modes of sustained oscillation. These two design
criteria have assumed even greater importance in the wake of recent interconnection blackouts
in the U.S., Canada, and Europe. The focus has almost exclusively been on the second
criterion, oscillation instability as cured by suitably tuned power system stabilizers (PSS)
attached to appropriate generators. Particular emphasis is given to the performance of PSS
devices. On the contrary, this research deals with overall power system stability, i.e., both
the system transient stability as provided by the AVR devices and the system oscillation
stability\damping as provided by the PSS devices. Power system stabilizers (PSSs) are the
most well-known and eective tools to damp power system oscillation caused by disturbances.
To gain a good transient response, the design methodology of the PSS is quite important.
Existing techniques for the analysis of small signal stability such as eigenvalues analysis are
based on a few selected points from the wide range of possible operating conditions. Based
on engineering judgment, and experience, small signal stability limits can be approximately
determined. Nevertheless, these techniques do not guarantee acceptable performance or
even stability other than at the design condition. More importantly, the known methods
do not produce an indication of the stability margin that is needed for the development
of remedial measures. Ever increasing complexity of electric power systems has increased
research interests in developing more suitable methodologies for power system stabilizers
(PSSs). PSSs are the most eective devices for damping low frequency oscillations and
increasing the stability margin of the power systems. In fact, a PSS provides the excitation
system with a proper supplementary signal in-phase with the rotor speed deviation resulting
stable operation of the synchronous generator.
In the last two decades, various types of PSSs have been introduced. Fuzzy Logic Based PSS
(FLPSS) and adaptive controller-based PSS with some capabilities have been developed in
recent years. Conventional power system stabilizers (CPSS) are one of the premiere PSSs
composed by the use of some xed lag-lead compensators. CPSSs still are widely being used
in the power systems and this may be because of some diculties behind using the new
techniques. To overcome the diculties on the PSS design, intelligent optimization based
techniques have been introduced. These techniques can be divided into two categories:
time domain and frequency domain methods. In the time domain design, generally after
1.2 Problem Statement 3
applying a disturbance to the power system, some of the main signals are optimized.
However, disturbances at various locations may excite dominant modes with quite dierent
specications, leading to dierent PSS tuning parameters. Also this method may require
heavy computational burden for big power systems simulation. This research formulates
the robust PSS design as a mono-objective optimization problem and employs GA to solve
it. Results are demonstrated through a real-time environment.
1.2 Problem Statement
In this research, the formulation is based on the following optimization problem,
min f (x) =

kMB
[
ref

k
(x, t)]
2
(1-1)
where x is the vector of variables (PSSs gains and time constants);
ref
is the reference
angular speed;
k
(x, t) is the k-th synchronous machine angular speed after some
disturbance(s); t is time; MB is the set of synchronous machines equipped with PSS.
In order to take dierent conditions of operation into account, Eq.(1-1) may include an
additional summation which includes such conditions. In such case, Eq.(1-1) may be
weighted to formulate the problem by,
min f (x) =

iOC

i
{

kMB
[
ref

k
(t)]
2
} (1-2)
index OC is related to conditions of operation, and
i
is a weighting factor. Conventionally,

i=OC

i
= 1 (1-3)
In order to illustrate the PSS performance, in this research Eq.(1-2) is solved by a genetic
algorithm (GA), taking three operating points into account. Genetic algorithms have been
extensively used to solve optimization problems. Thus, it is well known as a powerful method
and it is utilized in this research. The Fig.1-1 depicts a owchart of the proposed strategy.
4 1 Introduction
Start
Estimation of
Parameters by GA
optimizer.
Evaluate objective
function (f)
min f(x)
Transient Stability
Analysis.
End
no
yes
Initial Population.
Generate
disturbance.
Power Flow
(Initial operating
condition).
Electrical power
system.
Figure 1-1: Flowchart for the proposed coordination.
1.2.1 A Brief review of previous work
In modern power systems, the high-voltage transmission network interconnects the widely
dispersed remote generations with load demands across the grid. The dynamics of power
1.2 Problem Statement 5
transfer in such complex grids can be characterized by electromechanical oscillatory behaviors
amongst synchronous generators [1, 2, 3]. These can be categorized into two main types: (i)
local and (ii) inter-area mode.
Since the stability of inter-area oscillation deteriorates with increasing transmission stress
buildup, procedures and equipments for maintaining adequate damping become even more
crucial [4]. The traditional approach to ensure adequate damping of electromechanical
modes is to install Power System Stabilizers (PSS) that provides supplementary control
actions through the generator excitation systems [5, 6, 7]. The primary objective of PSS
design is oscillation-mode stabilization. More recently, supplementary modulation controlled
FACTS (Flexible AC Transmission System) devices are also popular-alternative solutions
for damping relevant inter-area modes. However, FACTS are still to be widely installed in
transmission grids across the globe.
The design criteria have assumed even greater importance in the wake of recent
interconnection blackouts in the U.S., Canada, and Europe [8, 9]. The focus has
almost exclusively been on the oscillation instability as cured by suitably tuned power
system stabilizers attached to appropriate generators. Particular emphasis is given to the
performance of PSS devices, since the power system stabilizers are the most well-known and
eective tools to damp out power system oscillation caused by disturbances. To gain a good
transient response, the design methodology of the PSS is quite important. Ever increasing
complexity of electric power systems has increased research interests in developing more
suitable methodologies for power system stabilizers [10].
To overcome the diculties of PSS design, intelligent optimization-based techniques have
been introduced [11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22]. These techniques can be
divided into two categories: (i) time domain [20] and, (ii) frequency domain methods. In the
time domain design, generally after applying a disturbance to the power system, some of the
main signals are optimized. However, disturbances at various locations may excite dominant
modes with quite dierent specications, leading to dierent PSS tuning parameters. Also,
this method may require heavy computational burden for big power systems simulation.
Abdel-Magid and Abido[11] and Zhang and Coonick [22] have proposed frequency domain-
based techniques that seem more complete than the others.
Several modern control techniques can be used to design dierent PSS. However, power
systems companies prefer to choose the lead-lag structure because of its simplicity and
reliability in actual power systems implementation. Conventional PSSs (CPSSs) [23, 24]
use transfer functions designed for linear models representing the generators at a certain
operating point. However, they have some major limitations [25, 26]: (i) as they are
designed o-line, they require further tuning during commissioning; (ii) as they are tuned
for one operating condition, they cannot achieve the same level of performance for varying
operating conditions [27, 28, 29]; (iii) as the power system conguration or conditions change
with time, they require retuning at appropriate intervals for continued good performance.
Due to the limitations of CPSSs, new and better performing controllers based on alternative
6 1 Introduction
synthesis techniques that do not require a mathematical model of the system can be found
in the literature. Pal and Chaudhuri [18] present a comprehensive review of coordinated
PSS design methods.
Considerable eorts have been directed toward developing adaptive PSS, e.g.,[27, 28, 29, 30].
The basic idea behind adaptive techniques is to estimate the uncertainties in the plant on-
line based on measured signals [31, 32]. However, adaptive PSSs deal with systems of
known structures. Furthermore, adaptive controllers cannot use human experience which is
expressed in linguistic descriptions. This problem is overcome using articial intelligence
(fuzzy logic, neural networks, and decision trees)-based techniques for the PSSs design
[33]. Because of its complexity, the simultaneously tuning of PSSs has been investigated
by heuristic methods [34, 35, 36] and many PSS tuning methods using genetic algorithms
(GAs) were presented in [37, 38, 39, 40]. These methods investigated the use of GAs in
order to simultaneously tune the parameters of PSS with values that stabilize multi machine
power system over a wide range of scenarios.
Besides tuning methods for PSS, authors as Yuwa Chompoobutrgool, Vanfretti and
Ghandhari study appropriate PSSs tuning together with wide-area measurement signals. It is
expected that real-time and control using synchrophasor measurements could help enhancing
system stability and security, particularly by enhancing inter-area damping control [41].
1.2.2 Thesis Objectives
The objective of this dissertation are:
General objective.
Proposition of a methodology to coordinate power system stabilizers (PSSs) in order to
damp out electromechanical oscillations. This is attained by an optimization procedure. Its
eectiveness is demonstrated through nonlinear simulations and by real-time studies.
Specic objectives.
Apply a heuristic technique in the context of Electric Power Systems (EPS) and
emphasize the problem of controlling power oscillations.
Justify the use of heuristic optimization techniques, showing its main advantages in
complex scenarios such as the electric power system.
Analyze the inclusion of PSS in power systems and to develop a multi-machine power
system.
State the problem of power oscillations in power networks and propose a methodology
for tuning PSSs in order to help in damping out such oscillations. This procedure is
based on an optimization formulation solved by a heuristic optimizer.
1.2 Problem Statement 7
Verify the proposition through simulations and real-time studies on a dSPACE
platform.
1.2.3 Reseach Contributions
Due to the development and the increasing complexity of the power systems, it is necessary
to make use of modern controllers to improve their operation and control, without aecting
the systems safety and reliability. A possibility is to carry out o line emulations with using
models fully studied.
The main contributions of this research are summarized as follows:
Propose a methodology for tuning power system stabilizers, which is formulated by
an objective function solved by means of genetic algorithms. It is fundamental the
dynamic analysis to minimize the objective function and to add the robustness.
Develop applications in Simulink and dSPACE for future emulations and prove
controllers.
Emulate the power system and its controllers through dSPACE.
1.2.4 Outline of the contents
The thesis organization is as follows:
Chapter II: This chapter presents a brief overview about importance of the PSSs in the
system, previous work and brief description about methodology most used for tuning PSSs.
The small signal basic concepts, generators model and optimization are reviewed, besides a
brief description of the dSPACE.
Chapter III: This Chapter presents the proposed methodology for tuning PSSs, the
optimization problem and results obtained: The single machine innite bus (SMIB), New
England Power System and Mexican Interconnected Power System (MIPS).
Chapter IV: This Chapter presents the dSPACE simulator, its operation and results
obtained with real time simulations.
Chapter V: This chapter summarizes the conclusions resulting from this research work and
recommendations for future research on this topic.
2 Preliminaries
2.1 Introduction
Nowadays, with the power systems enlargement, to implement strategies able to improve
the operation has been necessary. For instance, with the market liberalization the power
systems operation has changed, so that the economical pressure and increasing amount of
transactions systems are operated closer to their operating limits. These changes put the
system operators faced with rather dierent and much more problematic scenarios than in
the past.
The power systems transient stability deals with the systems response to disturbances.
The most important period of interest is after the fault clearing. On this interval, the
electromechanical oscillations have to be damped as soon as possible, as an important factor
of power system security. After clearing a perturbation in a power system the modes
that appear may be classied in two groups: (i) local modes and, (ii) inter-area modes.
Oscillations may occur in the power system because of load uctuations, generators being
brought on, or line transmission switching; they may even apparently occur spontaneously.
These oscillations can either decay slowly and disappear or grow causing instability in the
system. For proper system performance, the oscillation should be removed from the system
quickly. With the growth in the energy demand and the interconnection of large electric
power systems, low frequency oscillations have become one of the main problems for power
system operation, aecting the systems reliability and security. It is clear that transient
stability and power system security play an important role to guarantee a secure and reliable
operation of the power system [42]. Transient stability of a system refers to the ability
of generators to remain in synchronism when it is subject to large disturbance such as
three-phase faults and switching of lines. Hence, system linearization is not applicable and
the nonlinear equations of the systems have to be solved. This complicates the analysis
considerably. The time period for transient stability is a few seconds, because the loss of
synchronism happens rapidly in that short time and causes that the angular position begins
to increase under the inuence of positive accelerating power and the system will become
unstable if experiences a large excursion.These and other factors have been very important
in the design and use of devices based on power electronics to help in assist the stability of
power systems after disturbances.
Power System Stabilizer (PSS) has been used as the most common supplementary control
2.1 Introduction 9
in order to improve the operation and add damping to help in attenuate power oscillations.
The PSS have proved to be very eective to add damping to the generator rotor oscillations
whenever there is a disturbance by controlling its excitation using auxiliary stabilizing
signals. However, the conventional PSS may not provide enough damping under special
circumstances. An important objective in planning, design and operation of an electric power
system is to provide security enough in supplying the electric power to the users. Therefore
power system security is one of the fundamental criterions in the power system design and
the main objective during its operation, under normal and perturbed conditions. Power
system stability only refers to the dynamic studies of the systems while the power system
security includes the study of the static and dynamic phenomena. In order to overcome these
shortcomings and to enhance the electromechanical oscillation damping, other electronic
devices can be eective. The advent of power electronic ones provides new ways of improving
and expanding the performance and capacity of power transmission systems. An interesting
and eective alternative is the Flexible A.C. Transmission Systems (FACTS). Its main
function is to modulate an adequate input signal to enhance or degrade the damping of
electromechanical oscillations, so that the FACTS devices are able to handle several states for
a secure system operation as controlling voltages at critical buses, controlling the phase angles
between the ends of transmission lines or modifying the series reactance of transmission lines.
A particular problem of concern in the power industry is the mitigation of low frequency
oscillations associated with the electromechanical modes which often arise between areas in
a large interconnected power network. These oscillations may be sustained and can grow to
cause system separation if not enough damping is available. To provide system damping, the
generators are equipped with power system stabilizers that supply supplementary feedback
stabilizing signals in the excitation systems. Certain excitation and system parameter
combinations under certain loading conditions can introduce negative damping into the
system. In order to oset this eect and to improve system damping in general, articial
means of producing torques in phase with the speed are introduced. These are called
supplementary stabilizing signals and the networks used to generate these signals are known
as power system stabilizers. For example, a fast acting, high gain voltage regulator,
although useful for improving transient stability margin, often depletes the generators
natural damping, thus rendering the system response highly oscillatory. When the use of such
a high-response regulator-exciter system is indicated from a transient stability consideration,
the resulting system oscillations can be minimized or eliminated by employing power system
stabilizers. Stabilizing signals are introduced at the point where the reference voltage and
the signal proportional to the terminal voltage are compared to obtain the error signal. The
signal, usually obtained from speed, frequency or accelerating power, is processed through a
suitable network to obtain the desired phase relationship. Although power system stabilizers
extend the power system stability limit by enhancing the system damping, they may cause of
great variations in the voltage prole [43], reducing the power system stability margin under
10 2 Preliminaries
severe disturbances if they are not properly coordinated. The stabilizers robustness under all
operating conditions is of concern and their interaction must be considered to obtain their
adequate design. In the open research, several techniques have been proposed to achieve a
satisfactory coordination of power system stabilizers (PSS) and FACTS devices stabilizers
(FDS) [43, 44, 45, 46].
Optimization techniques have been widely used for solving power system operation and
control issues. Optimization deals with problems of minimizing or maximizing a function
with several variables usually subject to equality and inequality constrains. There are a large
number of dierent approaches that have been developed to attain optimum designs. One
class of these techniques is the evolutionary methods that are used to search for the optimum
solutions via some form of directed random search process. A relevant characteristic of the
evolutionary methods is that they search for solutions without previous problem knowledge.
The stabilizers coordination is carried out by means of an objective function based on second
order eigen-sensitivities which is solved by Genetic Algorithms (GA) [46] and Simulated
Annealing (SA) separately. The PSS and FDS parameters are obtained in face of various
operating conditions.
2.2 Synchronous machines modelling
The performance of a power system is aected when a fault occurs. This will result in
insucient or loss of power. In order to compensate for the fault and resume normal
operation, corrective measures must be taken to bring the system back to its stable operating
conditions. Controllers are used for this function. Some of the control methods used to
prevent loss of synchronism in power systems are:
1. Excitation control: During a fault the excitation level of the generator drops
considerably. The excitation level is increased to counter the fault.
2. An addition of a variable resistor at the terminals of the generator. This is to make
sure that the power generated is balanced as compared to the power transmitted.
3. An addition of a variable series capacitor to the transmission lines. This is to reduce
the overall reactance of the line. It will also increase the maximum power transfer
capacity of the transmission line.
4. Turbine valve control: During a fault the electrical power output (P
e
) of the generator
decreases considerably. The turbine mechanical input power (P
m
) is decreased to
counter the decrease of P
e
.
Stability studies are generally categorized into two major areas: steady-state stability and
transient stability. Steady-state stability is the ability of the power system to regain
2.2 Synchronous machines modelling 11
synchronism after encountering slow and small disturbances. Example of slow and small
disturbances is gradual power changes. The ability of the power system to regain synchronism
after encountering small disturbance within a long time frame is known as dynamic stability.
Transient stability studies refer to the eects of large and sudden disturbances. Examples of
such faults are the sudden outrage of a transmission line or the sudden addition of removal of
the loads. Transient stability occurs when the power system is able to withstand the transient
conditions following a major disturbance. When a major disturbance occurs, an imbalance
is created between the generator and the load. The power balance at each generating unit
(mechanical input power-electrical input power) diers from generator to generator. As a
result, the rotor angles of the machines accelerate or decelerate beyond the synchronous
speed of for time greater than zero (t > 0). This phenomenon is called the swinging of
the machines.
There are two possible scenarios when the rotor angles are plotted as a function of time:
1. The rotor angle increases together and swings in unison and eventually settles at new
angles. As the relative rotor angles do not increases, the system is stable and in
synchronism.
2. One or more of the machine angles accelerates faster than the rest of the others. The
relative rotor angles diverges as time increase. This condition is considered unstable
or losing synchronism.
The main element in studying the power systems electromechanical phenomena is the
synchronous machine. Transient stability models are employed for generators, equipped
with a static excitation system; its representation is described as follows,
d
dt
=
0
(2-1)
d
dt
=
1
T
j
[T
m
T
e
D(
0
)] (2-2)
dE

q
dt
=
1
T

d0
_
E

q
(x
d
x

d
) i
d
+E
fd

(2-3)
dE

d
dt
=
1
T

d0
_
E

d

_
x
q
x

q
_
i
q

(2-4)
dE
fd
dt
=
1
T
A
[E
fd
K
A
(V
ref
+V
s
|V
t
|)] (2-5)
12 2 Preliminaries
where (rad) and (rad/s) represent the rotor angular position and angular velocity; E

d
(pu) and E

q
(pu) are the internal transient voltages of the synchronous generator; E
fd
(pu)
is the excitation voltage; i
d
(pu) and i
q
(pu) are the d-and q-axis currents; T

d0
(s) and T

q0
(s)
are the d-and q-open-circuit transient time constants; x

d
(pu) and x

q
(pu) are the d- and q-
transient reactances; T
m
(pu) and T
e
(pu) are the mechanical and electromagnetic nominal
torque; T
j
is the moment of inertia ; D is the damping factor; K
A
and T
A
(s) are the system
excitation gain and time constant; V
ref
is the voltage reference; V
t
is the terminal voltage;
V
s
is the PSS output fed into the excitation. The corresponding parameters are selected as
typical [47].
2.3 Optimization
Optimization is the mathematical discipline which is concerned with nding the maxima
and minima of functions, possibly subject to constraints. Nowadays, optimization is used
in a variety of applications related to Architecture, nutrition, Electrical circuits, Economics,
Transportation, etc. In power systems, the dynamic economic dispatch (DED) is one of the
most relevant problems. It is a complicated nonlinear optimization problem to determine
the optimal production levels of the scheduled units over a short-term horizon to meet load
demands while satisfying equality and inequality constraints in a way that the total fuel cost
is minimized. Traditionally, the valve-point loading eects of the large steam turbines were
ignored and a convex quadratic fuel cost function was considered for the thermal units, which
transformed the problem to a mathematically simple formulation resulting in the imprecise
dispatch results. Thus, in order to have a better illustration of the valve-point eects, the
non-convex and the non-smooth characteristics are used in the fuel cost function, which
makes the DED problem so complicated that it is quite hard to nd the optimum solution.
In general, there are two facets of optimization:
1. Modelling.
a) Translate the problem into mathematical language (sometimes trickier than you
might think).
b) Formulation of an optimization problem.
2. Solving.
Develop and implement algorithms that are ecient in theory and in practice. Likewise,
there is a close relationship between:
a) Formulate models that you know how to solve.
b) Develop methods applicable to real-world problems.
Likewise, there is a close relationship between: (1) formulate models that you know how to
solve; (2) develop methods applicable to real-world problems.
2.3 Optimization 13
2.3.1 Classical formulation
minf(x) xX R
N
(nite dimension.) (2-6)
Often we dene,
X = {x R
n
/g
i
(x) 0 and h
j
(x) = 0 for i I and j J} (2-7)
where x is the decision variable, and functions g(x) and h(x) are called inequality constraints
and equality constraints, respectively. When there are several decision variables, x =
[x
1
, x
2
, ..., x
n
] is a vector. In this case, the problem is a multidimensional one. One of
the most known methods to solve the unconstrained problem Eq. (2-6) is the Levenberg-
Marquardt algorithm.
Levenberg-Marquardt algorithm. Within the conventional optimization methods, the
Levenberg-Marquardt method has been widely used due to its robustness. Frequently, it
is possible to express the optimization problem as a sum of squares,
J(x) = [r(x)]
T
[r(x)] (2-8)
where x is the vector of stabilizers parameters. To minimize J(x), dierentiate Eq. (2-8)
and equate to zero; must satisfy the nonlinear equation Eq. (2-9)
dJ(x)
dx x= x
= 2[F( x)]
T
r( x) = 0 (2-9)
where
F( x) =
dr(x)
dx

x= x
(2-10)
is the Jacobian matrix.
One method of solving Eq. 2-9 is based on the Taylor series approximation of r(x) around
a nominal point x
0
, i.e.
r(x) = r(x
0
) +F(x
0
)[ x x
0
] (2-11)
Substituting Eq. 2-9 into Eq. 2-11 yields
[F
T
(x
0
)F(x
0
)][ x x
0
] = F
T
(x
0
) r(x
0
) (2-12)
[F
T
(x
q
)F(x
q
)] x
q+1
= F
T
(x
q
) r(x
q
) (2-13)
14 2 Preliminaries
Solve iteratively x
q+1
= xq + x
q+1
. The iterations of Eq. 2-12-2-13 are continued until
J( x) approaches a minimum.The method of estimating x by solving Eq. 2-12-2-13 is also
called the Gauss-Newton method. According to the Levenberg-Marquardt algorithm [48],
Eq. 2-12-2-13 may be solved by adding positive numbers to the diagonal of the matrix
F
T
(x
q
)F(x
q
) in case of oscillatory behavior in convergence and/or ill-conditioning of the
matrix. Thus, Eq. 2-12-2-13 becomes
[F
T
(x
q
)F(x
q
) + D] x
q+1
= F
T
(x
q
) r(x
q
) (2-14)
where D is a diagonal matrix and the constant >0. A small gives a Newtons step and a
large gives a steepest descent step. It is convenient to adjust by comparing the actual
reduction J(x) in the sum of squares, to the reduction that would have occurred if the
linear model
r(x
0
+ x) = r(x
0
) +F(x
0
) x (2-15)
A test for optimality of the point x
q
often carried out is:
_
_
_
_
dJ(x
q
)
dx
_
_
_
_
(2-16)
x
q
is optimum and hence stops the process. Smaller convergence values () result in a
better estimate of the model parameters. Despite its wide use, conventional optimization
algorithms may experience convergence diculties due to convexity problems or non-smooth
characteristics. These are some reasons that gave origin to alternative techniques, such as
the genetic algorithms.
2.4 A small signal formulation
Stability programs designed for large-disturbance (transient) stability studies simulate
system response in time domain following a disturbance. The simulations are normally
limited to a short duration, usually a few seconds. If the generator rotors swing back
before reaching a specied angle, the power system is considered large-disturbance stable.
In the early days when electric power networks were relatively conned, and sophisticated
control equipment were not generally utilized on the generators, the above criterion, also
called rst swing stability, was enough to assure eventual stability of the system against
that particular disturbance. Instability in the initial period following a large disturbance is
generally due to insucient (or lack of) synchronizing torque between the interconnected
generators. Automatic control equipment, e.g., fast acting excitation control, help improve
the rst swing stability by increasing the synchronizing torque. However, in the process they
often reduce the damping torque, sometimes even rendering the overall damping negative,
thereby causing oscillatory instability. With the growth of interconnection and application of
2.4 A small signal formulation 15
advanced control equipment, consideration of proper damping of oscillations became more
important. In a system capable of withstanding the initial shock of the disturbance, as
evidenced by rst swing stability, oscillations could continue at a reduced amplitude for a
while, only to increase a few cycles later and eventually cause cascading line tripping and
possibly system separation. This type of instability can manifest itself not only following a
major disturbance but also following a sudden small change in system condition not generally
classied as major disturbance, e.g., a moderate amount of load tripping, a sudden addition
of a large load, tripping of a minor transmission line, etc. In order for a power system to
be operable, it must have an equilibrium point that is stable. This means it must be small-
disturbance stable. In an interconnected system, due to an improper selection of control
parameters, a stable equilibrium point may not exist at all. Oscillations once started can
build up gradually although they may not be apparent during the rst few cycles. On
the other hand a well designed control system can extend the stability limit considerably.
Immediately following a small disturbance, or following a large disturbance after the system
has survived the initial shock (i.e., its rst swing) and entered a state of oscillation, the system
nonlinearities do not play a major role. The power system can therefore be linearized about
the equilibrium point and useful information on the system small-disturbance performance
can be obtained from the linearized model. This would permit ecient design of the control,
allowing necessary xes to combat instability. Certain aspects of the system stability problem
are more readily detectable, and hence correctable, by studying the linearized system.
The multi-machine model linearized around an equilibrium point is represented by the state
equation
dx
dt
= Ax +Bu (2-17)
y = Cx (2-18)
where x R
n
is the state vector, u R
q
is the input signal vector, and y R
p
is the output
signal vector. Common inputs are the exciter reference voltages, and the outputs are the
machine angular velocities or electrical powers. Lets dene v
i
and w
i
the right and left
eigenvectors associated with the i th eigenvalue
i
. p
fik
= v
ik
w
ki
is the participation
factor;
I
c
=

w
t
i
B
j

is the controllability index of eigenvalue

i
from station j;
I
ow
=

C
K
v
i

is the observability index of eigenvalue

i
from station k, considering angular velocities as
outputs. A large p
fik
suggests that the k th eigenvalue is quite sensitive to local feedback
16 2 Preliminaries
around the i th state variable. If I
ow
is large,
i
is strongly observable at the k th station.
If I
c
is large,
i
is strongly controllable at station j. Hence, based on these indexes, an
appropriate location for PSSs can be selected, to enhance the damping of those modes.
In this section a methodology to enhance the damping of power oscillations is reviewed
[46]. The power system dynamic equations are nonlinear, however, suited solutions for
small deviations around an operating condition may be found. Therefore, in the following a
linearized model of the power system around several equilibrium points are used. We assume
that the system model is composed of generators, loads and excitation systems. Likewise,
we assume that it had been made a complete linear analysis and we know where PSSs should
be installed [49]. The PSSs can be coordinated by the procedure described in the following.
2.4.1 Eigenvalue sensitivities
The eigenvalue sensitivity analysis has been used as an important tool for power system
dynamic studies and the controller design. It has been found that the trajectories of
the dominant eigenvalues for system parameter changes are nonlinear and the rst-order
estimates are not accurate [50, 51].
A method to consider the second-order eigenvalues of the augmented system matrix has been
presented in [51]. In this technique, sensitivity analysis of a particular mode is performed
with only its own left and right-eigenvectors; additional information is not needed [51]. The
well-known linearized model for the power system stability analysis may be represented as
follows:
_
x
0
_
=
_
A
G
B
G
C
G
Y
_ _
x
v
_
+
_
U
0
_
[u] (2-19)
or

x =

A x +

U u (2-20)
where:
x is the state vector.
v is the vector of the network bus voltages.
u is the vector of control input variables.
indicates perturbed values.

A is the augmented system state matrix.
2.4 A small signal formulation 17
The eigenvalue, right eigenvector v
i
, and left eigenvector w
i
of the augmented matrix are
evaluated by,

Av
i
=
i
Bv
i
(2-21)

A
T
w
i
=
i
Bv
i
(2-22)
where matrix B has values of 1s in its rst n-diagonal positions and 0s in all other positions,
and n is the number of state variables. Considering changes in the system parameters, the
augmented matrix

A can be change to a new augmented matrix

A
N
and its perturbation
results as E =

A
N

A. The rst and second-order eigenvalue perturbations are given as
follows [51]:

(1)
i
= w
i
T
Ev
i
(2-23)

(2)
i
= w
i
T
Ev
i
(2-24)
To compute the vector v
i
to solve the following equations simultaneously is required,
_

A
i
B
_
v
i
=
_

(1)
i
B E
_
v
i
(2-25)
w
i
T
Bv
i
= 0 (2-26)
As aforementioned, existing techniques for the analysis of small signal stability such as
eigenvalues analysis are based on a few selected points from the wide range of possible
operating conditions. Based on engineering judgment, and experience, small signal stability
limits can be approximately assumed. Nevertheless, these techniques do not guarantee
acceptable performance or even stability other than at the design condition. More
importantly, the known methods do not produce an indication of the stability margin that is
needed for the development of remedial measures. Trying to overcome such diculties,
an optimization function to compute de best PSSs parameters in order to damp the
electromechanical oscillations caused by perturbations was proposed. The following objective
function is solved [46]:
min
K

k=1

jZ
Re {
j
} (2-27)
where the eigenvalues perturbation can be represented as a second-order perturbation

j
=
j
(1)
+
j
(2)
, K = {set of operating conditions} and Z={all
i
of concern on
18 2 Preliminaries
the k-th operating condition}. The proposed objective function Eq. (2-27) minimizes the
real part of the sensitivities of the modes of concern in order to compute the best parameters
of the stabilizers, taking dierent operating points into account.
The conventional PSSs studied in this research have the transfer function:
y (s)
u (s)
=
skT
1 +sT
1 +sT
1
1 +sT
2
1 +sT
3
1 +sT
4
(2-28)
The rst term represents a washout lter used to eliminate steady state bias in the output
of PSS, which will modify the generator terminal voltage. Essentially, the washout acts as
a high pass lter and it must pass all frequencies that are of interest. The second term is a
dynamic compensator in series with the washout lter and is used to improve the phase lag
through the system [2]. To simplify the procedure, we consider T
1
= T
3
and T
2
= T
4
. For
each stabilizer, we are going to estimate just T
1
and K, because of T and T
2
are previously
selected. T ranging between 7.5 15s, is chosen to ensure a negligible phase-shift and
gain contributed by the washout block in the range of oscillating frequencies of concern. T
2
ranging between 0.020 0.10s, is chosen from a consideration of physical realization [2]. So
that, for N stabilizers, we are going to estimate 2N parameters (N time constants T
1
, and N
gains K). Notice that with this procedure robust stabilizer parameters are computed, able
to increase the damping of the modes of concern and operate satisfactorily over a wide range
of conditions. Changing the parameters of the stabilizers from a set of typical parameters,
creates the perturbation matrix E of Eq. (2-25-2-26),Table 2-1.
Table 2-1: Stabilizers parameters used as typical.
PSSs

Parameter V alue
T 7.5
K 0.1
T
1
= T
3
0.045
T
2
= T
4
0.015
2.5 Evolutionary techniques.
Unfortunately, in dierent propositions the optimization process requires computations
of sensitivity factors and eigenvectors at each iteration. This gives rise to the heavy
computational burden and slow convergence. Besides, the search process is susceptible to
be trapped in the local minima and the solution obtained will not be optimal. Furthermore,
the problem of the PSS design is a multimodal optimization problem (i.e., there exists more
than one local optimum). Hence, local optimization techniques are not suitable for such
a problem. Moreover, there is no local criterion to decide whether a local solution is also
the global one. Thus, conventional optimization methods that make use of the derivatives
and gradients are, in general, not able to locate or identify the global optimum, but for the
2.5 Evolutionary techniques. 19
real-world applications, one is often content with a good solution, even if it is not the best.
In recent years, global optimization techniques like Tabu search, Genetic Algorithms (GA),
simulated annealing, rule based bacteria foraging and strength pareto evolutionary algorithm
[10, 11, 12, 16, 44] have been applied for the PSS parameter optimization. These evolutionary
based methods are heuristic population-based search procedures that incorporate random
variation and selection operators. Although, these methods seem to be good approaches
for the solution of the PSS parameter optimization problem, however, when the system
has a highly epistatic objective function (i.e. where parameters being optimized are highly
correlated), and the number of parameters to be optimized is large, then they have degraded
eectiveness to obtain the global optimum solution and also simulation process takes a lot
of computing time. Shayeghi et. al [52] proposed a particle swarm optimization (PSO) for
the design of PSS parameters at dierent operating conditions in comparison with the GA
based tuned PSSs. However, the performance of the classical PSO greatly depends on its
parameters, and it often suers the problem of being trapped in the local optima so as to
be premature convergence.
Genetic Algorithms (GA) are stochastic search techniques considered global search methods
based on the mechanism of natural selection and natural genetics. GA acts as a biologi-
cal metaphor and tries to emulate some of the processes observed in natural evolution, for
instance plants and animals. They are viewed as randomized, yet structured, search and
optimization techniques. GA, diering from conventional search techniques, starts with an
initial set of random solutions called population. Each individual in the population is called
a chromosome, representing a solution to the problem. A chromosome is a set of genotypes,
which store the characteristics of solutions. The chromosomes evolve through successive
iterations named generations. For each generation the objective function (tness measuring
criterion) determines the suitability of each solution. Based on these values, some of them
(parent chromosomes) are selected for reproduction. The performance of a GA depends on
the tness evaluation criterion to a large extent. Genetic operators are applied on these (se-
lected) parent chromosomes and new chromosomes (ospring) are generated. The operators
frequently employed in GA are selection/reproduction, crossover, and mutation which are
used to generate a new population. Some of the parents form ospring by rejecting others
so as to keep the population size constant. Fitter chromosomes have higher probabilities of
being selected. After several generations, the algorithms converge to the best chromosome,
which hopefully represents the optimum or suboptimal solution to the problem. The general
structure of genetic algorithms is described in Fig.2-1.
20 2 Preliminaries
Generate initial
population
A small number of
individuals in the offspring
population will mutate
Randomly pair the
individual as a parent, then
perform crossover on the
pair to generate two
offspring.
Select individuals. The
probability of selection of
each individual is
proportional to its fitness.
Evaluate each individual by
fitness function.
Stopping does
criteria met?
Output results
Replace all individuals of
the previous generation
with the new offspring that
have completely gone
through selection,
crossover and mutation.
Yes
No
Figure 2-1: GA owchart
2.6 Emulation 21
The main elements of Genetic Algorithms can be described as follows [53]:
Selection. This operator selects the tter chromosomes in the population for
reproduction. The objective of GA is to converge to an optimal individual, and the
driving force is the selection pressure that determines the convergence rate.
Crossover. The idea of crossover is to mimic the natural of mating which recombine two
chromosomes of two organisms. This operator chooses one/many cutting point(s) and
exchanges the subsequences before and after that cutting point between two osprings.
Mutation. This operator simulates the natural mutation by randomly ipping some of
the bits in a chromosome.
Population. GA usually starts with a set of initial guess solutions that are randomly
generated. This set of possible solutions in each generation is called population. The
size of a population may vary from one generation to another or it can be constant.
Fitness Function. A tness function is used to evaluate the tness of each candidate
solution. The tness of a chromosome depends on how well that chromosome solves
the problem.
Representation. The basic feature of GA is that they work on coding space and solution
space. The coding is a mapping from the coding space to the solution space that
transforms the set of solutions to a nite length string. The codes represented by
genes in the chromosome enable the GA to proceed in a manner that is independent
of the solution space.
2.6 Emulation
A very attractive new kind of simulation has been available for some years: fully digital real-
time simulation. Traditionally used in transient analysis of electric power systems, digital
real-time simulators could also be very helpful to study systems. These simulations can
indeed be interfaced directly with industrial controllers, thus allowing extensive performance
evaluation without the cumbersome laboratory installations required by the real equipment.
Real-time simulation would therefore provide an economical tool for drive controller testing
and oer all the exibility needed to simulate machines in all power ranges. Fully digital real-
time simulation is based on parallel computing. In order to demonstrate the applicability of
the proposed methodology, results of real-time simulations are presented in the nal chapter.
These studies are carried out on a dSPACE platform [54].
22 2 Preliminaries
2.6.1 Basics
Real-time power system simulators have been used by power system engineers since several
decades. The complexity of electrical network phenomena, the constant growth of power
system networks, and the use of more sophisticated control equipments, has increased
the need for real-time simulation tool. Since several years, with the rapid progress of
computer technology, real-time simulation of power system network can now be conducted
fully numerically. While o-line simulation and modeling signicantly helps in preparing
a new product or prototype for eld deployment, thorough testing of the hardware under
dynamic and transient conditions is essential for a swift transition of technology from the
experimental phase to prototype and, subsequently, product phase. This approach is named
hardware-in-the-loop (HIL). The method of real-time simulation of electric power systems
and its application in hardware-in-the-loop (HIL) experiments can be a very powerful tool in
this process. As an advanced design and/or test method, HIL simulations allow the prototype
of a novel apparatus to be investigated in a virtual system under a wide range of realistic
conditions repeatedly, safely and economically. HIL simulations have been successfully used
for the tests of protection relays [55], [56], power converter controllers [57] and power quality
regulators [58]. Typically, the hardware under test consists only of controllers. This may
be referred to as a controller hardware-in-the-loop (CHIL) simulation. With CHIL, various
controllers such as protective relays, power electronic control boards, and rotating machine
controllers can be tested in a simulated, closed-loop environment which allows reproducing
of highly dynamic and transient power system phenomena under real-time constraints with
the full feedback between system and hardware. All the signals exchanged between the
simulator and the hardware are at low voltage and low power levels (typically in the range
of 10V and a few mA). The interface can be easily realized by ADCs (Analogue/Digital
Converters) or DACs (Digital/Analogue Converters). Such a simulation is therefore called
power hardware-in-the-loop (PHIL) simulation. Because the hardware under test absorbs
and/or sinks real power, appropriate power amplication and conversion apparatuses must
be included [59].
2.6.2 dSPACE
The dSPACE simulator is a hardware-in-the-loop (HIL) system. dSPACE allows:
1. Real-time simulations.
2. Generates and measures various signals, for example: PWM sensor signals and Hall
sensor signals (e.g., wheel speed, fuel level), resistance-based sensor signals (e.g.,
temperature), analog and digital sensor signals (e.g., throttle, switches, lamps, relays),
algorithm and waveform-based signal generation (e.g., crankshaft, camshaft, knock
signals), angle-based measurement of injection and ignition pulses, measurement
2.6 Emulation 23
of PWM actuator signals (e.g., solenoid valves),connection to CAN, LIN, FlexRay,
MOST, Ethernet, etc., and serial interfaces. Other buses are also possible.
To work with dSPACE systems, you must rst:
1. Generate a real-time application (RTI). For this, you rst create the Simulink model.
The Real Time Interface is the implementation software for single board hardware,
MicroAutoBox hardware and modular hardware with a single processor board. It
connects your model to the I/O of the board.
2. Add the I/O interface to the model using dSPACE RTI blocks.
3. Generate C Code. The Real-time Workshop generates C code from the model
automatically.
4. The cross compiler environment compiles the generated C code and links the object
les and libraries into an executable application for the real time processor.
3 Optimal Tuning of Power System
Stabilizers
A proposition is made to design and coordinate multiple power system stabilizers (PSS)
in order to enhance the electromechanical transient behaviour of power systems. A
technique for tuning stabilizers by an optimization problem is presented. The stabilizers
parameters are obtained through the minimization of an objective function based on
quadratic functions. The eectiveness and robustness of the procedure are demonstrated
through digital simulations on a dynamic equivalent of the Mexican power system.
3.1 Introduction
With the increasing electric power demand power systems can reach stressed conditions.
So that, measures need to be taken in order to prevent not allowed voltage and frequency
levels. For many years, considerable eorts have been made for designing and applying new
damping sources. The power system stabilizers (PSS) are one of the most common controls
used to damp out oscillations and to oset the negative damping of the automatic voltage
regulators [60]. The major role of PSS is to introduce modulating signal acting through
the excitation system to add damping to rotor oscillations. However, if the PSSs are not
properly coordinated, such devices may not provide enough damping, especially to inter-area
oscillating modes. That is the reason to look for smart strategies for coordinating PSSs.
Thus, the interaction among stabilizers may enhance or degrade the damping of certain
modes of rotors oscillating modes. The improvement hinges on an adequate coordination
of controllers in order to solve marginal operating problems, ensuring robustness for several
operating conditions.
Optimization techniques have been widely used for solving power system operating and
control strategies. Classical optimization deals with problems of minimizing or maximizing
an objective function with several variables, usually subject to equality and inequality
constraints. There are a large number of dierent approaches that have been developed to
attain optimum results. The evolutionary methods constitute an approach to search for the
optimum solutions via some form of directed random search process. A relevant characteristic
of the evolutionary methods is that they search for solutions without previous problem
3.2 Proposition 25
knowledge. In this research, the stabilizers coordination is carried out by a quadratic
objective function.
The power system dynamic equations are non-linear in nature, without a closed form
solution. However, it is possible to nd useful solutions for small deviations about a chosen
steady-state equilibrium point. A plethora of techniques to coordinate PSSs have been
presented based on a power systems linearized version. That will not be the case in this
research, where non-linear models are used. It is assumed that the power system is composed
of generators, loads and excitation systems, and dierent operating conditions are taken into
account.
3.2 Proposition
In this research, the formulation is based on the following optimization problem,
min f (x) =

kMB
[
ref

k
(x, t)]
2
(3-1)
where x is the vector of variables (PSSs gains and time constants);
ref
is the reference
angular speed;
k
(x, t) is the k-th synchronous machine angular speed after some
disturbance(s); t is time; MB is the set of synchronous machines equipped with PSS. In order
to take dierent operating conditions into account,the Eq. (3-1) may include an additional
summation which includes such conditions. That is,the Eq. (3-1) may be weighted to
formulate the problem by,
min f (x) =

iOC

i
{

kMB
[
ref

k
(t)]
2
} (3-2)
index OC is related to conditions of operation, and
i
is a weighting factor. Conventionally,

i=OC

i
= 1 (3-3)
In order to illustrate the PSS performance, in this paper the Eq. (3-2) is solved by a genetic
algorithm (GA), taking three operating points into account. Additionally, a multi-objective
formulation may be used to improve some other indexes, for instance, account for some power
ow limits through transmission lines, but it isnt objective in this document. Assuming a
conventional PSS with one wash-out block and two lead-lag blocks, which transfer function
is,
V
i

=
K
w
Ts
1 +sT
1 +T
1
s
1 +T
2
s
1 +T
3
s
1 +T
4
s
(3-4)
The following parameters are going to be estimated: (i) the wash-out gain,K
w
; (ii) the time
constant T
1
= T
3
. For simplicity, it is assumed that T = 7.5s, and T
2
= T
4
= 0.01s. Hence,
26 3 Optimal Tuning of Power System Stabilizers
each conventional PSS contributes with two parameters to the optimization problem. To
simplify, we consider T
1
= T
3
and T
2
= T
4
; so that, for each stabilizer, we are going to
estimate just the time constant T
1
and the gain K, because of the time constants T and
T
2
are previously selected. T ranging between 7.5 and 15s, is chosen to ensure a negligible
phase-shift and gain contributed by the wash-out block in the range of oscillating frequencies
of concern. T
2
ranging between 0.010 and 0.10s, is chosen from a consideration of physical
realization [49]. Therefore, for N stabilizers, we are going to estimate 2N parameters (N time
constants T
1
, and N gains K). It is emphasized that with this procedure robust stabilizers,
able to operate satisfactorily over a wide range of operating conditions, are obtained.
The optimization algorithm is concerned with calculating a vector x so as to optimize f(x);
x = x
1
, x
2
, ..., x
D
. D is the dimensionality of vector x. Ordinarily, the variables have an
allowed interval dened by their lower and upper bounds: x
jlow
, x
jupp
;j {1, ..., D}. The
initial population is selected uniform randomly between the lower (x
jlow
) and upper (x
jupp
)
bounds dened for each variable (x
j
). These bounds are specied by the user according to
the nature of the problem.
Genetic algorithms have been extensively used to solve optimization problems. Thus, it is
well known as a powerful method and it is utilized in this research [11, 38, 61]. For each
problem is important dene the following GAs parameters: (i) population size; (ii) crossover
probability; (iii) mutation probability; (iv) maximum number of generations. The Fig. 3-1
depicts a owchart of the proposed strategy.
3.2 Proposition 27
Start
Estimation of
Parameters by GA
optimizer.
Evaluate objective
function (f)
min f(x)
Transient Stability
Analysis.
End
no
yes
Initial Population.
Generate
disturbance.
Power Flow
(Initial operating
condition).
Electrical power
system.
Figure 3-1: Flowchart for the proposed coordination.
28 3 Optimal Tuning of Power System Stabilizers
3.3 Single Machine Innite Bus Power System (SMIB).
To illustrate the applicability of the proposed technique three dierent test power systems are
used, where stabilizers are tuned with the purpose of adding damping to power oscillations
and to preserve security. In SMIB, synchronous machine model without damper windings
with typical parameters are employed and equipped with a static excitation system.
To assess the eectiveness and robustness of the technique for tuning stabilizers, simulation
studies are carried out for three-phase faults and clearing faults applied in various buses
under dierent scenarios. Lets start showing the results obtained for the SMIB under four
operating conditions.
The SMIB is composed by a synchronous generator, static excitation system, conventional
power system stabilizer -CPSS, transformer, transmission line, load and innite bus, Fig.
3-2
The synchronous generator parameters are (p.u): D = 0, X
d
= 0.065, X
pd
= 0.025,
T
pd0
= 6.0, R
a
= 0, X
q
= 0.055, X
pq
= 0.02, T
pq0
= 0.535, H = 6.4.
Network equivalent: This varies under dierent operating conditions. R
e
= 0.
The pre-assigned static excitation system parameters are K
a
= 35 and T
a
= 0.015.
Taking into account the above considerations the PSSs parameters are: T = 7.5
and T
2
= 0.01 . The following bound are imposed over parameters K and T
1
:
[K
low
, T
1low
] = [0.002, 0.01] and [K
up
, T
1up
] = [0.75, 0.1].
Figure 3-2: Single-machine innite-bus power system.
The following GAs parameters have been used: (i) population size=30; (ii)
crossover probability=0.9; (iii) probability of mutation=0.1; (iv) maximum number of
generations=30.
3.3 Single Machine Innite Bus Power System (SMIB). 29
3.3.1 Study
In this application, there are two PSSs parameters (decision variables, x): (i) the gain K;
(ii) the time constant T
1
. Table 3-1 shows four operating conditions employed to calculate
the PSS parameters. The optimization problem is Eq. (3-2)
Table 3-1: Operating conditions.
Case 1 Case 2 Case 3 Case 4
P +jQ 0.5 +j0.30 0.5 +j0.30 0.75 +j0.40 0.5 j0.30
X
e
j0.175 j0.30 j0.10 j0.15
The estimated parameters become: K = 0.730 and T
1
= 0.0977
3.3.2 Simulations
A three-phase fault is applied for the dierent operating conditions. The fault is cleared
after 3 cycles and the original system is restored after the fault clearance. Fig. 3-3-3-4
show a comparison with and without PSS for case 2. Without PSS, an oscillatory instability
appears in the power system for a local oscillation mode which lacks of sucient damping
torque. With PSS, the system retrieves steady state conditions after fault clearance in a
time 0.6 seconds.
0 0.5 1 1.5 2 2.5
0.998
0.9985
0.999
0.9995
1
1.0005
1.001
1.0015
1.002
1.0025


A
n
g
u
l
a
r

s
p
e
e
d

(
p
.
u
)
Time (sec)
Angular SpeedCase 2.
ThreePhase Fault.
Gen 1 without pss
Gen 1 with pss
Figure 3-3: Angular speed-comparing with and without PSS.
Without PSS the active power ow through the line is completely oscillatory after the fault,
therefore this instantaneous value could jeopardize the line,Fig. 3-4
30 3 Optimal Tuning of Power System Stabilizers
0 0.5 1 1.5 2 2.5
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
A
c
t
i
v
e

p
o
w
e
r

f
l
o
w

(
p
.
u
)
Time(sec)
Electric Power Flow.
ThreePhase Fault.


Without Pss
With Pss
Figure 3-4: Active power ow: comparison with and without PSS.
Figures 3-5-3-7 depict the transient behavior of some representative signals after a three-
phase fault, where the stability of the system is maintained. The power oscillations are
eectively suppressed with the application of the conventional power system stabilizer. It is
observed that damping can be added satisfactorily through the proposed objective function.
0 0.5 1 1.5 2 2.5
0
0.2
0.4
0.6
0.8
1
1.2
1.4
A
c
t
i
v
e

p
o
w
e
r

f
l
o
w

(
p
.
u
)
Time(sec)
Electric Power Flow.
ThreePhase Fault.


Case 1
Case 2
Case 3
Case 4
Figure 3-5: Active power ow for each case.
3.3 Single Machine Innite Bus Power System (SMIB). 31
0 0.5 1 1.5 2 2.5
376.8
376.9
377
377.1
377.2
377.3
377.4
377.5
A
n
g
u
l
a
r

s
p
e
e
d

(
r
a
d
/
s
)
Time(sec)
Angular speed in Gen 1.
ThreePhase Fault.


Case 1
Case 2
Case 3
Case 4
Figure 3-6: Angular speed.
Thus, the estimated parameters obtained by the genetic algorithm, minimize the dierence
between the angular speed at time t and the reference speed after the fault clearance. That
is, the strategy is to push the postfault value of the angular speed to the prefault value.
0 0.5 1 1.5 2 2.5
0.2
0.1
0
0.1
0.2
0.3
0.4
0.5
D
w

(
r
a
d
/
s
e
c
)
Time (sec)
DwGen 1.
ThreePhase Fault.


Case 1
Case 2
Case 3
Case 4
Figure 3-7: Dierence between the angular speed and reference speed at time t.
32 3 Optimal Tuning of Power System Stabilizers
3.4 Multimachine Test System - New England power
system.
The power system model presented in the following is the New England power system model,
constituted by ten machines, thirty nine bus, and 46 transmission lines. All generators
are represented through a transient stability model, and they are equipped with a static
excitation system.
The diagram of this system is shown in the Fig. 3-8. Details of the system data are given
in [60].
01
02
03
05 04
06
07
09
08
10
29
9
28 26
25
8
10
12
27
37
24
36
38
13
11
2
39
35
34 14
15
16 32
31
33
30
4
5
21 22
6
23
7
17
18
19
20
1
3
Figure 3-8: New England Power System.
3.4.1 Study
With the purpose of illustrating the proposition, it is assumed that conventional PSSs are
installed at synchronous generators 1-8. The main objective is estimates simultaneously the
3.4 Multimachine Test System - New England power system. 33
set of optimal variables K and T
1
associated to the PSSs, under various system operating
conditions, by minimizing the objective function. Thus, for 8 generators, there are 16
parameters (decision variables, x) to be estimated.
The multi-machine system is composed by a synchronous generators, static excitation
systems, conventional power system stabilizers-PSSs, transformers, transmission lines, and
loads. The lead-lag structure of PSS controller is considered in this study. The PSSs
parameters used are wash-out time constant T = 7.5s and the time constant T
2
= 0.01s.
The PSSs variables to be estimated are the gain K and the lead time constant T
1
. The
fast exciter model contain the automatic voltage regulator (A.V.R) characterized by gain
K
A
= 50 p.u, time constant T
A
= 0.015s, maximum voltage regulator output V R
max
= 20
p.u, and minimum voltage regulator output V R
min
= 20 p.u.
In this case, robustness means the fact that the PSSs have to perform well against a wide
scenarios of the system parameters, loading conditions, disturbance size and location [62].
Three operating conditions are considered:
Lightly loading system-Case 70%.
Nominal Case-Case 100%.
Heavily loading of the system-Case 130%.
Taking into account the optimization problem Eq.(3-2), the PSSs parameters were tuned
optimally to improve the overall system dynamic stability, where a random load change in all
buses gives rise to the transient behavior. A normal distribution with zero mean is utilized
to generate the increment (decrement) in all buses. The variation is limited to a maximum
of 50%. To account for each operating condition into the objective functions, the same
weighted factors have been utilized (w
i
= 1/3).
After variables K and T
1
are estimated, the evaluation of the eectiveness and robustness
of PSS was performed under dierent operating conditions by applying three-phase fault in
several bus.
The design problem formulated into Eq.(3-2) is constrained to:
2 10
5
K 1
2 10
5
T
1
0.1
T
1
=T
3
T
2
=T
4
The following GAs parameters have been used: (i) population size=160; (ii)
crossover probability=0.9; (iii) probability of mutation=0.1; (iv) maximum number of
generations=50. Table 3-2 summarizes the estimated parameters for the eight generators,
according to the objective function.
34 3 Optimal Tuning of Power System Stabilizers
Table 3-2: Estimated parameters for the conventional PSSs.
K T
1
0.85748726 0.04679
0.51392467 0.09538521
0.88216716 0.09686705
0.27235732 0.09474711
0.42142959 0.08513015
0.75819379 0.09011494
0.72907428 0.09220556
0.65306817 0.09236284
3.4.2 Simulations
For performances evaluation, non linear time domain simulations were carried out using
matlab. A three-phase fault is applied in all cases aforementioned; the fault is cleared after
3 cycles, and after fault clearance the original system is restored.
Table 3-3 summarizes the 3 cases under study for three-phase faults on dierent buses.
Table 3-3: Operating Cases.
Case Buses
70% 14 27 37
100% 13 19 28
130% 11 22 39
The results are depicted in Fig. 3-9-3-11 for case 70%, in Fig. 3-12-3-14 for case 100%
and Fig. 3-15-3-17 for case 130%.
For the three cases with fault in dierent buses, the following details can be remarked:
In the faulted condition, the electrical variables of the power system drooped from a
steady state to new value that changes (for each time t), looking for a new operating
point as a result of generatores dynamic response. The simulations without PSSs show
that the damping is not enough and the system become oscillatory.
The conventional power system stabilizer PSS provided damping to the rotor
oscillations independently of the fault location. With the application of PSSs in
Fig.3-9-3-17 it is observed that the stability of the system is maintained and power
system oscillation is suppressed in a time lower than 1 second, in all the studied
operating conditions.
With the PSS inclusion the variations speed angular, position angular y active power
system of all machines are greatly reduced. This contributes to the machines do not
lose synchronism and the system remains stable.
3.4 Multimachine Test System - New England power system. 35
The proposed coordination enable to estimate the best PSSs parameters
simultaneously to ensure damping after a fault.
Non-linear time-domain simulations allow to validate the eectiveness and robustness
of the optimized PSS parameters under various operating conditions and under three
phase faults in various buses.
0 0.5 1 1.5 2 2.5 3
0.75
0.8
0.85
0.9
0.95
Time (sec)
A
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
r
a
d
)
Angular position in bus 7 and 8.
ThreePhase Fault at bus 14.


Gen 7 without PSS
Gen 8 without PSS
Gen 7 with PSS
Gen 8 with PSS
0 0.5 1 1.5 2 2.5 3
0.999
0.9995
1
1.0005
1.001
1.0015
1.002
Time (sec)
A
n
g
u
l
a
r

s
p
e
e
d

(
p
u
)
Angular speed in bus 3 and 4.
ThreePhase Fault at bus 14.


Gen 3 without PSS
Gen 4 without PSS
Gen 3 with PSS
Gen 4 with PSS
0 0.5 1 1.5 2 2.5 3
0.4
0.2
0
0.2
0.4
0.6
Time (sec)
D
w

(
r
a
d
/
s
)
DwGen 3 and 4.
ThreePhase Fault at bus 14.


Dw
3
without PSS
Dw
4
without PSS
Dw
3
with PSS
Dw
4
with PSS
0 0.5 1 1.5 2 2.5 3
2.5
3
3.5
4
4.5
5
5.5
6
6.5
Time (sec)
E
l
e
c
t
r
i
c

P
o
w
e
r

(
p
u
)
Electric Power in bus 5 and 6.
ThreePhase Fault at bus 14.


Bus 5 without PSS
Bus 6 without PSS
Bus 5 with PSS
Bus 6 with PSS
Figure 3-9: From left to right: Angular position gen 7 and 8, angular speed gen 3 and 4, w gen
3 and 4 and active electric power gen 5 and 6 to a fault in the bus 14-Case 70%.
36 3 Optimal Tuning of Power System Stabilizers
0 0.5 1 1.5 2 2.5 3
0.2
0
0.2
0.4
0.6
0.8
1
1.2
Time (sec)
A
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
r
a
d
)
Angular position in bus 9 and 10.
ThreePhase Fault at bus 27.


Gen 9 without PSS
Gen 10 without PSS
Gen 9 with PSS
Gen 10 with PSS
0 0.5 1 1.5 2 2.5 3
0.999
0.9995
1
1.0005
1.001
1.0015
1.002
Time (sec)
A
n
g
u
l
a
r

s
p
e
e
d

(
p
u
)
Angular speed in bus 5 and 6.
ThreePhase Fault at bus 27.


Gen 5 without PSS
Gen 6 without PSS
Gen 5 with PSS
Gen 6 with PSS
0 0.5 1 1.5 2 2.5 3
0.4
0.2
0
0.2
0.4
0.6
0.8
1
Time (sec)
D
w

(
r
a
d
/
s
)
DwGen 3 and 4.
ThreePhase Fault at bus 27.


Dw
3
without PSS
Dw
4
without PSS
Dw
3
with PSS
Dw
4
with PSS
0 0.5 1 1.5 2 2.5 3
0
1
2
3
4
5
6
7
Time (sec)
E
l
e
c
t
r
i
c

P
o
w
e
r

(
p
u
)
Electric Power in bus 1 and 2.
ThreePhase Fault at bus 27.


Bus 1 without PSS
Bus 2 without PSS
Bus 1 with PSS
Bus 2 with PSS
Figure 3-10: From left to right: Angular position gen 9 and 10, angular speed gen 5 and 6, w
gen 3 and 4 and active electric power gen 1 and 2 to a fault in the bus 27-Case 70%.
3.4 Multimachine Test System - New England power system. 37
0 0.5 1 1.5 2 2.5 3
0.2
0.3
0.4
0.5
0.6
0.7
Time (sec)
A
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
r
a
d
)
Angular position in bus 1 and 2.
ThreePhase Fault at bus 37.


Gen 1 without PSS
Gen 2 without PSS
Gen 1 with PSS
Gen 2 with PSS
0 0.5 1 1.5 2 2.5 3
0.999
0.9995
1
1.0005
1.001
1.0015
1.002
Time (sec)
A
n
g
u
l
a
r

s
p
e
e
d

(
p
u
)
Angular speed in bus 7 and 8.
ThreePhase Fault at bus 37.


Gen 7 without PSS
Gen 8 without PSS
Gen 7 with PSS
Gen 8 with PSS
0 0.5 1 1.5 2 2.5 3
0.2
0.1
0
0.1
0.2
0.3
0.4
0.5
Time (sec)
D
w

(
r
a
d
/
s
)
DwGen 3 and 4.
ThreePhase Fault at bus 37.


Dw
3
without PSS
Dw
4
without PSS
Dw
3
with PSS
Dw
4
with PSS
0 0.5 1 1.5 2 2.5 3
2
3
4
5
6
7
8
Time (sec)
E
l
e
c
t
r
i
c

P
o
w
e
r

(
p
u
)
Electric Power in bus 9 and 10.
ThreePhase Fault at bus 37.


Bus 9 without PSS
Bus 10 without PSS
Bus 9 with PSS
Bus 10 with PSS
Figure 3-11: From left to right: Angular position gen 1 and 2, angular speed gen 7 and 8, w gen
3 and 4 and active electric power gen 9 and 10 to a fault in the bus 37-Case 70%.
38 3 Optimal Tuning of Power System Stabilizers
0 0.5 1 1.5 2 2.5 3
0.75
0.8
0.85
0.9
0.95
1
Time (sec)
A
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
r
a
d
)
Angular position in bus 5 and 6.
ThreePhase Fault at bus 13.


Gen 5 without PSS
Gen 6 without PSS
Gen 5 with PSS
Gen 6 with PSS
0 0.5 1 1.5 2 2.5 3
0.999
0.9995
1
1.0005
1.001
1.0015
1.002
1.0025
Time (sec)
A
n
g
u
l
a
r

s
p
e
e
d

(
p
u
)
Angular speed in bus 1 and 2.
ThreePhase Fault at bus 13.


Gen 1 without PSS
Gen 2 without PSS
Gen 1 with PSS
Gen 2 with PSS
0 0.5 1 1.5 2 2.5 3
0.2
0
0.2
0.4
0.6
0.8
Time (sec)
D
w

(
r
a
d
/
s
)
DwGen 9 and 10.
ThreePhase Fault at bus 13.


Dw
9
without PSS
Dw
10
without PSS
Dw
9
with PSS.
Dw
10
with PSS
0 0.5 1 1.5 2 2.5 3
4
5
6
7
8
9
10
Time (sec)
E
l
e
c
t
r
i
c

P
o
w
e
r

(
p
u
)
Electric Power in bus 3.
ThreePhase Fault at bus 13.


Bus 3 without PSS
Bus 3 with PSS
Figure 3-12: From left to right: Angular position gen 5 and 6, angular speed gen 1 and 2, w gen
9 and 10 and active electric power gen 3 to a fault in the bus 13-Case 100%.
3.4 Multimachine Test System - New England power system. 39
0 0.5 1 1.5 2 2.5 3
0.2
0
0.2
0.4
0.6
0.8
1
1.2
Time (sec)
A
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
r
a
d
)
Angular position in bus 9 and 10.
ThreePhase Fault at bus 19.


Gen 9 without PSS
Gen 10 without PSS
Gen 9 with PSS
Gen 10 with PSS
0 0.5 1 1.5 2 2.5 3
0.9996
0.9998
1
1.0002
1.0004
1.0006
Time (sec)
A
n
g
u
l
a
r

s
p
e
e
d

(
p
u
)
Angular speed in bus 7 and 8.
ThreePhase Fault at bus 19.


Gen 7 without PSS
Gen 8 without PSS
Gen 7 with PSS
Gen 8 with PSS
0 0.5 1 1.5 2 2.5 3
0.2
0.1
0
0.1
0.2
0.3
0.4
0.5
Time (sec)
D
w

(
r
a
d
/
s
)
DwGen 1 and 2.
ThreePhase Fault at bus 19.


Dw
1
without PSS
Dw
2
without PSS
Dw
1
with PSS
Dw
2
with PSS
0 0.5 1 1.5 2 2.5 3
5
5.5
6
6.5
7
7.5
8
8.5
Time (sec)
E
l
e
c
t
r
i
c

P
o
w
e
r

(
p
u
)
Electric Power in bus 3 and 4.
ThreePhase Fault at bus 19.


Bus 3 without PSS
Bus 4 without PSS
Bus 3 with PSS
Bus 4 with PSS
Figure 3-13: From left to right: Angular position gen 9 and 10, angular speed gen 7 and 8, w
gen 1 and 2 and active electric power gen 3 and 4 to a fault in the bus 19-Case 100%.
40 3 Optimal Tuning of Power System Stabilizers
0 0.5 1 1.5 2 2.5 3
0.75
0.8
0.85
0.9
0.95
1
1.05
1.1
1.15
Time (sec)
A
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
r
a
d
)
Angular position in bus 3 and 4.
ThreePhase Fault at bus 28.


Gen 3 without PSS
Gen 4 without PSS
Gen 3 with PSS
Gen 4 with PSS
0 0.5 1 1.5 2 2.5 3
0.9995
1
1.0005
1.001
1.0015
1.002
Time (sec)
A
n
g
u
l
a
r

s
p
e
e
d

(
p
u
)
Angular speed in bus 9 and 10.
ThreePhase Fault at bus 28.


Gen 9 without PSS
Gen 10 without PSS
Gen 9 with PSS
Gen 10 with PSS
0 0.5 1 1.5 2 2.5 3
0.5
0
0.5
1
Time (sec)
D
w

(
r
a
d
/
s
)
DwGen 5 and 6.
ThreePhase Fault at bus 28.


Dw
5
without PSS
Dw
6
without PSS
Dw
5
with PSS.
Dw
6
with PSS
0 0.5 1 1.5 2 2.5 3
3
4
5
6
7
8
9
10
Time (sec)
E
l
e
c
t
r
i
c

P
o
w
e
r

(
p
u
)
Electric Power in bus 7 and 8.
ThreePhase Fault at bus 28.


Bus 7 without PSS
Bus 8 without PSS
Bus 7 with PSS
Bus 8 with PSS
Figure 3-14: From left to right: Angular position gen 3 and 4, angular speed gen 9 and 10, w
gen 5 and 6 and active electric power gen 7 and 8 to a fault in the bus 28-Case 100%.
3.4 Multimachine Test System - New England power system. 41
0 0.5 1 1.5 2 2.5 3
0.9
0.95
1
1.05
1.1
1.15
Time (sec)
A
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
r
a
d
)
Angular position in bus 7 and 8.
ThreePhase Fault at bus 11.


Gen 7 without PSS
Gen 8 without PSS
Gen 7 with PSS
Gen 8 with PSS
0 0.5 1 1.5 2 2.5 3
0.9995
1
1.0005
1.001
Time (sec)
A
n
g
u
l
a
r

s
p
e
e
d

(
p
u
)
Angular speed in bus 9 and 10.
ThreePhase Fault at bus 11.


Gen 9 without PSS
Gen 10 without PSS
Gen 9 with PSS
Gen 10 with PSS
0 0.5 1 1.5 2 2.5 3
0.4
0.3
0.2
0.1
0
0.1
0.2
0.3
Time (sec)
D
w

(
r
a
d
/
s
)
DwGen 5 and 6.
ThreePhase Fault at bus 11.


Dw
5
without PSS
Dw
6
without PSS
Dw
5
with PSS.
Dw
6
with PSS
0 0.5 1 1.5 2 2.5 3
2
4
6
8
10
Time (sec)
E
l
e
c
t
r
i
c

P
o
w
e
r

(
p
u
)
Electric Power in bus 1 and 2.
ThreePhase Fault at bus 11.


Bus 1 without PSS
Bus 2 without PSS
Bus 1 with PSS
Bus 2 with PSS
Figure 3-15: From left to right: Angular position gen 7 and 8, angular speed gen 9 and 10, w
gen 5 and 6 and active electric power gen 1 and 2 to a fault in the bus 11-Case 130%.
42 3 Optimal Tuning of Power System Stabilizers
0 0.5 1 1.5 2 2.5 3
0.8
0.85
0.9
0.95
1
1.05
Time (sec)
A
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
r
a
d
)
Angular position in bus 5 and 6.
ThreePhase Fault at bus 22.


Gen 5 without PSS
Gen 6 without PSS
Gen 5 with PSS
Gen 6 with PSS
0 0.5 1 1.5 2 2.5 3
0.9975
0.998
0.9985
0.999
0.9995
1
1.0005
1.001
1.0015
Time (sec)
A
n
g
u
l
a
r

s
p
e
e
d

(
p
u
)
Angular speed in bus 1 and 2.
ThreePhase Fault at bus 22.


Gen 1 without PSS
Gen 2 without PSS
Gen 1 with PSS
Gen 2 with PSS
0 0.5 1 1.5 2 2.5 3
1
0.5
0
0.5
1
1.5
2
Time (sec)
D
w

(
r
a
d
/
s
)
DwGen 3 and 4.
ThreePhase Fault at bus 22.


Dw
3
without PSS
Dw
4
without PSS
Dw
3
with PSS
Dw
4
with PSS
0 0.5 1 1.5 2 2.5 3
6
6.5
7
7.5
8
8.5
Time (sec)
E
l
e
c
t
r
i
c

P
o
w
e
r

(
p
u
)
Electric Power in bus 7.
ThreePhase Fault at bus 22.


Bus 7 without PSS
Bus 7 with PSS
Figure 3-16: From left to right: Angular position gen 5 and 6, angular speed gen 1 and 2, w gen
3 and 4 and active electric power gen 7 to a fault in the bus 22-Case 130%.
3.5 Mexican Interconnected Power System (MIPS). 43
0 0.5 1 1.5 2 2.5 3
0.4
0.5
0.6
0.7
0.8
0.9
1
Time (sec)
A
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
r
a
d
)
Angular position in bus 1 and 2.
ThreePhase Fault at bus 39.


Gen 1 without PSS
Gen 2 without PSS
Gen 1 with PSS
Gen 2 with PSS
0 0.5 1 1.5 2 2.5 3
0.9985
0.999
0.9995
1
1.0005
1.001
1.0015
1.002
Time (sec)
A
n
g
u
l
a
r

s
p
e
e
d

(
p
u
)
Angular speed in bus 5 and 6.
ThreePhase Fault at bus 39.


Gen 5 without PSS
Gen 6 without PSS
Gen 5 with PSS
Gen 6 with PSS
0 0.5 1 1.5 2 2.5 3
1.5
1
0.5
0
0.5
1
1.5
2
2.5
Time (sec)
D
w

(
r
a
d
/
s
)
DwGen 9 and 10.
ThreePhase Fault at bus 39.


Dw
9
without PSS
Dw
10
without PSS
Dw
9
with PSS.
Dw
10
with PSS
0 0.5 1 1.5 2 2.5 3
5
6
7
8
9
10
Time (sec)
E
l
e
c
t
r
i
c

P
o
w
e
r

(
p
u
)
Electric Power in bus 8.
ThreePhase Fault at bus 39.


Bus 8 without PSS
Bus 8 with PSS
Figure 3-17: From left to right: Angular position gen 1 and 2, angular speed gen 5 and 6, w gen
9 and 10 and active electric power gen 8 to a fault in the bus 39-Case 130%.
3.5 Mexican Interconnected Power System (MIPS).
In this paper, a simplication of the Mexican interconnected power system is used to test
the proposition. It consists of 190 buses and 46 synchronous machines, Fig.3-18 [63]. The
Mexican Interconnected System comprises an overall generating capacity of approximately
57 GW, and a main transmission system of about 91,311 km among 400 kV, 230 kV and 115
44 3 Optimal Tuning of Power System Stabilizers
kV lines. The MIPS is composed of seven interconnected regions including the NorthWestern
(NW), Northern (N), NorthEastern (NE), Western, (W), Central (C), SouthEastern (SE)
and Peninsular (P) areas. The system is characterized by a longitudinal conguration, with
geographical areas interconnected through long transmission lines, and remotely-located
generation resources from the major consumption centers like the metropolitan areas of
Mexico City (in the C region), Monterrey (in the NE region) and Guadalajara (in the W
region) [26]. In this paper, a dynamic equivalent of the grid in Fig.3-18 is utilized. The
subsystem on the right of the dotted line is considered as the system under study. Thus,
the subsystem on the left is the external one. There are ve frontier nodes (86, 140,142,
148 and 188), and six frontier lines (86-184, 140-141, 142-143, 148-143 and 188-187). Thus,
the equivalent electrical grid has ve ctitious generators at nodes 86, 140, 142, 148 and
188. PSSs are installed in the twelve generators of the internal system. All generators
are represented through a transient stability model, and they are equipped with a static
excitation system.
Figure 3-18: 190 buses-46 generators power system.
3.5 Mexican Interconnected Power System (MIPS). 45
3.5.1 Study
In this application, there are two PSSs parameters (decision variables, x) per each equivalent
generator: (i) the gain K; (ii) the time constant T
1
. Thus, for twelve equivalent generators,
there are 24 parameters to be estimated.
A random load change in all buses gives rise to the transient behavior. A normal distribution
with zero mean is utilized to generate the increment (decrement) in all buses. The variation
is limited to a maximum of 50%. The disturbance lasts for 0.12 s and then it is eliminated;
the studied time is 2s. Fig. 3-1 depicts a owchart of the followed strategy to calculate an
optimal solution.
Three operating points are taken into account:
Nominal case-Case 1 100% [64].
Heavily loading of the system-Case 2 140%.
Lightly loading system-Case 3 70%.
To account for each operating condition into the objective functions, the same weighted
factors have been utilized (w
i
= 1/3), eqs. (2)-(3). Table 3-4 summarizes the estimated
parameters for the twelve generators at the internal system, according to the objective
function. These values are the average after 25 runs, estimated through the GA with
the parameters above mentioned. Columns 1-2 in Table 3-4 are parameters associated
to the conventional PSS model. In the case of conventional PSSs, it was assumed that the
wash-out time constants T = 7.5s and the time constants T
2
= T
4
= 0.01s. During the
optimization procedure, the following bounds are taken into account: (i) wash-out gain, K
: [2.5e 4, 1e 3]; (ii) time constants, T
1
= T
3
= [1e 2, 0.1].
Table 3-4: Estimated parameters for the conventional PSSs.
K T
1
0.000999 0.0989
0.000986 0.0303
0.000997 0.0998
0.00033 0.0998
0.001 0.1
0.001 0.1
0.000995 0.076
0.00025 0.0899
0.00025 0.0871
0.000973 0.1
0.001 0.0998
0.000992 0.0102
46 3 Optimal Tuning of Power System Stabilizers
3.5.2 Simulations
Figures 3-19-3-21 depict the transient behavior of some representative signals after a three-
phase fault at buses 172-Fig.3-19 for the Case 1; 168- Fig.3-20 for the Case 2; and 144-
Fig.3-21 for the Case 3. Bus 39 is selected as the slack bus.
Figure 3-19: Case 1: From top to bottom (i) electrical torque 40; (ii) angular speed 28; (iii) angular
position 37 (referred to slack), after a three-phase fault at bus 172.
3.5 Mexican Interconnected Power System (MIPS). 47
Figure 3-20: Case 2: From top to bottom (i) electrical torque 38; (ii) angular speed 37; (iii) angular
position 40 (referred to slack), after a three-phase fault at bus 168
48 3 Optimal Tuning of Power System Stabilizers
Figure 3-21: Case 3: From top to bottom (i) electrical torque 39; (ii) angular speed 30; (iii) angular
position 29 (referred to slack), after a three-phase fault at bus 144.
These results exhibit the non-linear behavior of the some real electrical powers, angular
velocities, and the rotor positions, after a sudden three-phase fault applied at nodes 172-
168-144, taken into account the aforementioned operating conditions. It is noteworthy that
robust stabilization is attained through the coordinated stabilizers. Thus, the problem of
selecting the PSSs parameters in order to enhance the damping of power oscillations for a set
of operating conditions is formulated as an optimization problem. A systematic procedure
3.5 Mexican Interconnected Power System (MIPS). 49
based on an optimal criterion has been proposed. It leads to robust stabilizers able to have
a suitable performance under dierent operating environments and network structures. The
numerical examples exhibit the coordination of PSSs; a notorious improvement of the power
system transient behavior for dierent operating requirements is reached.
4 PSSs Testing on a Real Time
Environment
Real time power system simulation has been used with the purpose to corroborate the
dynamic behaviour of the designed PSSs. The real time simulation is used to illustrate the
eectiveness of the proposed PSSs coordination. It is worth noting that this tool represents
a powerful strategy to test conventional and novel controllers, in order to assess its feasibility.
4.1 dSPACE
The dSPACE real time hardware is an obvious choice when working within the
Matlab/Simulink environment. The Matlab/Simulink models can be implemented and tested
in real-time.
To work with dSPACE systems, it is required to rst generate a real time application.
Generating a real time application is a successive process, which is dened below [54].
Creating a Simulink model: A Simulink model is created using Matlab and Simulink.
Instead of programming C code manually, we can implement the control algorithm
graphically using Simulink blocks. The models are saved as MDL les.
Specifying RTI I/O interfaces: RTI is the link between dSPACE hardware and the
Matlab/Simulink software. It acts as a driver to the dSPACE hardware. RTI is
integrated into the Simulink model like any other Simulink blocks. To connect the
simulation model to the physical world, it is necessary to introduce I/O interfaces into
the model. These allow to replace parts of the simulated model with real hardware.
dSPACEs RTI (Real Time Interface) blocks provide I/O interfaces for accessing the
dSPACE hardware.
Generating C code: We can build the model created with Simulink and RTI blocks
using the Real Time Workshop. The Real Time Workshop generates C code from the
model automatically.
Compiling and linking the real time application: The cross compiler environment
compiles the generated C code and links the object les and libraries into an executable
application for the real time processor.
4.2 Experimental Results 51
As a result of the above process, a real time application is generated,which consists of four
les [65]:
PPC-Program code to be downloaded to the hardware.
MAP-Map le with address information of variables.
TRC-Variable description le used by Control Desk.
SDF-System description le with references to the PPC, MAP,and TRC le. In Control
Desk, drag and drop this le to the dSPACE board to start the real time application
and load the variables contained in the related TRC le.
Finally, Control Desk, an experimentation tool, is used to control, tune and monitor the
running process. With the software Control desk (dSPACE) a virtual instrument panel is
built. It enables the operator to change parameters and monitor signals in realtime without
regenerating the code. In addition, Control desk displays time histories of any variable being
used by the application.
4.2 Experimental Results
For simulations in real time two plataforms are used: Matlab/Simulink and dSPACE. The
power system employed is the New England and the PSSs parameters are those described
in Chapter 3. These models were built in Simulink to generate the simulation in dSPACE
real time hardware; the dSPACE station is DS1104. The signals obtained through dSPACE
are observed by the oscilloscope, Fig.4-1.
52 4 PSSs Testing on a Real Time Environment
Figure 4-1: Experiment with dSPACE.
For performances evaluation, non linear time domain simulations were carried out using
dSPACE. A three-phase fault is applied for all cases shown in Table 4-1; the fault is cleared
after 3 cycles and after fault clearance the original system is restored. In this case, robustness
means the fact that the PSSs must to perform properly under a wide scenarios of the system
parameters, loading conditions, disturbance size and location.
Table 4-1: Operating Cases.
Case Buses
Case70% 2 20 25
Case100% 13 16 28
Case130% 11 36 39
Fig. 4-2-4-10 depict the transient behaviour of some representative signal as angular speed
deviation w, angular speed w, active electric power pelec and angular position afther a
three-phase fault. The signals have been scaled with several factors to be observed on the
oscilloscope.
4.2 Experimental Results 53
Figure 4-2: Case 70%: Three-phase fault in bus 2. CH1=w
gen8
, CH2=w
gen2
, CH3=pelec
gen10
, CH4=
3
.
Figure 4-3: Case 70%: Three-phase fault in bus 20. CH1=w
gen7
, CH2=w
gen3
, CH3=pelec
gen5
, CH4=
2
.
54 4 PSSs Testing on a Real Time Environment
Figure 4-4: Case 70%: Three-phase fault in bus 25. CH1=w
gen8
, CH2=w
gen8
, CH3=pelec
gen9
, CH4=
6
.
Figure 4-5: Case 100%: Three-phase fault in bus 13. CH1=w
gen1
, CH2=w
gen2
, CH3=pelec
gen4
,
CH4=
6
.
4.2 Experimental Results 55
Figure 4-6: Case 100%: Three-phase fault in bus 16. CH1=w
gen2
, CH2=w
gen1
, CH3=pelec
gen3
,
CH4=
10
.
Figure 4-7: Case 100%: Three-phase fault in bus 28. CH1=w
gen5
, CH2=w
gen9
, CH3=pelec
gen8
,
CH4=
3
.
56 4 PSSs Testing on a Real Time Environment
Figure 4-8: Case 130%: Three-phase fault in bus 11. CH1=w
gen8
, CH2=w
gen10
, CH3=pelec
gen1
,
CH4=
3
.
Figure 4-9: Case 130%: Three-phase fault in bus 36. CH1=w
gen7
, CH2=w
gen6
, CH3=pelec
gen5
,
CH4=
8
.
4.2 Experimental Results 57
Figure 4-10: Case 130%: Three-phase fault in bus 39. CH1=w
gen4
, CH2=w
gen5
, CH3=pelec
gen6
,
CH4=
2
.
From these simulations the following may be concluded:
By means of time simulation is observed the dynamic behaviour, in this case the
PSSs behaviour provide damping to system oscillations over a wide range of loading
conditions and with three-phase fault in dierent buses. This implies robustness respect
to the operating condition.
dSPACE Simulations results corroborate the eectiveness of the proposed coordination.
The dSPACE is a hardware easy to use and ecient. Its interface is accessible so that
experiments can be developed under dierent conditions.
Real-time simulations have gained importance due to the need to do devices proofs
with real data without aecting system operation in the control centers.
Real time simulations represent a powerful tool to design novel controllers.
5 Conclusions and future work
Based on the literature revision, the analysis realized,the proposed coordination and the
simulations obtained the following may be concluded:
The combination of excitation and system parameters under certain loading conditions
can introduce negative damping. In order to oset this eect and to improve system
damping in general, articial means of producing torques in phase with the speed are
introduced called supplementary stabilizing signals. The controllers used to generate
these signals are known as power system stabilizers.
Power System Stabilizer (PSS) is a supplementary control in order to improve the
operation and add damping to help in attenuate power oscillations; they are able to
extend the stability limit by enhancing the system damping.
The stabilizers robustness under all operating conditions is of concern and their
interaction must be considered to obtain their adequate design.
Conventional optimization algorithms may experience convergence diculties due to
convexity problems or non-smooth characteristics.
Genetic Algorithms (GA) are stochastic search techniques considered global search
methods based on the mechanism of natural selection and natural genetics. GA acts
as a biological metaphor and tries to emulate some of the processes observed in natural
evolution, for instance plants and animals.
The genetic algorithms are good choice for PSSs tuning. For this problem it is
important to dene the following GAs parameters: (i) population size, (ii) crossover
probability, (iii) mutation probability, (iv) maximum number of generations.
The proposed methodology estimates the best parameters K and T
1
simultaneously.
The eectiveness and robustness of the proposed coordination is validated by means of
non-linear simulations in the time domain and was evaluated in three power systems:
Single machine innite bus, New England and Mexican Interconnected. It shows that
the oscillations are suppressed independent of the operation scenarios and the bus
where a three-phase fault is simulated.
59
The estimated parameters obtained by the genetic algorithm K and T
1
, minimize the
dierence between the angular speed at time t and the reference speed after the fault
clearance. That is, the strategy is to push the postfault value of the angular speed to
the prefault value.
The real time simulations are realized through the dSPACE order to validate the
performance of the designed PSSs. The PSS provides damping to systems oscillations
under faults in dierent buses and operating conditions, this corroborates the
eectiveness and robustness of proposed coordination.
Real-time simulations have gained importance due to the need to carry out devices
proofs with real data without aecting system operation in the control centers and
represent a powerful tool to design novel controllers.
As future work the following is proposed.
With the proposed methodology realize to estimate the PSSs parameters through GA,
including constraints related to the PSSs output voltage limits, exciter voltage limits
and turbine model.
Estimate the best PSSs parameters T and T
2
together K and T
1
.
Realize simulations with others heuristic methods and dierent PSS.
Design physical PSS based on Digital Signal Processor (DSP) or microcontrollers in
order to close the loop within the real time environment.
Feedback phasor measurement unit (PMU) signals into the PSSs under the real time
environment.
Bibliography
[1] J. Chow, J. Sanchez-Gasca, H. Ren, and S. Wang, Power system damping controller
design-using multiple input signals, Control Systems, IEEE, vol. 20, no. 4, pp. 82 90,
aug 2000.
[2] P. Kundur, Power system stability and control, N. J. Balu and M. G. Lauby, Eds.
McGraw-Hill, 1994.
[3] G. Rogers, Power system oscillations, ser. Kluwer international series in engineering
and computer science: Power electronics & power systems, K. Academic, Ed. Kluwer
Academic, 2000.
[4] J.-H. Peng, N.-K. Nair, A. Maryani, and A. Ahmad, Adaptive power system stabilizer
tuning technique for damping inter-area oscillations, in Power and Energy Society
General Meeting, 2010 IEEE, july 2010, pp. 1 6.
[5] E. Larsen and D. Swann, Applying power system stabilizers part I: General concepts,
Power Apparatus and Systems, IEEE Transactions on, vol. PAS-100, no. 6, pp. 3017
3024, june 1981.
[6] , Applying power system stabilizers part II: Performance objectives and tuning
concepts, Power Apparatus and Systems, IEEE Transactions on, vol. PAS-100, no. 6,
pp. 3025 3033, june 1981.
[7] , Applying power system stabilizers part III: Practical considerations, Power
Apparatus and Systems, IEEE Transactions on, vol. PAS-100, no. 6, pp. 3034 3046,
june 1981.
[8] G. Andersson, P. Donalek, R. Farmer, N. Hatziargyriou, I. Kamwa, P. Kundur,
N. Martins, J. Paserba, P. Pourbeik, J. Sanchez-Gasca, R. Schulz, A. Stankovic,
C. Taylor, and V. Vittal, Causes of the 2003 major grid blackouts in north america
and europe, and recommended means to improve system dynamic performance, Power
Systems, IEEE Transactions on, vol. 20, no. 4, pp. 1922 1928, nov. 2005.
[9] A. Dysko, W. Leithead, and J. OReilly, Enhanced power system stability by
coordinated PSS design, Power Systems, IEEE Transactions on, vol. 25, no. 1, pp.
413 422, feb. 2010.
Bibliography 61
[10] H. Yassami, A. Darabi, and S. Raei, Power system stabilizer design using strength
pareto multi-objective optimization approach, Electric Power Systems Research,
vol. 80, no. 7, pp. 838 846, 2010.
[11] Y. Abdel-Magid and M. Abido, Optimal multiobjective design of robust power system
stabilizers using genetic algorithms, Power Systems, IEEE Transactions on, vol. 18,
no. 3, pp. 1125 1132, aug. 2003.
[12] M. Abido, Robust design of multimachine power system stabilizers using simulated
annealing, Energy Conversion, IEEE Transactions on, vol. 15, no. 3, pp. 297 304, sep
2000.
[13] M. Abido and Y. Abdel-Magid, Robust design of multimachine power system
stabilisers using tabu search algorithm, Generation, Transmission and Distribution,
IEE Proceedings-, vol. 147, no. 6, pp. 387 394, nov 2000.
[14] M. Dubey and A. Dubey, Simultaneous stabilization of multimachine power system
using genetic algorithm based power system stabilizers, in Universities Power
Engineering Conference, 2006. UPEC 06. Proceedings of the 41st International, vol. 2,
sept. 2006, pp. 426 431.
[15] R. Jabr, B. Pal, N. Martins, and J. Ferraz, Robust and coordinated tuning of power
system stabiliser gains using sequential linear programming, Generation, Transmission
Distribution, IET, vol. 4, no. 8, pp. 893 904, august 2010.
[16] S. Mishra, M. Tripathy, and J. Nanda, Multi-machine power system stabilizer design
by rule based bacteria foraging, Electric Power Systems Research, vol. 77, no. 12, pp.
1595 1607, 2007.
[17] T. Nguyen and R. Gianto, Optimal design for control coordination of power system
stabilisers and exible alternating current transmission system devices with controller
saturation limits, Generation, Transmission Distribution, IET, vol. 4, no. 9, pp. 1028
1043, september 2010.
[18] B. Pal and B. Chaudhuri, Robust control in power systems, Springer, Ed. Springer,
2005.
[19] K. Sebaa, H. Gueguen, and M. Boudour, Mixed integer non-linear programming via the
cross-entropy approach for power system stabilisers location and tuning, Generation,
Transmission Distribution, IET, vol. 4, no. 8, pp. 928 939, august 2010.
[20] N. Sumathi, M. Selvan, and N. Kumaresan, A hybrid genetic algorithm based
power system stabilizer, in Intelligent and Advanced Systems, 2007. ICIAS 2007.
International Conference on, nov. 2007, pp. 876 881.
62 Bibliography
[21] E. Viveros, G. Taranto, and D. Falcao, Coordinated tuning of AVRs and PSSs by
multiobjective genetic algorithms, in Intelligent Systems Application to Power Systems,
2005. Proceedings of the 13th International Conference on, nov. 2005, p. 6 pp.
[22] P. Zhang and A. Coonick, Coordinated synthesis of PSS parameters in multi-machine
power systems using the method of inequalities applied to genetic algorithms, Power
Systems, IEEE Transactions on, vol. 15, no. 2, pp. 811 816, may 2000.
[23] J. Chow, G. Boukarim, and A. Murdoch, Power system stabilizers as undergraduate
control design projects, Power Systems, IEEE Transactions on, vol. 19, no. 1, pp. 144
151, feb. 2004.
[24] T. Hussein, M. Saad, A. Elshafei, and A. Bahgat, Damping inter-area modes of
oscillation using an adaptive fuzzy power system stabilizer, in Control and Automation,
2008 16th Mediterranean Conference on, june 2008, pp. 368 373.
[25] G. Ramakrishna and O. Malik, Adaptive PSS using a simple on-line identier and
linear pole-shift controller, Electric Power Systems Research, vol. 80, pp. 406 416,
2010.
[26] M. Ramirez-Gonzalez, R. Castellanos B, and O. Malik, Application of simple fuzzy
PSSs for power system stability enhancement of the mexican interconnected system,
in Power and Energy Society General Meeting, 2010 IEEE, july 2010, pp. 1 8.
[27] G. Chen, O. Malik, G. Hope, Y. Qin, and G. Xu, An adaptive power system stabilizer
based on the self-optimizing pole shifting control strategy, Energy Conversion, IEEE
Transactions on, vol. 8, no. 4, pp. 639 645, dec 1993.
[28] K. El-Metwally, N. Rao, O. Malik, and G. Ramakrishna, Application of a neural
network as an integrated excitation controller, Electric Power Systems Research,
vol. 42, no. 2, pp. 121 126, 1997.
[29] D. A. Pierre, A perspective on adaptive control of power systems, Power Systems,
IEEE Transactions on, vol. 2, no. 2, pp. 387 395, may 1987.
[30] Y.-Y. Hsu and K.-L. Liou, Design of self-tuning PID power system stabilizers for
synchronous generators, Energy Conversion, IEEE Transactions on, vol. EC-2, no. 3,
pp. 343 348, sept. 1987.
[31] K. Astrom and B. Wittenmark, Adaptive Control, Addison-Wesley, Ed. Addison-
Wesley, 1995.
[32] A. Khodabakhshian and R. Hemmati, Robust decentralized multi-machine power
system stabilizer design using quantitative feedback theory, International Journal of
Electrical Power; Energy Systems, vol. 41, pp. 112119, 2012.
Bibliography 63
[33] V. S. Peric, A. T. Saric, and D. I. Grabez, Coordinated tuning of power system
stabilizers based on fourier transform and neural networks, Electric Power Systems
Research, vol. 88, pp. 7888, 2012.
[34] S. Abd-Elazim and E. Ali, Coordinated design of PSSs and SVC via bacteria foraging
optimization algorithm in a multimachine power system, International Journal of
Electrical Power; Energy Systems, vol. 41, no. 1, pp. 4453, 2012.
[35] A. El-Zonkoly, A. Khalil, and N. Ahmied, Optimal tunning of lead-lag and fuzzy
logic power system stabilizers using particle swarm optimization, Expert Systems with
Applications, vol. 36, pp. 20972106, 2009.
[36] S. Panda, Robust coordinated design of multiple and multi-type damping controller
using dierential evolution algorithm, International Journal of Electrical Power;
Energy Systems, vol. 33, no. 4, pp. 10181030, 2011.
[37] Y. Abdel-Magid, M. Abido, S. Al-Baiyat, and A. Mantawy, Simultaneous stabilization
of multimachine power systems via genetic algorithms, Power Systems, IEEE
Transactions on, vol. 14, no. 4, pp. 14281439, nov 1999.
[38] A. Do Bomm, G. Taranto, and D. Falcao, Simultaneous tuning of power system
damping controllers using genetic algorithms, Power Systems, IEEE Transactions on,
vol. 15, no. 1, pp. 163169, feb 2000.
[39] K. A. Folly, Performance evaluation of power system stabilizers based on population-
based incremental learning (PBIL) algorithm, International Journal of Electrical
Power; Energy Systems, vol. 33, no. 7, pp. 12791287, 2011.
[40] A. Hasanovic and A. Feliachi, Genetic algorithm based inter-area oscillation damping
controller design using MATLAB, in Power Engineering Society Summer Meeting,
2002 IEEE, vol. 3, july 2002, pp. 11361141 vol.3.
[41] Y. Chompoobutrgool, L. Vanfretti, and M. Ghandhari, Survey on power system
stabilizers control and their prospective applications for power system damping using
synchrophasor-based wide-area systems, European Transactions on Electrical Power,
vol. 21, no. 8, pp. 20982111, 2011.
[42] F. Demello and C. Concordia, Concepts of synchronous machine stability as aected by
excitation control, Power Apparatus and Systems, IEEE Transactions on, vol. PAS-88,
no. 4, pp. 316 329, april 1969.
[43] A. Rahim and S. Nassimi, Synchronous generator damping enhancement through
coordinated control of exciter and svc, Generation, Transmission and Distribution,
IEE Proceedings-, vol. 143, no. 2, pp. 211 218, mar 1996.
64 Bibliography
[44] Y. Abdel-Magid, M. Abido, and A. Mantawy, Robust tuning of power system
stabilizers in multimachine power systems, in Power Engineering Society Winter
Meeting, 2000. IEEE, vol. 2, 2000, p. 1425 vol.2.
[45] M. Abido, Simulated annealing based approach to PSS and FACTS based stabilizer
tuning, International Journal of Electrical Power ; Energy Systems, vol. 22, pp. 247
258, 2000.
[46] J. M. Ramrez and I. Castillo, PSS and FDS simultaneous tuning, Electric Power
Systems Research, vol. 68, pp. 33 40, 2004.
[47] P. Anderson, A. Fouad, I. of Electrical, and E. Engineers, Power system control and
stability, ser. IEEE Press power engineering series. IEEE Press, 2003.
[48] J. J. More, The Levenberg-Marquardt algorithm: implementation and theory, W. GA,
Ed. Springer, 1977.
[49] P. Kundur, M. Klein, G. Rogers, and M. Zywno, Application of power system
stabilizers for enhancement of overall system stability, Power Systems, IEEE
Transactions on, vol. 4, no. 2, pp. 614 626, may 1989.
[50] K. Wang, C. Chung, C. Tse, and K. Tsang, Multimachine eigenvalue sensitivities of
power system parameters, in Power Engineering Society Winter Meeting, 2000. IEEE,
vol. 2, 2000, p. 1434 vol.2.
[51] H.-K. Nam, Y.-K. Kim, K.-S. Shim, and K. Lee, A new eigen-sensitivity theory of
augmented matrix and its applications to power system stability analysis, Power
Systems, IEEE Transactions on, vol. 15, no. 1, pp. 363 369, feb 2000.
[52] H. Shayeghi, H. Shayanfar, A. Safari, and R. Aghmasheh, A robust PSSs design using
PSO in a multi-machine environment, Energy Conversion and Management, vol. 51,
pp. 696 702, 2010.
[53] D. Goldberg, Genetic algorithms in search, optimization, and machine learning, ser.
Articial Intelligence, Addison-Wesley, Ed. Addison-Wesley, 1989.
[54] dSPACE, ControlDesk-Experiment Guide for Release 6.5, dSPACE, Ed. c 2012
dSPACE GmbH, 2009.
[55] J. Langston, M. Steurer, S. Woodru, T. Baldwin, and J. Tang, A generic real-time
computer simulation model for superconducting fault current limiters and its application
in system protection studies, Applied Superconductivity, IEEE Transactions on, vol. 15,
no. 2, pp. 2090 2093, june 2005.
Bibliography 65
[56] A. Mackay, S. Galloway, C. Booth, and J. McDonald, Real-time assessment of relay
protection schemes on integrated full electric propulsion systems, in Electric Ship
Technologies Symposium, 2005 IEEE, july 2005, pp. 230 236.
[57] K. Lian and P. Lehn, Real-time simulation of voltage source converters based on time
average method, Power Systems, IEEE Transactions on, vol. 20, no. 1, pp. 110 118,
feb. 2005.
[58] Y. Li, D. Vilathgamuwa, and P. C. Loh, Design, analysis, and real-time testing of a
controller for multibus microgrid system, Power Electronics, IEEE Transactions on,
vol. 19, no. 5, pp. 1195 1204, sept. 2004.
[59] M. Steurer, F. Bogdan, W. Ren, M. Sloderbeck, and S. Woodru, Controller and power
hardware-in-loop methods for accelerating renewable energy integration, in Power
Engineering Society General Meeting, 2007. IEEE, june 2007, pp. 14.
[60] K. Padiyar, Power System Dynamics: Stability and Control, B. Publications, Ed. BS
Publications, 1996.
[61] M. Abido and Y. Abdel-Magid, Coordinated design of robust excitation and TCSC-
based damping controllers, in Power Engineering Society General Meeting, 2003,
IEEE, vol. 3, july 2003, p. 4 vol. 2666.
[62] H. E. Talaat, A. Abdennour, and A. A. Al-Sulaiman, Design and experimental
investigation of a decentralized GA-optimized neuro-fuzzy power system stabilizer,
International Journal of Electrical Power & Energy Systems, vol. 32, no. 7, pp. 751
759, 2010.
[63] J. M. Ramirez, R. E. Correa-Gutierrez, and N. J. Castrillon-Gutierrez, A study
on multiband PSS coordination, International Journal of Emerging Electric Power
Systems, vol. Volume 10, pp. 118, 2010.
[64] A. Messina, J. Ramirez, and J. Canedo C., An investigation on the use of power
system stabilizers for damping inter-area oscillations in longitudinal power systems,
Power Systems, IEEE Transactions on, vol. 13, no. 2, pp. 552 559, may 1998.
[65] J. Reddy and M. J. Kishore, Real time implementation of H

loop shaping robust

pss for multimachine power system using dspace, International Journal of Electrical
Power; Energy Systems, vol. 33, pp. 1750 1759, 2011.