t

1s IEEE International Conference on Power Electronics. Intelligent Control and Energy Systems (ICPEICES-2016)

Power System Oscillation Damping by Intelligent

Power System Stabilizer

Sragdhara Bhattacharya
Department ofElectrical Engineering, National Institute of Technology Patna

Abstract-Power Systems are inherently non-linear signal that is injected into the AVR. However the design
systems that are frequently subj ected to various disturbances of a PSS and the optimal tuning of the PSS parameters is a
causing oscillations at low frequencies that may lead to multimodal optimization problem. Therefore it is not
instability.
Generators are usually provided with power
possible to use conventional optimization techniques to
system stabilizers minimize the effect of these oscillations.
determine the local and global minima points. PSO is a
The obj ective of this paper is find the optimal parameters for
a conventional lead-lag compensator based Power System
very modern optimization technique that has been
Stabilizer ( PSS) for a system comprising of a generator
developed according to the concept of swarm behaviour of
connected to an infinite bus and containing a STIA type organisms such as fish-schooling and bird-tlocking and it
excitation system. The tuning of the parameters of the Power has been shown to enhance the procedure of exploration of
System Stabilizer is accomplished using the Particle Swarm local and global minima points.
Optimization (PSO) algorithm. In this paper, a Fuzzy Power This paper mainly focuses on the optimal tuning of
System Stabilizer (FPSS) where the optimal values of the the parameters of a lead-lag type power system stabilizer
parameters of the FPSS are decided using the PSO algorithm
and also the parameters of the STlA type excitation
is also designed. The Particle Swarm Optimization based
system in a single-machine infmite bus system (SMIB)
conventional PSS and the Particle Swarm Optimization
using the Particle Swarm Optimization technique. Also a
based Fuzzy PSS are also incorporated in a system
containing multiple machines to check the system responses
fuzzy logic based power system stabilizer is designed in
under different loading conditions and faults of different which, the input and output scaling factors are chosen
types. The simulation results c1early prove the efficiency of optimally using the PSO technique. The PSO based lead­
the PSO based conventional and fuzzy power system lag PSS (PSOLLPSS) and the PSO based fuzzy logic PSS
stabilizers in damping the low frequency speed and power (PSOFLPSS) are now incorporated in the Single-machine
oscillations occurring in the power system due to various infinite bus system and also in a multi-machine system
disturbanees. (Kundur's 4-machine 2-area system ) [1] and the resuIts
Keywords-FuZ7JI Logic based Power System Stabilizer; are simulated under various loading conditions and also
Lead-lag Power System Stabilizer; Multi-machine system;
under different types of fauIt conditions. Simulation
Oscillation damping; Particle Swarm Optimization Algorithm
resuIts show, that the performance of the PSOFLPSS is
I. INTRODUCTION improved as compared to that of the PSOLLPSS in tenns
of the speed deviation response and the tenninal voItage
Power systems are highly non-linear systems and response of the system under consideration.
hence are frequently subjected to a wide variety of
disturbances or transient conditions which cause under­ 11. THE PARTICLE SWARM OPTIMIZATION TECHNIQUE
damped low frequency speed as weil as power system
The Particle Swarm optimization technique (PSO) is
oscillations which if not minimized within a short period analgorithm developed initially from the idea of birds
of time would cause the system to reach the verge of tlocking in a swarm. The technique exhibits stable
instability. The function of the generator excitation convergence, can be easily implemented and is
system is to maintain the terminal voItage of the computationally efficient. In the PSO technique, each
generator constant and also to control the reactive power particles travel in the search space with a velocity that
through the Automatic VoItage Regulator (AVR) . The needs to be updated as the particle gains movement
AVR is responsible for maintaining the steady-state experiences and also experiences of other particles'
stability of the system. However, maintaining the movements. Basically PSO may be developed by the
transient stability of the power system has to be taken care simulation of the bird-tlocking behaviour in a two­
of. During transients such as faults, the power system is dimensional space. Each particle remembers the best
subjected to a large perturbation for a short duration of position in the searching space that it has obtained so far
time due to which there is fall in the tenninal voltage of and this value is called Pbest . It also records the best
the generator and power transfer ability. For minimizing position obtained by any particle in the group and it is
the low frequency power system oscillations, a power termed gbest . The basic concept of the PSO technique is
system is usually provided with a Power System thus to push each individual towards its Pbest position
Stabilizer (PSS) which generates an auxiliary stabilizing and the gbest position at each time step. Acceleration
978-1-4673-8587-9/16/$31.00 ©2016 IEEE [1]
 t
1s IEEE International Conference on Power Electronics. Intelligent Control and Energy Systems (ICPEICES-2016)

towards the Pbest and gbest positions may be weighted III. THE SINGLE-MACHINE
by separate random numbers. In the search space, each INFINITE Bus SYSTEM
particle position is defined by X and Y coordinates and its
velocity is represented by the components Vx (along X­ A. WithL ead-Lag Type Conventional PSS
axis) and vy (along Y-axis) . Thus mainly depending on The schematic diagram of the single-machine infmite
four criteria, each agent in the swarm tries to update its bus system for small-signal stability studied, with a
position and velocity: lead-lag type power system stabilizer (PSS) installed is
a. The present coordinates(x,y). shown in Fig. 1. The model that has been used here is the
b. The present movement velocities ( vx, vy ) . Heffron-Philips model.
c. The distance between the present coordinate and
pbest.
d. The distance between the present coordinate and
gbest.
The modification of each agent may be represented by
the following equation:
V f+1 = wv f + Cl rand1 (pbesti - sf) +
c2rand2 gbest - s ( n (1)
where,
vf = velocity of the agent i at iteration k
w = weighting function
Ci = weighting factor
rand = random number between 0 and 1 Fig. I: Schematic Diagram of SMlB System with Lead-Lag Type PSS
sf position of the agent i at iteration k
=

pbesti pbest of the agent i

= B. Determination 0/ the Lead-Lag PSS Parameters and
gbest best value of the group
=
the Exciter Parameters
The expression for wused is given in (2). For the study of the small-signal performance of the
w wmax - Wmax-Wmin
=
t er iter (2) single-machine infmite bus system, the transfer function
i max
block diagram model that is taken into consideration, is
where
shown in the figure below. In this model, the synchronous
wmax = initial weight
generator is equipped with the excitation system of the
w min = final weight STlA model. In the STlA model, the exciter is
itermax maximum number of iterations
=
represented as a first order transfer function with a gain KA
iter current iteration number
=
and a time constant TA- The lead-lag power system
Using the above two equations, an updated or stabilizer (PSS) is represented by a gain KSTAB' a washout­
modified velocity, that gets closer to pbest and gbest filter block with time constant Tw and a dynamic
may be calculated. The position of each agent is updated compensator block with time constants Tl and T2 as
by the following equation as in (3). shown in the diagram. A Power System Stabilizer
S f +1 sf + V f +1
= (3) basically provides an auxiliary stabilizing signal Vpss
where, comes as a positive feedback to be provided to the
S f +1 = modified position excitation system along with the reference voltage and the
terminal voltage of the generator. If the overall transfer­
sf = current position
function of the plant from Vret to !:J.w can be calculated,
Vf +1 = modified velocity
then the PSS may be considered to be a controller placed
vf = current velocity in the feedback path and providing a positive feedback
The values of the PSO variables are Iisted as in Table with !:J.w as its input.
land the Fig.l below shows the graphical plot for the PSO Thus, the parameters to be optimized by the particle
search algorithm by which the positions and velocities of swarm optimization algorithm are the exciter parameters
the particles are updated. that is KA and TA, the parameters of the lead-lag PSS that
TABLE I: TABLE SHOWING THE PSO PARAMETERS is KSTAB,Tw,T1 and T2 and the time constant TR of the
PSO Parameters Values voltage transducer block. The inputs to the excitation
Initial Population Size 50 system are the terminal voltage of the generator V t , the
c ,c? 2 reference voltage Vret and the stabilizing voltage VPSS
Wmax 0.9
0.4
generated by the power system stabilizer. The power
Wmin

[2]
 t
1s IEEE International Conference on Power Electronics. Intelligent Control and Energy Systems (ICPEICES-2016)

system stabilizer used in this section is a speed input PSS output of the FLC are divided into seven membership
that takes the generator speed deviation I::.w as its input. functions namely Negative Big (NB), Negative Medium
The performance of the proposed particle swarm (NM), Negative Small (NS), Zero (ZE), Positive Small
optimization technique for the tuning of the exciter (PS), Positive Medium (PM) and Positive Big (PB) [2].
parameters and those of the lead-lag PSS is thus The FLC has two input scaling factorsKw for I::.w and Kp
investigated. The single-machine infmite bus system for I::.Pe and a single output scaling factor Kp ss. The
considered, is fIrst linearized about a specifIed operating choice of the scaling factors calls for proper understanding
point and the eigenvalues of the system are fIrst of the input-output behaviour of the system.The Fuzzy
calculated. The particle swarm optimization technique Logic rule base used for the design of the PSOFLPSS is as
proposed in this paper, aims at minimizing a suitable shown below in Table III.
objective function which, in this case is chosen as
shown in (4).
OB] (1- e-ß).(Mp + esJ + e-ß(ts - tr) (4)
=

where,
OB] objective function to be evaluated
=

The time domain specifIcations of the system, used

for formulating the objective function are the maximum
overshoot (Mp), the rise time (tr), the settIing time (tJ
and the steady-state error (ess)' ß is the value of the
weighting factor which enables the system to satisfy the
requirements of the designer. The value of ß is usually
chosen to be in the range of 0.8 to l.5. In this paper, the
value of ß has been chosen as l.2. This work proposes the
above new performance index in the time domain for
Fig.2. Schematic diagram of the SMIB system with Fuzzy based PSS
evaluating the parameters of the exciter and the lead lag
TABLE 3: THE Fuzzy RULES USED"
power system stabilizer. The purpose behind choosing
such an objective function is the fact that if the parameters
Spee d Change in acti",e Ilo,wer deriation
of the PSS and exciter are properly chosen then that would
produce a good step response in the time domain. For de-iati •. P'B, Pi\i PS ZE )l'S N�I );B
executing the PSO optimization algorithrn, the limits of Ilw
the variables are fIrst set. The higher and lower limits of PB P'B, PB PM P'B PM PS ZE
the target variables and the corresponding optimal PZt.1 p�, PB PM P:-"1 PS ZE �S
values obtained by The PSO technique are given below in PS P'B,I PM PI\! PS ZE �s �f
Table 11.
TABLE 2: THE OPTIMUM VALUES OF THE LEAD-LAG PSS
ZE F�! I Pli.1 PS zr �s N1.1 ),1:"'1
Parameter :"amE' KA TR KSTA B Tw Tl Tz )l'S PM PS ZE :\5 �1I,1 !'öl :\B
Lower B oWld 200 0.01 3 1 0.01 0.001
�"M PS ZE ::-""$ :.")� �1\,r :;.m �B
Upper B o und 450 0.2 60 3 0.4 0.008

Op ti lll i zed ,"alue 333.86 0.05 48.76 1.5 0.078 0.0035

)lB zr I :-lS �s :-m :-lB �B �B

D. Determination 0/ the Parameters 0/ the Fuzzy L ogic

C. With FuzzyL ogic based Power System Stabilizer
based PSS using PSO
In this paper, the model used for study and simulation
The tuning of the parameters that is the input and
is the Single-Machine Infmite Bus model as shown in
output scaling factors of the FLC is very essential in order
Fig. 2. The control signal generated by the Fuzzy logic
to obtain the desired output. For a proper choice of the
controller is equivalent to the supplementary stabilizing
FLC parameters, one needs to have prior knowledge of the
signal provided by a conventional power system stabilizer
and is injected into the AVR system as a feedback. The operation of the system and also details of the input output
main aim of the implementation of a fuzzy logic controller signals. However, in most of the cases, such information
is to stabilize the power system and to improve the about the system is lacking. As a resuIt, the choice of the
damping of the power system oscillations occurring due to FLC parameters depends on a trial and error procedure
generator speed deviation and generator rotor deviation for which is not so reliable. Therefore, optimization
all possible operating conditions. The speed deviation I::.w techniques like PSO is made use of for the design of a
and the deviation in active power I::.Pare the input signals Power System Stabilizer. Determining the parameter set
for the Fuzzy Logic Controller and the output stabilizing appropriately enables the system give a good step
signal I::.vs is injected into the AVR . For the formulation response which further minimizes the performance index
of the fuzzy inference system, each of the inputs and the in the time domain[5]. The time domain specifIcations of

[3]
 t
1s IEEE International Conference on Power Electronics, Intelligent Control and Energy Systems (ICPEICES-2016)

the system, used for formulating the objective function are B. Multi-Machine System (Kundur's Two-Area Four
the maximum overshoot (Mp) , the rise time (tr) , the Machine System) ..
settling time (tJ and the steady-state error (ess)' Thus,
the objective function may be formulated as in (5). The The single line diagram of the multi-machine
objective function stated below is basically a combination system[l] used in this paper is as shown in Fig.4.The
of the above-mentioned time-domain specifications so that system basically comprises of fOUT synchronous machines
optimization of this objective function would ultimately arranged in two areas, two in each with the area 1
result in a better step response of the system. exporting a power of about 413MW to the area 2. Each
OB] (1 - e-ß).(Mp + esJ + e-ß(ts - tr) (5)
= generator is equipped with the STlA type excitation
where ß is the value of the weighting factor which system. The PSOLLPSS and the PSOFLPSS designed in
enables the system to satisty the requirements of the the above sections are incorporated in the machine 1 of
designer. The value of ß is usually chosen to be in the area 1 and thereafter the results are simulated for the
range of 0.8 to 1.5. In this paper, the value of ß has been following large scale disturbances.
chosen as 1.2. The upper and lower bounds for the • SLG fault at bus 8.
PSOFLPSS are as given in Table IV. • LLG fault at bus 8.
TABLE 4: THE OPTIMUM INPUT AND OUTPUT SCALING
FACTORS FOR THE Fuzzy LOGIC CONTROLLER DESIGNED
• Permanent tripping of one transmission line .
• 3-<pto ground fault at bus 8.
Parameter Kw Kp Kpss 1106<m 9
Upper Bound 40 4 I
G1
Lower Bound 60 8 4
Optimized Values 56.5 3.98 1.145

IV. RESULTS AND DISCUSSIONS

After the design of the PSO based lead-lag power

system stabilizer (PSOLLPSS) and the PSO based Fuzzy Fig. 4: The 2-Area 4-Generator Test System C. !:J.w and !:J.Vt
Logic PSS (PSOFLPSS) , these are incorporated in the Responses of GI for SLG Fault at Bus 8
SIMULINK model of the Single-machine Infinite Bus From Fig. 5 and Fig. 6 shown below, it is observed
system and also a multi-machine system, thus showing the
that the speed deviation I:J.w is better for PSOLLPSS than
improvement in the damping of the power system
for PSOFLPSS having lower under-shoot and also
oscillations.
comparatively lesser settling time. However, the terminal
A. Single-Machine Infinite Bus System .. voltage I:J.Vtresponse is much superior with the PSOFLPSS
having lower magnitude of peak overshoot and nearly
A system comprising a single generator connected to
equal settling time.Hence, the PSOFLPSS is seen to
an infmite bus and having a STlA exciter is considered.
The response of deviation in generator speed I:J.w of the exhibit a much superior performance.
......

SMIB system with the PSOLLPSS and the PSOFLPSS are

?SOfL?$
observed as shown in Fig. 3. It may be inferred from the
results that with PSOLLPSS and PSOFLPSS, the
generator speed deviation signal I:J.w are almost equally
damped. However, if the response for tenninal voltage of
the synchronous generator I:J.Vt is also plotted then it may
be observed that it is better with a PSOFLPSS having
improved overshoot and settling time than with a
PSOLLPSS. Thus the PSOFLPSS exhibits a better
p�rformance. Fig. 5: !:J.w Response of GI Due to SLG Fault at Bus 8
,"

• ?SO L l ?SS
, f?SOfL?SS

, ' .

.----' /
, ,
" .. -._-

f
�

, \
, ,
, ,

r I�I
��
"

Fig. 3: !:J.W of Single Machine with PSOLLPSS and PSOFLPSS Fig. 6: !:J.V, at GI due to SLG Fault at Bus 8

[4]
 t
1s IEEE International Conference on Power Electronics. Intelligent Control and Energy Systems (ICPEICES-2016)

C. I:J.w and I:J.Vt Responses ofG] forLLG Fault at Bus 8 From Fig. 9 and Fig. 10 shown below, it may be
observed that for a 3-<pfault occurring at bus 8, the speed
From Fig. 7 and Fig. 8 shown below, it may be deviation I:J.w is better with the PSOFLPSS than with the
observed that for a LLG fault occurring at bus 8, the PSOLLPSS. Though the PSOLLPSS gives comparatively
speed deviation I:J.w is slightly better for PSOLLPSS than lesser settling time and a less oscillatory response, the
for PSOFLPSS having slightly lesser undershoot and undershoot is very high. However, the terminal voltage
lower settling time. However, the tenminal voltage I:J.Vt is I:J.Vt is better with the PSOFLPSS than with the
much better for PSOFLPSS than for PSOLLPSS having PSOLLPSS having lesser settling time and lower peak
much lesser settling time and lower peak overshoot. overshoot.
......

E. I:J.w and I:J.Vt Responses of G] with Tripping of One

L ine
From Fig. 11 and Fig. 12 shown below, it may be
observed that for a LLG fault occurring at bus 8, the speed
deviation I:J.w is much better for PSOFLPSS than for
PSOLLPSS having lowerovershootsand lower settling
time. Also, in this case the terminal voltage I:J.Vt is much
better for PSOFLPSS than with PSOLLPSS showing
Fig. 7: /J.w Response of GI Due to LLG Fault at Bus 8
I?wer settling time and lower undershoots and overshoots.

?SOLL?SS
?WfL?SS

?n Fa];lt ?MiO<!

...J-- -- +- -- -+- -- --+

-- ---;f- -- ;\- +
-- -- ---ic ----;f-
-- ;\-
-- -- �
.�
.• -- � ---- � --+- -- � -- ��--� -- +.- -- � -- �-- �
Fig. 11: /J.w Response of GI Due to Tripping of One Transmission Line
Fig. 8: /J.V, Response of GI Due to LLG Fault at Bus 8

D. I:J.w and I:J.Vt Responses ofG] for 3-<p Fault at Bus 8

?SOLL?SS

f---+-
. •. -- -+- -- --+
---- J--
-- +.- --i;-
-- -- -+. ---- I;- +.-
-- -- -,l

Fig. 12: /J.V, Response of GI Due to Tripping of One Transmission Line

T ABLE 5: COMPARISONS BETWEEN TIME DOMAIN SPECIFICA T10NS

FOR DIFFERENT TYPES OF FAULTS USING TwO DIFFERENT PSO-PSS
••• f- -- +- -- -+- -- --+ -- ---;f- -- ;\-
-- + ---ic
-- -- ----;f-
-- ;\- --�
IeH Condition Response PSOPSS Onrshoot lindershoot PeakIime SettJin"
Fig. 9: /J.w Response of GI Due to 3-<p Fault at Bus 8 in �t, in �'o
"� Time
Single line to Speed PSOFLPSS 0.016 .o.Oi81 O.�; ;
ground dedation PSOLLPSS 0.0238 ·0.0699 0.42 ]
fault(SLG Temrinal PSOFLPSS O. I � ·0.1 0.09 4
faul.) \'oltage PSOLLPSS 0.13 .0.11 0.1 �
Double line 10 Speed PSOFLPSS Nu -0.725 0.91 ;.8
ground de\'iation PSOLLPSS 0.06 ·0.64; 0.87 U
faul.(LLG Tenninal PSOFLPSS O. B
. .Oj j 1.33 6
faul.) yoltage PSOLLPSS OA I .0.61 1.33 6
Three-phasf [0 Speed PSOFLPSS 0;; -3.4 2.3 9
ground fauh deüation PSOLLPSS Oj -3.2 1.1 gj
Tenninal PSOFLPSS 0.4 ·0.67 0.11 11
yoltage PSOLLPSS 0.6 ·0.66 0.1 10

V. C ONCLUSION
This paper proposes the design of a conventional
lead-lag PSS (LLPSS) and also a fuzzy logic PSS
Fig. 10: /J.V, Response of GI Due to 3-<p Fault at Bus 8
(FLPSS) for a SMIB system with a STlA exciter. Also the

[5]
 t
1s IEEE International Conference on Power Electronics. Intelligent Control and Energy Systems (ICPEICES-2016)

paper proposes the procedure for the determination of the [5] Ghazanfar Shahgholian, Amir Movahedi, and Jawad Faiz,
"Coordinated Design of TCSC and PSS Controllers using
optimal values of the exciter parameters , the parameters
VURPSO and Genetic A1gorithms for Multi-machine Power
of the LLPSS and also the scaling factors of the FLPSS by System Stability", International Journal of Control, Automation,
the Particle Swarm Optimization (PSO) technique. The and Systems (20I5) 13(2): 398-409,Springer.
implementation of the PSOLLPSS and PSOFLPSS in the [6] Mohamed Jalaluddin, Dr.H.V.Saikumar, Dr.Bansilal,
SMIB and multi-machine systems has been presented "Performance Evaluation of Multi Machine Power System with
Fuzzy Based Power System StabiIizer", 20I5 IEEE International
through simulations that c1early justify the fact that the
Conference on Technological Advancements in Power & Energy.
PSOFLPSS gives comparatively better performance than
[7] Abe. S., & Doi, A., "A new power system stabilizer synthesis in
PSOLLPSS as a lead-lag based conventional PSS may be multi-machine power systems",IEEE Transactions on Power
designed only for a particular operating point while the Apparatus and Systems,I983, PAS-I02, 39I0-39I8.
fuzzy logic based PSS is found to provide satisfactory [8] Do Bmfim, A.L.B., Taranto, G.N., et al.,"Simultaneous tuning of
operation at all operating points within the range of power system damping controllers using genetic
algorithms",2000,IEEE Transactions on Power Systems, 15(1),
operation of the system.
163-169.

REFERENCES [9] EL-Hawary, M.E., "Electric power applications of fuzzy

systems",I998,IEEE press.
[I] P. Kundur, "Power System Stability and Control", McGraw-Hill, [10] EL-Zonkoly, A.M.,"Optimal tuning of power systems stabilizers
Inc. and A VR gains using particle swarm optimization", 2006,
[2] S.P. Ghoshal,A. Chanerjee,V. Mukherjee, "Bio-inspired fuzzy International Journal of Expert Systems with Applications (Vol.
logic based tuning of power system stabiIizer", Expert Systems 31(939)). pp. 551-557.
with Applications 36 (2009) 9281-9292,Elsevier. [II] Shi.Y,Eberhart.R."Amodified particle swarm
[3] P.K. Dash, A.C. Liew, B.R. Mishra, "An adaptive PID stabilizer optimizer",I998,Proc. IEEE Int. Conf Evol. Comput. Anchorage,
for power systems using fuzzy logic", Electric Power Systems UK (pp. 69-73).
Research 44 (1998) 2I3-222, Received 18 April 1997; received in [12] Anderson,P.M,Fouad.A.A,"Power system dynamics and stability",
revised form 15 August 1997; accepted 8 September 1977, Ames, IA: lowa State Univ. Press.
1997,Elsevier.
[4] Zwe-Lee Gaing, "A Particle Swarm Optimization Approach for
Optimum Design of PID Controller in AVR System", IEEE
Transactions on Energy Conversion, Vol. 19, No. 2, June 2004.

[6]