International Journal of Science and Research (IJSR), India Online ISSN: 2319-7064

Analysis of Automatic Generation Control Two

Area Network using ANN and Genetic Algorithm
Deepika Lathwal1, Priyanka Malik2, Sunil Kumar3
1
Murthal University, Sonipat, Haryana, India
[email protected]
2
Department of Electronic Engineering, YMCA UST, Faridabad, India
[email protected]

Abstract: This Paper presents decentralized control scheme for Load Frequency Control in a two-area Power System by appreciating
the performance of the methods in a single area power system. A number of modern control techniques are adopted to implement a
reliable stabilizing controller. A serious attempt has been undertaken aiming at investigating the load frequency control problem in a
power system consisting of two power generation unit and multiple variable load units. The robustness and reliability of the various
control schemes is examined through simulations. This paper deals with the automatic generation control (AGC) of interconnected
thermal systems with combination of the automatic voltage control (AVR) and Demand Side Management (DSM). In this particular
work thermal unit is considered with four area concept. The primary purpose of the AGC is to balance the total system generation
against system load and losses so that the desired frequency and power interchange with neighboring systems are maintained. Any
mismatch between generation and demand causes the system frequency to deviate from scheduled value. Thus high frequency deviation
may lead to system collapse. Further the role of automatic voltage regulator is to hold terminal voltage magnitude of synchronous
generator at a specified level. The interaction between frequency deviation and voltage deviation is analyzed in this paper.

Keywords: Automatic generation control (AGC), ANN, Genetic Algorithm (GA)

1. Introduction controller has been reconstructed with applying the μ-

based robust controller to power systems in different
Automatic Generation Control (AGC) is one of the most operating points under different load disturbances by
important issues in electric power system design and using the learning capability of the neural networks.
operation. The objective of the AGC in an interconnected
power system is to maintain the frequency of each area Genetic algorithm (GA) is an optimization method based
and to keep tie-line power close to the scheduled values on the mechanics of natural selection. In nature, weak and
by adjusting the MW outputs the AGC generators so as to unfit species within their environment are faced with
accommodate fluctuating load demands. The automatic extinction by natural selection. The strong ones have
generation controller design with better performance has greater opportunity to pass their genes to future
received considerable attention during the past years and generations. In the long run, species carrying the correct
many control strategies have been developed [1 4] for combination in their genes become dominant in their
AGC problem. The availability of an accurate model of population. Sometimes, during the slow process of
the system under study plays a crucial role in the evolution, random changes may occur in genes. If these
development of the most control strategies like optimal changes provide additional advantages in the challenge for
control. However, an industrial process, such as a power survival, new species evolve from the old ones.
system, contains different kinds of uncertainties due to Unsuccessful changes are eliminated by natural selection.
changes in system parameters and characteristics, loads In real-coded genetic algorithm (RCGA), a solution is
variation and errors in the modeling. On the other hand, directly represented as a vector of real parameter decision
the operating points of a power system may change very variables, representation of the solutions very close to the
much randomly during a daily cycle. Because of this, a natural formulation of the problem [8], [4]. The use of
fixed controller based on classical theory [2 3] is certainly floating-point numbers in the GA representation has a
not suitable for AGC problem. Thus, some authors have number of advantages over binary encoding. The
suggested a variable structure [4 5] and neural networks efficiency of the GA gets increased as there is no need to
methods [6 9] for dealing with parameter variations. All encode/decode the solution variables into the binary type.
the proposed methods are based on the state-space
approach and re-quire information about the system states The objective here is to minimize the deviation in the
which are not usually known or available. frequency of two areas and the deviation in the tie line
power flows and these variations are weighted together by
In this paper, because of the inherent nonlinearity of a linear combination to a single variable called the ACE.
power systems we address a new nonlinear Artificial The fitness function is taken as the Integral of time
Neural Network (ANN) controller based on μ-synthesis multiplied absolute value of ACE at every discrete time
technique. The motivation of using the μ-based robust instant in the simulation [1], [2]. An optional penalty term
controller for training the proposed controller is to take the is added to take care of the transient response
large parametric uncertainties and modeling error into ac- specifications viz. transient response specifications on
count. To improve the stability of the overall system and system frequency settling time, over shoots, etc. Integral
also its good dynamic performance achievement, the ANN
Volume 2 Issue 3, March 2013
www.ijsr.net 274
 International Journal of Science and Research (IJSR), India Online ISSN: 2319-7064

of time multiplied absolute value of the Error (ITAE), is inductive components. The speed-load characteristic of
given by the composite load is given by
ΔPe=ΔPL+DΔω (2)
2. Mathematical Modeling of Generator
where ΔPL is the non-frequency- sensitive load change,
�� �� ∆� DΔω is the frequency sensitive load change.
= ∆P� - ∆P�
� ���
D is expressed as percent change in load by percent
Taking Laplace Transform, we obtain change in frequency
�
∆Ω�s�= [∆P� (s) -∆P� (s) (1)
���

Figure 1: Modeling of Generator

Figure 2: Represent the Block diagram of Load
3. Mathematical Modeling of Load
4. Mathematical Modeling for Governor
The load on the power system consists of a verity of
�
electrical drives. The equipments used for lighting ΔPg=ΔPref - ∆f (3)
�
purposes are basically resistive in nature and the rotating Or in s- domain
devices are basically a composite of the resistive and �
ΔPg(s)= ΔPref - ∆f (4)
�
Figure 1 Represent the Block diagram of Generator
5. Two Area Network with neural network
In this section we study the Response of two areas AGC
using Neural Network in terms of frequency load variation
for different gain value. In fig 4 as neural networks, in we
used NARMA-L2 controller to controlling the load
frequency. Also we used feed forward neural network to
controlling load frequency, the ANN controller
architecture employed here is Non linear Auto Regressive
Figure 3: Represent the Block diagram of Governor Model reference Adaptive Controller. It consists of
reference, plant output and control signal. The plant
The command ΔPg is transformed through hydraulic output is forced to track the reference model output. Here,
amplifier to the steam valve position command ΔPV. We the effect of controller changes on plant output is
assume a linear relationship and consider simple time predicted. It permits the updating of controller parameters.
constant T we have the following s-domain relation: In the study, the frequency deviations, tie-line power

� deviation and load perturbation of the area are chosen as

ΔPV (s) = ∆P� �s� (5)
���� the neural network controller inputs. Control signals
applied to the governors in the area act as the outputs of
the neural network. The data required for the ANN
controller training is obtained by designing the Reference
Model Neural Network and applying to the power system
with step response load disturbance

Volume 2 Issue 3, March 2013

www.ijsr.net 275
 International Journal of Science and Research (IJSR), India Online ISSN: 2319-7064

Figure 4: Simulink Block Diagram Two Area network with Neural Network

6. Optimal Control Method Using Genetic Where K is the feedback gain matrix, In this paper
Algorithm Evolutionary Genetic algorithms is used to optimize the
feedback gains of the controller. Genetic algorithm (GA)
An optimal AGC strategy based on the linear state is an optimization method based on the mechanics of
regulatory theory requires the feedback of all state natural selection. In nature, weak and unfit species within
variables of the system for its implementation, and an their environment are faced with extinction by natural
optimal control feedback law is obtained by solving the selection. The efficiency of the GA gets increased as there
non-linear Riccati equation using suitable computational is no need to encode/decode the solution variables into the
technique. To illustrate the effectiveness of the proposed binary type.
control design and the algorithm to tune the feedback
gains to the controller, a two area Restructured power 7. Chromosome structure
system having two GENCOs and two DISCOs in each
area is considered. The time-invariant state space for In GA terminology, a solution vector known as an
figure 5 representation as: individual or a chromosome. Chromosomes are made of
discrete units called genes. Each gene controls one or
X� = A.X +B.U (6) more features of the chromosome [9]. The chromosome
Y = C.X (7) string comprises of all feedback gains encoded as a string
of real numbers.
Where X is the state vector and U is the vector of
contracted and un-contracted power demands of the 8. Fitness-Objective function evaluation
DISCOs
The objective here is to minimize the deviation in the
X � � Δf1 Δf2 ΔPg1 ΔPg2 ΔPg3 ΔPg4 � ACE1 � ACE2 ΔPtie12, act�^T frequency of two areas and the deviation in the tie line
(8) power flows and these variations are weighted together by
And U � � ΔPL1 ΔPL2 ΔPL3 ΔPL4 ΔPd1 ΔPd2�� (9) a linear combination to a single variable called the ACE.
The fitness function is taken as the Integral of time
for the system defined by the Eq.(6) and (7), the feedback multiplied absolute value of ACE at every discrete time
control law is given as, instant in the simulation [1], [2].

U= -K.Y (10)

Volume 2 Issue 3, March 2013

www.ijsr.net 276
 International Journal of Science and Research (IJSR), India Online ISSN: 2319-7064

Figure 5: Simulink Block diagram of Two Area network with GA

9. Simulation and Result using ANN and Genetic Algorithm controllers and
integral controllers to demonstrate the performance of
In this study simulations were performed using load frequency control using MATLAB/SIMULINK
MATLAB® Simulink and Fuzzy Logic Toolbox on a two- package. Fig. 5, 6, 7 & 8 respectively represent the plots
equal area interconnected power system having steam of change in system frequency and tie-line power
turbine type thermal units in both areas Two area network respectively for 1% step load variation.
are. Interconnected power system has been developed

Figure 8: Response of load frequency of two area using

GA for different gain value
Figure 6: Response of load frequency of two area using
ANN for different gain value. The above figure’s is
showing that response of AGC two are network for
controlling the load frequency at 1-2%.for different gain
value using ANN

Figure 9: Complete Response of 2% load frequency of

two areas using GA for a gain value

10. Conclusions
This paper has offered a brief introduction to the basic
Figure 7: Complete Response of 2% load frequency of
application of Neural Network and Genetic algorithm for
two area using ANN for a gain value In this section we the controlling load frequency of Automatic Generation
find the combined response of two are network for control two are network for different gain value. For
controlling the load frequency at 1-2% for different gain genetic algorithm we used optimization theory to find the
value using ANN.
exact or optimum value of network for different input
data. The two area network consist of, a plant connected
though tie line to delivered power in terms of frequency
Volume 2 Issue 3, March 2013
www.ijsr.net 277
 International Journal of Science and Research (IJSR), India Online ISSN: 2319-7064

across load. The load value is depends on gain of governor

of plant which is vary with power across load.

Reference
[1] Karrari, m. shayeghi, h. menhaj, m. B. abedi, m.:
design of h1 controller load frequency control in
electric power systems, amirkabir journal of science
& technology 11 no. 41 (1999), 79 88.
[2] Nanda, j. kavi, bl.: automatic generation control of
interconnected power systems, ieee proc. On gen.
Tran. And dis.125 no. 5 (1988), 385 390
[3] Das,d. nanda, j. khothari,ml khothari, dp. : Automatic
generation control of hydrothermal system with new
area control error considering generation rate
constant, electric machines and power systems 18
(1990), 461 467.
[4] jiang, h. cai, h. dorsey, f. qu, z. : toward a glob- ally
robust decentralized control for large-scale power
systems, ieee trans. On control system technology 5
no.3, (1997), 309 319
[5] Al-hamouz, z. M. al-duwaish, h. N.: a new load
frequency variable structure controller using genetic
algorithms, electric power systems research 55 no. 1
(2000), 1 6.
[6] Beaufays, f. abdel-magid, y. Widrow, b.: application
of neural networks to load frequency control in power
system, neural networks 7 no. 1 (1994), 183 194.
[7] Zeynelgil, h. L. demiroren, a. Sengor, n. S.: the
application of ann technique to automatic generation
control for multi-area power system, electrical power
and energy systems 24 (2002), 545 354.
[8] A. Konak et al, (2006) - multi-objective optimization
using genetic algorithms: a tutorial- / reliability
engineering and system safety 91 992 1007
[9] Pingkang li1 xiuxix du, (2009) - multi-area agc
system performance improvement using ga based
fuzzy logic control- the international conference on
electrical engineering
[10] Y.l.karnavas,k.s.dedousis, (2010)- overall
performance evaluation of evolutionary designed
conventional agc controllers for interconnected
electric power system studies in a deregulated market
environment- international journal of engineering,
science and technology,vol. 2, pp150-166
[11] Elyas rakhshani , javad sadeh, (2010) practical
viewpoints on load frequency control problem in a
deregulated power system-energy conversion and
management 51,pp 1148 1156
[12] A. Konak et al, (2006) - multi-objective optimization
using genetic algorithms: a tutorial- / reliability
engineering and system safety 91 992 1007
[13] V. Donde, m. Pai, i. Hiskens, (2001), Simulation and
optimization in an agc system after seregulation. Ieee
transactions on power systems, 16(3): 481 488

Volume 2 Issue 3, March 2013

www.ijsr.net 278