I. INTRODUCTION A.

Load frequency Control (LFC)

In power system, both active and reactive The aim of LFC is to maintain real power
power demands are never steady they continuously balance in the system through control of system
change with the rising or falling trend. Steam input frequency. Whenever the real power demand
to turbo generators(or water input to hydro changes, a frequency change occurs. This
generators) must therefore, be continuously frequency error is amplified, mixed and changed to
regulated to match the active power demand, a command signal which is sent to turbine
failing which the machine speed will vary with governor. The governor operates to restore the
consequent change in frequency, which may be balance between the input and output by changing
highly undesirable. The quality of power supply the turbine output. This method is also referred as
must meet certain minimum standards with regard Megawatt frequency or Power-frequency (P-f)
to constancy of voltage and frequency. The control .PSS is a fixed parameter analogue type
function of excitation control is to regulate device with lead-lag compensation, washout, and
generator voltage and reactive power output. The amplifier gains, which are limited and may lose
desired real power outputs of the Individual effective damping robustness for overall operation.
generating units are determined by the system
generation control. The voltage and frequency
controller has gained importance with the growth
of interconnected system and has made the
operation of power system more reliable.
The LFC and AVR of an isolated power
system have been reported and a number of control Fig 2: Block diagram of LFC with PID controller
schemes like Proportional and Integral (PI),
Proportional, Integral and Derivative (PID) and B.Automatic voltage regulator (AVR)
optimal control. The conventional method exhibits The aim of this control is to maintain the
relatively poor dynamic performance as evidenced system voltage between limits by adjusting the
by large overshoot and transient frequency excitation of the machines. The automatic voltage
oscillations. The artificial intelligence search regulator senses the difference between a rectified
techniques are used to find parameters voltage derived from the stator voltage and a
optimization of LFC and AVR considering reference voltage. This error signal is amplified
proportional-integral-derivative controller (PID) and fed to the excitation circuit. The change of
excitation maintains the VAR balance in the
for an isolated power system. The selection of the network. This method is also referred as Megawatt
optimal parameters of the PID controller is Volt Amp Reactive (MVAR) control or Reactive-
converted to a simple optimization problem which Voltage(QV)control.
is solved by a PSO algorithm.

Fig 3:Block diagram of AVR with PID controller

III.PARTICLE SWARM OPTIMIZATION

(PSO)

PSO, a proposed algorithm by James

Kennedy and R.C.Eberhart is one of the modern
Fig 1: Diagram of a LFC and AVR of a synchronous generator
heuristic algorithms, it has been motivated by the
behaviour of organisms, such as fish schooling and
II. PLANT MODEL
bird flocking. Generally, PSO is characterized as a
In an interconnected power system, LFC
simple concept, easy to implement, and
and AVR Equipment are installed for each
computationally efficient. Unlike the other
generator. The schematic diagram of the voltage
heuristic techniques, PSO has a flexible and well-
and frequency control loop is represented in fig.1.
balanced mechanism to enhance the global and
The controllers are set for a particular operating
local exploration abilities. It shares many
condition and take care of small changes in load
similarities with evolutionary computation
demand to maintain the frequency and voltage
techniques such as Genetic Algorithms (GA). The
magnitude within the specified limits.
PSO algorithm starts with random initialization of
population and velocity. The search for the
optimum solution is continued unless one of the
stopping criteria is reached. The stopping criteria
are either the maximum iterations are reached or
there is no further improvement in the optimal
solution. The PSO is initialized with a group of
random particles (solutions) and then searches for
optima by updating generations. In every iteration,
each particle is updated by following two best
values. The first one is the best solution (fitness) it
has achieved so far (the fitness value is also
stored) this value is called pbest. Another “best”
value that is tracked by the particle swarm
optimizer is the best value, obtained so far by any
particle in population. This best value is a global
best called as gbest. The particle takes part of the
population as its topological neighbours the best
value is a local best and is called lbest.
Let x and v denote a particle
coordinates (positions)and its corresponding flight
speed (velocity)in a search space, respectively.
Therefore, the i-th particle is represented as
xi=(xi1,xi2,…..xid) in the d-dimensional space .The
best previous position of the i-th particle is
recorded and represented as pbesti=(pbesti1,pbesti2,
….,pbestid).the index of the best particle among the
all particles in the group is represented by the
gbest. The rate of velocity for the particle i can
be represented as vi=(vi1,vi2…..vid).The modified
velocity and positions of the each particle can be
Calculated using the current velocity and the
distance from the following equations,

Vid(t+1)=ω.vid(t)+c1*rand()*(pbestid-xid(t))
+c2*rand()*(gbestd-xid(t)) (5)

Xid(t+1)=xid(t)+vid(t+1), i=1,2,3...n.,d=1,2...m
(6)
where c1 and c2 are cognition and social
parameters respectively, rand1() and rand2() are Fig 4: PSO flow chart
constant numbers in the range of [0,1], w is the
inertia weight. Vi represents the velocity of the ith The PID controller parameters are
particle and Xi is its position, pbesti and gbest are adjusted to keep the LFC and AVR at an optimum
local best and global best positions respectively. level of performance under varying operating
The velocity of particle in equation depends on its conditions. It requires a performance index which
previous velocity, its own thinking and social is the function of system parameters. The
psychological adaptation of the population. commonly used performance indexes are, integral
square error (ISE), integral of the absolute
The inertia weight ω is set according to magnitude of error (IAE) ,integral time square
the following equation , error (ITSE), integral of time multiply by absolute
error (ITAE). ITAE performance index which
(ωmax −ω min ⁡) produces smaller overshoots and oscillations.
ω=ω max− ×iter (7) The ITAE is given as,
iter max ∞

The advantages of PSO over the other

ITAE=∫ t |e (t)|dt (7)
0
optimization techniques are Lower sensitivity to In this approach an ITAE criterion is used for
the nature of the objective function Derivative free evaluating the PID controller. A set of good
property unlike genetic algorithm Easy control parameters P, I and D can yield a good step
implementation and Fast convergence. response that will result in performance criteria
minimization in the time domain. Therefore, the
performance criterion is defined as follows,
Ji¿ min ¿
(8)
The PID parameters are subjected to the following
constraints.
0 < Kp< 100
0 < Ki < 100
0 < Kd< 20
.
IV. SIMULATION MODEL OF LFC AND Fig 6: Simulink model of LFC and AVR with PSO based PID
AVR WITH PSO BASED PID CONTROLLER controller

To verify the efficiency of proposed V. SIMULATION RESULTS

algorithm a practical higher order system in Fig.4 The LFC and AVR are simulated using
is considered. The PSO algorithm is used to search PSO based PID controller, for different values of
an optimal parameter set containing Kp, Ki and load and regulation.
Kd. The optimum values generated by the
algorithm are stored in work space and shared with
the LFC and AVR simulink model.

Fig 7:Performance of AVR with PSO based PID controller for

∆PL=0.1 p.u
Figure 5:Simulink model of AVR with PID
controller
It is observed that the settling time of
AVR with PSO based PID controller is 4 seconds
and there is no transient peak overshoot. As seen
from the result, the PSO-PID controller could
create very perfect step response of the terminal
voltage in AVR system.
The LFC model was simulated with
different loads and regulations. The change
frequency for a load of 0.1 p.u and regulation ‘R’
Fig 6:Simulink model of LFC with PID controller value as 10.
Table I: Simulation Parameters
Simulation Parameters
Tg = 0.095, Tt = 0.5, H = 10,
LFC D = 0.8, R= 10, ΔPL = 0.10, 0.20
Ka =1.1165, Ta =0.2, Te = 0.4,
AVR Kg = 0.75, Tg =1.4, Kr= 1,
Tr = 0.05
Fig 8:Performance of LFC with PSO based PID controller for
∆PL=0.1 p.u
Table II: PSO Parameters
It is observed that the settling time of
No.of particles 20 LFC with PSO based PID controller is 8 seconds
No. of swarms 3 and the peak overshoot is -0.07. These results
No. of iterations 200 clearly indicate that the PID controller parameters
c1,c2 2,2 quickly and efficiently. In order to demonstrate the
Wmax,Wmin 0.9,0.4 effectiveness of the proposed controller,
Maximum particle (Upperbound-lower simulations are carried out for various
velocity bound)/No. of iterations performance indexes.
Table III: Comparison results of AVR for various [5] Haluk GOZDE et al, PSO based Load
performance indexes Frequency Control in a single area power system,
University of Pitesti, Scientific Bulletin,
Perform Kp Ki Kd Mp Ts Vol.2,No.8, 2008, pp106-110.
ance
indexes [6] Kennedy.J and Eberhart, R.C.,
“ParticleSwarmoptimization”,Proceedings of IEEE
ISE 8.4 8.2 2.2 1.16 8.2
International conference on Neural
ITSE 1.6 1.03 0.19 Nil 10.5
Networks,Australia, IV:1942-1948,1995.
IAE 1.63 1.37 0.54 1.05 7.4
ITAE 0.84 0.61 0.26 Nil 3.8 [7] A.Soundarrajan and S.Sumathi, “GA based
Table IV: Comparison results of LFC for various PID for Load Frequencycontrol”, Journal for
performance indexes Advances in Computational science
andtechnology, 2009, Vol.1, No.2.
Perform Kp Ki Kd Mp Ts
[8] A.Soundarrajan, et al., “Intelligent controllers
ance
for AutomaticGeneration Control”., In the
indexes
proceedings of International conferenceon
ISE 7.79 2.25 1.53 -0.09 10.8
Robotics, Vision, Information and signal
ITSE 11.4 3.8 4.14 -0.07 8.20
Processing, January,2003. pp307-311.
IAE 16.9 8.64 6.16 -0.06 11.5
ITAE 16.9 7.31 7.22 -0.05 6.20 [9] A.Khadobakshian and N.Golbon., “Robust
Load frequency Controllerdesign for Hydro Power
systems”., Proc. of IEEE conference onControl
VII. CONCLUSIONS Application, 2005, pp1510-1515.
The conventional controllers used for this problem
[10] Chen Y., Ye. L., “Anthropromorphic
have large settling time, overshoot and oscillations. Hence,
when evolutionary algorithms are applied to control system Intelligent PID controller and its applications in
problems, their typical characteristics show a faster and the Hydro turbine governor”, Proc.of International
smoother response. An intelligent technique has been conference on machine learning and cybernetics.,
proposed for combined voltage and frequency control in an 2002, pp391-395.
isolated power system. The proposed PSO-PID controller [11] Karnavas,K.L. and Papadopoulos, D.P, “AGC
provides a satisfactory stability between frequency overshoot
for autonomous power system using combined
and transient oscillations with zero steady state error. The
intelligent techniques”. Electric PowerSystem
work can be extended in future by including non-linear Research , 2002,62: 225-239.
parameters to the system modelled and thereby comparison
can be made with respect to linear system. Also the other [12] Zeynelgil, H.L, Demiroren, A. and
evolutionary computing techniques like Hybrid Particle Sengor,N.S., “The application ofANN technique
Swarm optimization (HPSO), GA-PSO, Hybrid GA, Fuzzy for automatic generation control for multi-
PSO, Distributed Evolution etc., and can be implemented to areapower system.”, Electric power and Energy
improve the performance characteristics. systems, 24(2002), 345-354.
VIII. REFERENCES
[1] Hadi saadat, “Power SystemAnalysis”,
Mcgraw-Hill International edition, 2005.
[2] H.D. Mathur and S.Ghosh, “A comprehensive
analysis of intelligent control for load frequency
control”, IEEE Power India conference,2006.
[3] D.M.Vinod Kumar, “Intelligent controllers for
Automatic Generation Control”, Proc. of IEEE
region 10 International conference on global
connectivity in Energy, Computer,
Communication andControl,1998, pp557-574.
[4] Zwe-Lee Gaing, “A particle Swarm
Optimization approach foroptimum design of PID
controller in AVR system”, IEEETransactions on
Energy Conversion, Vol.19, No.2, 2004, pp384-
391.