International Journal of Electronics and Electrical Engineering 6 2012

Tuning of Power System Stabilizers in a Multi-

Machine Power System using C-Catfish PSO
M. H. Moradi, S. M. Moosavi, A. R. Reisi

Abstract—The main objective of this paper is to investigate the Thus, maintaining reliable operation of the power system
enhancement of power system stability via coordinated tuning of [2]. Several approaches based on modern control theory have
Power System Stabilizers (PSSs) in a multi-machine power system. been applied to power system stabilizer design problems.
The design problem of the proposed controllers is formulated as an
These include optimal control, adaptive control, variable
optimization problem. Chaotic catfish particle swarm optimization
(C-Catfish PSO) algorithm is used to minimize the ITAE objective structure control and intelligent control [3]–[5]. Novel
function. The proposed algorithm is evaluated on a two-area, 4- intelligent control design methods such as fuzzy logic
machines system. The robustness of the proposed algorithm is controllers [6] and artificial neural network controllers [7]
verified on this system under different operating conditions and have been used as the PSSs. Unlike other classical control
applying a three-phase fault. The nonlinear time-domain simulation methods, fuzzy logic and neural network controllers are
results and some performance indices show the effectiveness of the
model-free controllers; i.e. they do not require an exact
proposed controller in damping power system oscillations and this
novel optimization algorithm is compared with particle swarm mathematical model of the controlled system. Moreover, speed
optimization (PSO). and robustness are the most significant properties in
comparison to other classical schemes. H ∞ optimization
Keywords—Power system stabilizer, C-Catfish PSO, ITAE techniques [8] have been also applied to the robust PSS design
objective function, Power system control, Multi-machine power problem. However, the importance and difficulties in the
system
selection of weighting functions of the H ∞ optimization have
I. INTRODUCTION been reported. Recently, intelligent optimization techniques
like genetic algorithms (GA) [9]–[12], Tabu search [13],
I N1950s and to 1960s, many power-generating plants were
equipped with continuously acting automatic voltage
simulated annealing [14], evolutionary programming [15] and
rule based bacteria foraging [16] have been applied for PSS
regulators (AVRs). As the number of power plants with AVRs
parameter optimization. These evolutionary algorithms are
grew, it became apparent that the high performance of these
heuristic population-based search procedures that incorporate
voltage regulators had a destabilizing effect on the power
random variation and selection operators. Although, these
system. Power oscillations of small magnitude and low
methods seem to be good methods for the solution of PSS
frequency often persisted for long periods of time. Since the
parameter optimization problem, However, when the system
development of interconnected large electric power systems,
has a highly epistatic objective function (i.e. where parameters
there have been spontaneous system oscillations at very low
being optimized are highly correlated), and number of
frequencies in order of 0.2–3.0 Hz. Once started, they would
parameters to be optimized is large, then they have regarded
continue for a long period of time. In some cases, they
efficiency to obtain global optimum solution and also
continue to grow, causing system separation if no adequate
simulation process use a lot of computing time. An algorithm
damping is available. Moreover, low frequency oscillations
for computerized automatic tuning of power system stabilizers
present limitations on the power-transfer capability. In some
has been presented in [17]. Particle swarm optimization [18]
cases, this presented a limitation on the amount of the power to
has been applied for PSSs parameter optimization in 3-
be transmitted within the system [1]. Power system stabilizers
machines power system. In this paper, a comprehensive
(PSSs) are auxiliary control devices on synchronous
assessment of the effects of the coordinated design of PSSs
generators, used in conjunction with their excitation systems to
stabilizers on power system stability enhancement has been
provide control signals toward enhancing the system damping
carried out in multi-machine power system. Chaotic catfish
and extending power transfer limits.
particle swarm optimization (C-Catfish PSO) algorithm is used
to minimize the ITAE objective function for a two-area, 4-
machines system under different operating conditions and
applying a three-phase fault. Unlike the other heuristic
techniques, C-Catfish PSO has a flexible and well-balanced
mechanism to enhance the global and local exploration
M. H. Moradi is a lecturer at Bu Ali Sina University in Hamedan , Iran abilities. The nonlinear time-domain simulation results and
(phone:303-555-5555;fax: 303-555-5555; e-mail: mh_moradi@yahoo.co.uk). some performance indices show the effectiveness of the
S. M. Moosavi is M.SC. student at Bu Ali Sina University in Hamedan,
Iran (e-mail: masood_moosavi2100@yahoo.com). proposed controller in damping power system oscillations.
A. R. Reisi is M.SC. student at Bu Ali Sina University in Hamedan, Iran
(e-mail: ali_reza_reisi@yahoo.com).

120
 International Journal of Electronics and Electrical Engineering 6 2012

II. C-CATFISH PSO ALGORITHM Where itermax is the maximum number of iterations and
PSO is one of the optimization techniques and belongs to iteration is the current number of iteration. (3) presents that the
evolutionary computation techniques [19]. The method has inertia weight is updated, considering Wmax and Wmin are the
been developed through a simulation of simplified social initial and final weights, respectively. The underlying idea for
models. The features of the method are as follows: the development of Catfish PSO was derived from the catfish
1) The method is based on researches on swarms such as effect observed when catfish were introduced into large
fish schooling and bird flocking. holding tanks of sardines [22]. The catfish in competition with
2) It is based on a simple concept. Therefore, the the sardines, stimulate renewed movement amongst the
computation time is short and it requires few memories. sardines. Similarly, the introduced catfish particles stimulate a
According to the research results for bird flocking, birds are renewed search by the other ‘‘sardine’’ particles in Catfish
finding food by flocking (not by each individual). It leaded the PSO. In other words, the catfish particles can guide particles
assumption that information is owned jointly in flocking. trapped in a local optimum onto a new regions of the search
According to observation of behavior of human groups, space, and thus to potentially better particle solutions.
behavior pattern on each individual is based on several In Catfish PSO, a population is randomly initialized in a
behavior patterns authorized by the groups such as customs first step, and the particles are distributed over the
and the experiences by each individual (agent). The D-dimensional search space. The position and velocity of each
assumptions are basic concepts of PSO. particle are updated by (1)–(3). If the distance between gbest
The PSO starts with a population of random solutions and the surrounding particles is small, each particle is
"particles" in a D-dimension space. The ith particle is considered a part of the cluster around gbest and will only move
represented by Xi=(xi1, xi2, ..., xiD). Each particle keeps track a very small distance in the next generation. To avoid this
of its coordinates in hyperspace, which are associated with the premature convergence, catfish particles are introduced and
fittest solution it has achieved so far. The value of the fittest replace the 10% of original particles with the worst fitness
for particle i (Pbest) is also stored as Pi=(Pi1, Pi2, ..., PiD). The values of the swarm.
global version of the PSO keeps track of the overall best value In PSO, the parameters w, r1 and r2 are the key factors
(gbest), and its location, obtained thus far by any particle in the affecting the convergence behavior. The inertia weight
population. PSO consists of changing the velocity of each controls the balance between the global exploration and the
particle toward its Pbest and gbest, at each step according to (1). local search ability. A large inertia weight favors the global
The velocity of particle i is represented as Vi=(vi1, vi2, ..., viD). search, while a small inertia weight favors the local search. For
Acceleration is weighted by a random term, with separate this reason, an inertia weight that linearly decreases from 0.9
random numbers being generated for acceleration toward Pbest to 0.4 throughout the search process is usually used [23]. Since
and gbest. The position of the ith particle is then updated logistic maps are frequently used chaotic behavior maps and
according to (2) [20] and [21]. chaotic sequences can be quickly generated and easily stored,
V id =W ×V id + C 1 × r1 × ( Pid − X id ) + C 2 × r2 × ( Pgd − X id ) (1) there is no need for storage of long sequences [24]. In C-PSO,
sequences generated by the logistic map substitute the random
x id = x id + cv id (2) parameters r1 and r2 are modified by the logistic map based on
Where Pid and Pgd are Pbest and gbest. Several modifications the following equation.
have been proposed in literature to improve the PSO algorithm Cr(t +1) = k × Cr(t ) × (1 − Cr(t ) ) (4)
speed and convergence toward the global minimum. One In (4), Cr(0) is generated randomly for each independent run,
modification is to introduce a local-oriented paradigm (Ibest) with Cr(0) not being equal to {0, 0.25, 0.5, 0.75,1} and k equal
with different neighborhoods. to 4. The driving parameter k of the logistic map, controls the
It is concluded that gbest version performs best in terms of behavior of Cr(t) (as t goes to infinity) [25]. The velocity
median number of iterations to converge. However, Pbest update equation for C-PSO can be formulated as:
version with neighborhoods of two is most resistant to local V id =W ×V id + c1 × Cr × ( Pid − X id ) + (5)
minima. The positive constants C1 and C2 are the cognitive and c 2 × (1 − Cr ) × ( Pgd − X id )
social components that are the acceleration constants
responsible for varying the particle velocity towards Pbest and In (5), Cr is a function based on the results of the logistic map
gbest, respectively. Variables r1 and r2 are two random functions with values between 0.0 and 1.0.
based on uniform probability distribution functions in the In C-Catfish PSO, a logistic map is embedded into Catfish
range [0,1]. The inertia weight w is responsible for PSO, which updates the parameters r1 and r2 based on (4)
dynamically adjusting the velocity of the particles, so it is [26]. The logistic map improves the search capability of
responsible for balancing between local and global searches Catfish PSO significantly. The particle velocities are updated
and hence requiring less iteration for algorithm to converge. according to (5).
The following inertia weight is used in (1): This new approach features many advantages; it is simple,
W =W
W
- m ax
- W m in
itera tio n
(3) fast and easy to be coded. Also, its memory storage
m ax
iterm ax requirement is minimal. Another advantage of C-Catfish PSO

121
 International Journal of Electronics and Electrical Engineering 6 2012

is that the initial population of the PSO is maintained, and so Fig. 3. The operating function of a PSS is to produce a proper
there is no need for applying operators to the population, a torque on the rotor of the machine involved in such a way that
process that is time and memory-storage-consuming. It is the phase lag between the exciter input and the machine
shown in [26] that better solutions can be found by guiding the electrical torque is compensated. The supplementary
whole swarm to more promising regions in the search space. stabilizing signal considered is one proportional to speed. In
C-Catfish PSO achieved far better performance than PSO, C- Fig.3, Vt is the terminal bus voltage and Vref is the reference
PSO, Catfish PSO and several other advanced PSO algorithms. voltage for the AVR. Limits of +5.0 p.u and -5.0 p.u, for the
The proposed algorithm will proceed as follows: field voltage were used in the simulations.
01: Begin Whereis ∆ωi the deviation in speed from the synchronous
02: Randomly initialize particles swarm speed. This type of stabilizer consists of a washout filter, a
03: Randomly generate Cr(0) dynamic compensator. The output signal is fed as a
04: while (number of iterations, or the stopping criterion is not
supplementary input signal, Ui , to the regulator of the
met)
excitation system. The washout filter, which essentially is a
05: Evaluate fitness of particle swarm
high pass filter, is used to reset the steady state offset in the
06: for n = 1 to number of particles
07: Find pbest output of the PSS. The value of the time constant, Tw is usually
08: Find gbest not critical and it can range from 0.5 to 20 s. In this study, it is
09: for d = 1 to number of dimension of particle fixed to 10 s. The dynamic compensator is made up to two
10: update the Chaotic Cr value by (4) lead–lag stages and an additional gain
11: update the position of particles by (5) and (2) Fig. 1 The computational flow chart of PSO algorithm is
12: next d shown in
13: next n
14: if fitness of gbest is the same Seven times then
15: Sort the particle swarm via fitness from best to worst
16: for n = number of Nine-tenths of particles to number of
particles
17: for d = 1 to number of dimension of particle
18: Randomly select extreme points at Max or Min of the
search space
19: Reset the velocity to 0
20: next d
21: next n
22: end if
23: update the inertia weight value by (3)
24: next generation until stopping criterion
25: end

III. STUDY SYSTEM MODELING

A. Power system model

A four-machine, two-area study system, shown in Fig. 2, is
considered for the damping control design. Each area consists
of two generator units. The rating of each generator is 900
MVA and 20 kV. Each of the units is connected through
transformers to the 230 kV transmission line. There is a power
transfer of 400MW from area 1 to area 2. Each synchronous
generator of the multi-machine power system is simulated
using a third-order model. The detailed bus data, line data, and
the dynamic characteristics for the machines, exciters and
loads are given in [27]. The loads are modeled as constant
impedances. On the basis of participation factors [28], two Fig. 1 Flowcharts of C-Catfish PSO algorithm
PSSs are installed in generators 1 and 3. The dynamics of the
machines are given in the Appendix A.

B. PSS structure
A first order model of a static type AVR was used and the
structure of the AVR equipped with the PSS is presented in

122
 International Journal of Electronics and Electrical Engineering 6 2012

3
F = ∑ Ji (8)
i =1
Where tsim is the time range of the simulation. It is aimed to
minimize this objective function in order to improve the
system response in terms of the settling time and overshoots.
To reduce the computational burden in this study, the values of
T21PSS, T41PSS, T23PSS and T43PSS are kept constant at a
reasonable value of 0.05 s. Thus, the design problem can be
formulated as constrained optimization problem shown in table
I.
Fig. 2 Multi-machine power system
TABLE I
CONSTRAINS OF PSSS PARAMETERS
The transfer function of the ith PSS is:
K1PSS T11PSS T31PSS K3PSS T13PSS T33PSS
sTw  (1 + sT1i ) (1 + sT 3i )  (6)
Ui = Ki ∆ωi (s ) min 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001
1 + sTw  (1 + sT 2i ) (1 + sT 4i )  max 50 1 1 50 1 1
The adjustable PSS parameters are the gain of the PSS, Ki ,
and the time constants, T1i-T4i. The lead–lag block present in The proposed approach employs C-Catfish PSO technique
the system provides phase lead compensation for the phase lag to solve this optimization problem and search for optimal or
that is introduced in the circuit between the exciter input and near optimal set of PSS parameters.
the electrical torque. The required phase lead can be derived It is emphasized that with this procedure, robust stabilizer,
from the lead–lag block even if the denominator portion enable to operate satisfactorily over a wide range of the
consisting of T2i and T4i gives a fixed lag angle. Thus, to operating conditions, are obtained. The flowchart of the
reduce the computational burden in this study, the values of T2i optimization based coordinated designing is depicted in Fig. 4.
and T4i are kept constant at a reasonable value of 0.05 s and The optimization of the PSSs controller parameters is carried
tuning of T1i and T3i are undertaken to achieve the net phase out by evaluating the objective cost function as given in (8)
lead required by the system. which considers a multiple of the operating conditions (OP).
The operating conditions are given in Table II. In order to
acquire better performance, number of particle, particle size,
number of iterations, c1, c2, and c are chosen as 50, 6, 100, 2,
2 and 1, respectively. It should be noted that the C-Catfish
PSO algorithm is run several times and then optimal set of
coordinated controller parameters is selected. The final values
of the optimized parameters are given in Table III. Also, Fig. 5
shows the minimum fitness value evaluating process.

TABLE II
Fig. 3 Block diagram of the AVR equipped with PSS
THREE OPERATING CONDITION
CONDITION OP1 OP2 OP3
C. Simultaneous coordinated design using C-Catfish PSO P1 0.7778 0.5556 0.9911
Q1 0.2056 0.2056 0.1722
The proposed controller must be able to work well under P2 0.2885 0.5556 0.6283
different operating conditions, while the improvement for the Q2 -0.1084 0.2611 0.5836
P3 0.8020 0.5597 0.0095
damping of the critical modes is necessary. Since the selection 0.0697 0.2217 0.0712
Q3
of the PSS parameters is a complex optimization problem. P4 0.8889 0.5556 1.1110
Thus, to acquire an optimal combination, this paper employs Q4 0.2244 0.2244 0.2222
the C-Catfish PSO algorithm to improve the optimization
TABLE III
synthesis and find the global optimum value. A performance OPTIMAL PSSS PARAMETERS USING PSO AND C-CATFISH PSO TECHNIQUE
index based on the system dynamics after an impulse PSO C-Catfish PSO
disturbance alternately occurs in the system is organized and K T1 T3 K T1 T3
used as the objective function for the design problem. In this G1 10.6354 1.0000 0.0801 12.0528 0.0001 1.0000
G3 12.4578 1.0000 0.0001 14.0424 0.0001 1.0000
study, an ITAE is taken as the objective function. Since the
operating conditions in the power systems are often varied, a
performance index for three different operating points is IV. NONLINEAR TIME-DOMAIN SIMULATION
defined as follows [30]: The effectiveness and robustness of the performance of the
t sim
proposed controller under transient conditions is verified by
J = ∫ t . (w 1 −w 2 + w 1 −w 3 + w 1 −w 4 + w 3 −w 4 ) dt (7)
applying
0

123
 International Journal of Electronics and Electrical Engineering 6 2012

0.45
Objective function
0.4

0.35

Objective function
0.3

0.25

0.2

0.15

0.1
0 5 10 15 20 25 30
Iterations
Fig. 5 Variations of objective function

of the chaos system. In PSO, each particle only relies on its

individual pbest value and the global best position gbest to update
its position at each generation. If gbest is trapped in a local
optimum, the particles cluster together and lose their ability to
explore the search space in later generations. In order to avoid
such a scenario, the worst 10% of the swarm are replaced by
Catfish particles when gbest has not changed for a certain
number of generations. After the catfish particles are
introduced, they initialize a renewed search from extreme
points of the search space, and thus find the better solutions by
guiding the entire swarm to promising new regions. They also
improve the search efficiency of the swarm. The catfish
particles not only allow the swarm to discover better solutions
within the area of the swarm itself, but also to obtain better
solutions located outside the swarm area.
Now it is clear from the figures (6-8) that the PSSs which
are tuned by C-Catfish PSO, are more effective than the ones
tuned by PSO in damping the low frequency oscillations and
improves the stability performance of the example power
Fig. 4 Flowchart of optimization based coordinated designing system.
To demonstrate the performance and robustness of the
a three-phase fault of 100 ms duration at the middle of one of proposed method, two performance indices: the ITAE and FD
the transmission lines between bus-7 and bus-10. PSO based on the system performance characteristics are defined as
algorithm has been also used in this paper to minimize the [31]:
ITAE objective function and these two algorithms are tsim t . ( w1 −w 2 + w1 −w 3 +
compared according to nonlinear time-domain ITAE = 1000∫ (9)
simulation for different operating conditions. The inter-area
0
w1 −w4 + w3 −w 4 ) dt
and local mode of oscillations with coordinated design of the FD = (OS × 5000 )2 + (us × 5000 )2 +Ts2
controllers using PSO and C-Catfish PSO algorithms are Where w is the speed rotor, overshoot (OS), undershoot
shown in Figs (6-8) for different operating conditions. In C- (US) and settling time of ∆w13 of the system is considered for
Catfish PSO, a chaotic map was embedded to determine the
PSO parameters r1 and r2. The PSO parameters r1 and r2 cannot the evaluation of the ITAE and FD indices. It is worth
ensure optimal ergodicity in the search space because they are mentioning that the lower the value of these indices is, the
absolutely random i.e. the r1 and r2 are generated by a linear better the system response in terms of time-domain
congruential generator (LCG) with a random seed. The characteristics. Numerical results of the performance and
generated sequence of LCG consists of pseudo-random robustness for all system loading cases are shown in Fig. 9.
numbers that have periodic characteristics. It can be seen that the application of the PSSs damping
Furthermore, the generated sequence of a logistic map also controller where the controllers are tuned by
consists of pseudo-random numbers, but there are no fixed C-Catfish PSO, achieves better response than the ones tuned
points, periodic orbits, or quasi-periodic orbits in the behavior by PSO.

124
 International Journal of Electronics and Electrical Engineering 6 2012

-3 -3
x 10 x 10
2 4
PSSs PSO PSSs PSO
PSSs C-Catfish PSO PSSs C-Catfish PSO
W1-W2 (Rad/s)

W1-W2 (Rad/s)
1

0
0

-2

-1
0 5 10 15
-4
Time (s) 0 5 10 15
Time (s)
-3
x 10
2 0.01
PSSs PSO
PSSs PSO
PSSs C-Catfish PSO PSSs C-Catfish PSO
1 0.005

W 1-W 3 (Rad/s )
W1-W3 (Rad/s)

0 0

-1 -0.005

-2 -0.01
0 5 10 15 0 5 10 15
Time (s) Time (s)
-3
x 10 0.01
3
PSSs PSO
PSSs PSO
2 0.005 PSSs C-Catfish PSO
PSSs C-Catfish PSO
W 1-W 4 (Rad/s )
W1-W4 (Rad/s)

1
0
0
-0.005
-1

-2 -0.01
0 5 10 15 0 5 10 15
Time (s) Time (s)
-3
x 10
-3 4
x 10
3 PSSs PSO
PSSs PSO PSSs C-Catfish PSO
2 2
W3-W4 (Rad/s )

PSSs C-Catfish PSO

W3-W4 (Rad/s)

1
0
0
-2
-1

-2 -4
0 5 10 15 0 5 10 15
Time (s) Time (s)
Fig. 6 Inter-area and local mode of oscillations for OP1
Fig. 7 Inter-area and local mode of oscillations for OP2

125
 International Journal of Electronics and Electrical Engineering 6 2012

-3
x 10
5 (a)
PSSs PSO
PSSs C-Catfish PSO
W1-W2 (Rad/s)

-5

-10
0 5 10 15
Time (s)
OP1 OP2 OP3
0.01
PSSs PSO
0.005 PSSs C-Catfish PSO (b)
W 1-W 3 (Rad/s )

-0.005

-0.01

-0.015
0 5 10 15
Time (s)
0.01
PSSs PSO OP1 OP2 OP3
0.005 PSSs C-Catfish PSO
W1-W4 (Rad/s)

Fig. 9 Values of performance index; (a) FD and (b) ITAE

0
V. CONCLUSION
-0.005
Techniques such as PSO and C-Catfisho PSO are inspired
-0.01 by nature, and have proved themselves to be effective
solutions to optimization problems. The objective of this
-0.015 research is to compare the performance of these two
0 5 10 15 optimization techniques for PSSs controller design. In this
Time (s) paper, the enhancement of power system stability is
0.015 investigated via coordinated tuning of PSSs. For this reason,
PSSs PSO the PSSs parameters are tuned according to optimization of
0.01 PSSs C-Catfish PSO ITAE Objective function by PSO and C-Catfish PSO
W3-W4 (Rad/s)

algorithms. The proposed C-Catfish PSO algorithm for tuning

0.005 PSSs is easy to implement without additional computational
complexity. Thereby experiments this algorithm gives quite
0
promising results. The ability to jump out the local optima, the
-0.005 convergence precision and speed are remarkably enhanced and
thus the high precision and efficiency are achieved. The
-0.01 effectiveness of the proposed method is tested on a 2-area 4
0 5 10 15
machines power system for three different operating conditions
Time (s)
and applying a three-phase fault of 100 ms duration at the
middle of one of the transmission lines between bus 7 and bus
Fig. 8 Inter-area and local mode of oscillations for OP3
10. Compared with PSO technique in terms of damping low
frequency oscillations, ITAE and FD indices, the C-Catfish
PSO technique demonstrates its superiority in computational
complexity, success rate and solution quality. C-Catfish PSO

126
 International Journal of Electronics and Electrical Engineering 6 2012

has a flexible and well-balanced mechanism to enhance the [14] Abido MA. Robust design of multimachine power system stabilizers
using simulated annealing. IEEE Trans Energy Convers
global and local exploration abilities. The nonlinear time 2003;15(3):297–304.
simulation results confirm that the proposed C-Catfish PSO [15] Abido MA, Abdel-Magid YL. Optimal design of power system
based tuned PSSs can work effectively over three different stabilizers using evolutionary programming. IEEE Trans Energy
operating conditions and is superior to the PSO. The system Convers 2002;17(4):429–36.
[16] Mishra S, Tripathy M, Nanda J. Multi-machine power system stabilizer
performance characteristics in terms of ‘ITAE’ and ‘FD’ design by rule based bacteria foraging. Electr Power Syst Res
indices reveal that this control strategy is a promising control 2007;77:1595–607.
scheme for PSS design in the real world power systems. [17] Ricardo V. de Oliveira, Rodrigo A. Ramos, Newton G. Bretas. An
algorithm for computerized automatic tuning of power system
stabilizers. Control Engineering Practice 18 (2010) 45–54
APPENDIX A [18] H. Shayeghi, H.A. Shayanfar, A. Safari, R. Aghmasheh. A robust PSSs
design using PSO in a multi-machine environment. Energy Conversion
The dynamics of each synchronous machine is given by [28] and Management 51 (2010) 696–702
and [29]: [19] Fukuyama, Y. (1999). A particle swarm optimization for reactive power
. and voltage control considering voltage stability. Proceedings of IEEE
δ i = ωb (ωi − 1) (A1) international conference on intelligent system applications to power
. 1 systems (ISAP), Rio de Janeiro.
ωi = ( Pmi − Pei − D i (ωi − 1)) (A2) [20] Shayeghi H, Jalili A, Shayanfar HA. Multi-stage fuzzy load frequency
Mi control using PSO. Energy Convers Manage 2008;49:2570–80.
. 1 [21] Kennedy J, Eberhart R, Shi Y. Swarm intelligence. San Francisco:
E qi′ = ( E fdi − ( x di − x di′ )i di − E qi′ ) (A3)
Morgan Kaufmann Publishers; 2001.
T doi′ [22] dos Santos Coelho, V.C. Mariani, A novel chaotic particle swarm
. 1 optimization approach using Hénon map and implicit filtering local
E fdi = ( K A i (v refi − v i + u i ) − E fdi ) (A4)
search for economic load dispatch, Chaos, Solitons and Fractals 39
T Ai (2009) 510–518.
T ei = E qi′ i qi − ( x qi − x di′ )i di i qi (A5) [23] Y. Shi and R.C. Eberhart, Empirical study of particle swarm
optimization, in: Proceedings of Congress on Evolutionary
Computation, Washington, DC, 2002, pp. 1945-1949.
[24] H. Gao, Y. Zhang, S. Liang, D. Li, A new chaotic algorithm for image
REFERENCES encryption, Chaos, Solitons and Fractals 29 (2006) 393–399.
[25] D. Kuo, Chaos and its computing paradigm, IEEE Potentials Magazine
[1] Basler MJ, Schaefer RC. Understanding power system stability. IEEE 24 (2005) 13–15.
Trans Indus Appl 2008;44:463–74. [26] Li-Yeh Chuang, Sheng-Wei Tsai, Cheng-Hong Yang, Chaotic catfish
[2] Andreoiu A. On power systems stabilizers: genetic algorithm based particle swarm optimization for solving global numerical optimization
tuning and economic worth as ancillary services. Ph.D. Thesis. problems, Applied Mathematics and Computation 217 (2011) 6900–
Chalmers University; 2004. 6916
[3] Xia D, Heydt GT. Self-tuning controller for generator excitation [27] Kundur P. Power system stability and control. New York: McGraw Hill;
control. IEEE trans PAS 1983;102:1877-85. 1994.
[4] Cao Y, jiang L, Cheng S, Chen D, Malik OP, Hope GS. A nonlinear [28] Sadikovic R. Damping controller design for power system oscillations.
variable structure stabilizer for power system stability. IEEE Trans EC Zurich: Internal Report; 2004.
1994;9(3):489-95. [29] Mishra S. Neural network based adaptive UPFC for improving transient
[5] Abido MA, Abdel-Majid YL. A hybrid Neuro-fuzzy power system stability performance of power system. IEEE Trans Neural Networks
stabilizer for multimachine power systems. IEEE Trans PWRS 2006;17(2):461–70.
1998;13(4):1323-30. [30] Nguyen TT, Gianto R. Optimisation based control coordination of PSSs
[6] Fraile-Ardanuy J, Zufiria PJ. Design and comparison of adaptive power and FACTS devices for optimal oscillations damping in multi-machine
system stabilizers based on neural fuzzy networks and genetic power system. IEE Proc Gener Transm Distrib 2007;1(4):564–73.
algorithms. Neurocomputing 2007;70:2902–12 [31] Shayeghi H, Shayanfar HA, Jalili A. Multi stage fuzzy PID power
[7] Segal R, Sharma A, Kothari ML. A self-tuning power system stabilizer system automatic generation controller in deregulated environments.
based on artificial neural network. Elect Power Energy Syst Energy Convers Manage 2006;47:2829–45.
2004;26:423–30.
[8] Hardiansyah FS, Furuya S, Irisawa J. A robust H ∞ power system
stabilizer design using reduced-order models. Elect Power Energy Syst
2006;28:21–8.
[9] Abdel-Magid YL, Abido MA, AI-Baiyat S, Mantawy AH. Simultaneous
stabilization of multimachine power systems via genetic algorithms.
IEEE Trans Power Syst 1999;14(4):1428–39.
[10] Abido MA, Abdel-Magid YL. Hybridizing rule-based power system
stabilizers with genetic algorithms. IEEE Trans Power Syst
1999;14(2):600–7.
[11] Zhang P, Coonick AH. Coordinated synthesis of PSS parameters in
multimachine power systems using the method of inequalities applied to
genetic algorithms. IEEE Trans Power Syst 2000;15(2):811–6.
[12] Abdel-Magid YL, Abido MA. Optimal multiobjective design of robust
power system stabilizers using genetic algorithms. IEEE Trans Power
Syst 2003;18(3):1125–32.
[13] Abdel-Magid YL, Abido MA, Mantawy AH. Robust tuning of power
system stabilizers in multimachine power systems. IEEE Trans Power
Syst 2000;15(2):735–40.

127