Automatic Generation Control of Multi-Area Power System with Generating Rate Constraints Using Computational Intelligence Techniques

International Journal of Applied Power Engineering (IJAPE)
Vol. 2, No. 1, April 2013, pp. 27~38
ISSN: 2252-8792  27
Journal homepage: http://iaesjournal.com/online/index.php/IJAPE
Automatic Generation Control of Multi-Area Power System
with Generating Rate Constraints Using Computational
Intelligence Techniques
P. Subbaraj1
, K. Manickavasagam2
1
Sri Nandhanam College of Engineering and Technology, Vellore, Tamilnadu, India.
2
Department of Electrical and Electronics Engineering, AmritaVishwa Vidya peetham University, Amrita School of
Engineering, Bangalore, India
Article Info ABSTRACT
Article history:
Received Jun 6, 2012
Revised Dec 19, 2012
Accepted Jan 27, 2013
In a large inter-connected system, large and small generating stations are
synchronously connected and hence all stations must have the same
frequency. The system frequency deviation is the sensitive indicator of real
power imbalance. The main objectives of AGC are to maintain constant
frequency and tie-line errors with in prescribed limit. This paper presents two
new approaches for Automatic Generation Control using i) combined Fuzzy
Logic and Artificial Neural Network Controller (FLANNC) and ii) Hybrid
Neuro Fuzzy Controller (HNFC) with gauss membership functions. The
simulation model is created for four-area interconnected power system. In
this four area system, three areas consist of steam turbines and one area
consists of hydro turbine. The components of ACE, frequency deviation
( F) and tie line error ( Ptie) are obtained through simulation model and
used to produce the required control action to achieve AGC using i)
FLANNC and ii) HNFC with gauss membership functions. The simulation
results show that the proposed controllers overcome the drawbacks
associated with conventional integral controller, Fuzzy Logic Controller
(FLC), Artificial Neural Network controller (ANNC) and HNFC with gbell
membership functions.
Keyword:
Area Control Error
Automatic Generation Control
Change in Area Control Error
Frequency deviation
Fuzzy Logic
Fuzzy Logic Controller
Hybrid Neuro Fuzzy Controller
Neural Network controller
Tie line error
Copyright © 2013 Institute of Advanced Engineering and Science.
All rights reserved.
Corresponding Author:
K. Manickavasagam,
Department of Electrical and Electronics Engineering,
AmritaVishwa Vidya peetham University,
Amrita School of Engineering, Bangalore, India.
Email: manicavasagam2003@yahoo.com
1. INTRODUCTION
AUTOMATIC generation control (AGC) is an important problem in power system operation
and control. When ever a small load perturbation occurs, it causes the changes in tie-line power flow and
frequency deviation ( F). Many investigations in the area of AGC of interconnected power systems have
been carried out in the past [1]-[4] and number of control strategies have been proposed to improve the
performance of AGC. In the application of optimal control techniques, the controller design is normally based
on a fixed parameter model of the system derived by a linearization process. Power system parameters
are functions of the operating point. Therefore as the operating conditions change, system
performance with controllers designed for a specific operating point, most likely will not be
satisfactory [2]. Consequently, the non-linear nature of the load frequency control (LFC) problem makes it

 ISSN: 2252-8792
IJAPE Vol. 2, No. 1, April 2013 : 27 – 38
28
difficult to ensure stability for all operating points when an integral controller is used [6],[7]. The application
of adaptive control theory to the LFC problem eliminates some of the problems associated with classical and
modern control [5]. In recent years, intelligent techniques such as artificial neural networks (ANN), fuzzy
logic (FL) and genetic algorithms (GA), have gained increasing interest for the applications in LFC problem.
Some such applications using ANN can be found in [8]-[10]. Design and experience with a FLC for
Automatic Generation Control is explained in [11]. The method of constructing membership function using
stored data and formation of rule base are given in [12]-[14].
In this paper, four-area interconnected power system is considered. First three areas consist of steam
turbines and fourth area consists of hydro turbine. FLANNC and FLC to ANNC switching controller are used
for AGC for four area power system. A simulation model is created for four-area interconnected power
system and the performance of the following controllers is studied:
(i) Conventional integral controller
(ii) Fuzzy Logic Controller (FLC)
(iii) Artificial Neural Network controller (ANNC)
(iv) Combined Fuzzy Logic and Artificial Neural Network controller (FLANNC)
2. MODELING OF THE POWER SYSTEM FOR AGC
The multi-area AGC system used in this paper given in Fig.1 consists of four control areas, which
are connected by tie lines. In each control area, all generators are assumed to form a coherent group. The four
area interconnected power system consists of three re-heat thermal turbine units and one hydro unit.
The transfer function model of the four power system is given in [10], and system parameters are
given in Table 1. Each area supplies power to its user pool and tie lines allows electric power to flow between
areas. Therefore, any load disturbance in one area affects the frequency of its own area, frequency of other
areas and tie line power flow of other areas. Due to this, the control system of each area needs information
about the transient situation in all areas to bring the local frequency to its steady state value. The information
about each area during perturbation is found in its frequency and the information about the other area in the
perturbation is found in its tie-line power flow. In conventional control system, turbine reference power of
each area is tried to be set to its nominal value by an integral controller and the input of the integral controller
of each area is Bi ∆fi + ∆Ptiei (i=1,. 4) called Area Control Error (ACE) of the same area. Parameters Bi are
chosen as 1/Kpi+1/Ri.
Fig .1 Schematic diagram of the power system with four areas
The three areas consist of steam turbines which has governor, re-heater with generation rate
constraints. The hydro turbine in the fourth area is having generation rate constraint. Dead band effect of
governor of hydro turbine and steam turbines are ignored for simplicity. By considering the matters
mentioned above, the state space equations in discrete time domain are obtained as represented in Appendix.
As a result, the state space equations of the power system including four areas are written in discrete time
domain as follows:
x (k) =Ax (k-1) +Bu (k-1)+G (1)
where G is a vector containing non-linear terms. The expressions of matrices A, B and G in Eq. 1: are
obtained by arranging the equations given in Appendix A. The variables used in Appendix A are given in
nomenclature. The input state variables of the four-area power system are given below
xT
= [∆f1, ∆PR1, ∆PG1, ∆Pref 1, ∆xE1, ∆Ptie1, ∆f2, ∆PR2, ∆PG2, ∆Pref 2, ∆xE 2, ∆Ptie2, ∆f3, ∆PR3,
∆PG3, ∆Pref3,
∆xE3, ∆Ptie3, ∆f4,
∆Pref
4, ∆xE4, ∆PR4, ∆PG4, ∆Ptie4] (2)
uT
= [∆PD1, ∆PD2, ∆PD3, ∆PD4] (3)
where the above parameters are given in nomenclature.

IJAPE ISSN: 2252-8792 
Automatic Generation Control of Multi-Area Power System with Generating Rate Constraints (P. Subbaraj)
29
Fig 2. Transfer function simulation model of area1 controlled by FLC
The transfer function simulation model of area1 power system is shown in Fig.2. In practice, there is
a maximum limit on the rate of change in the generating power of a conventional power plant. Hence, if the
response of the applied controller and/or the load change is too fast under transient conditions, then system
non-linearities will prevent its achievement. It follows that a controller designed for the unconstrained
situation may not be suitable when the GRC is considered. The generation rate constraint (GRC) is taken into
account by replacing the turbine block with non-reheat turbine (non-linear). In order to project physical
constraints, generation rate limitations are chosen as 0.1 p.u./min (i.e. 0.0017 p.u. MW/s) for thermal area and
4.5% for hydro area [4].
B1,2,3,4 value is designed by using the expression
B1,2,3,4 = 1/Kpi + 1/Ri (4)
The three areas consist of steam turbines which have governor, re-heater with ramp rate constraints.
The hydro turbine in the fourth area is having generation rate constraint. Dead band effect of governor of
hydro turbine and steam turbines are ignored for simplicity [10].
3. DESIGN OF FLC
Fuzzy logic is a paradigm for an alternative design methodology, which can be applied in developing
both linear and non-linear systems for embedded control. A full control design requires developing a set of
control rules based on available inputs and designing a method of combining all rule conclusions. The design
procedure is given as follows:
3.1. CHOICE OF PROCESS INPUT AND CONTROL OUTPUT
A first step is to choose the correct input signals to the fuzzy logic controller. For this controller,
Area Control Error (ACE) and change in Area Control Error (∆ACE) signals are selected as input to the
controller which is the contents of the rule-antecedent (If-part of a rule). The control output (u) signals
(process input) represents the contents of the rule-consequent (then-part of the rule).
3.2. NORMALIZATION
Normalization performs a scale transformation and it also called input normalization. It maps the
physical values of the current process state variables into a normalized universe of discourse. It also maps the
normalized value of control output variable into its physical domain (output de-normalization). For this
controller, normalization is obtained by dividing each crisp input on the upper boundary value for the
associated universe.
3.3. FUZZIFICATION
Fuzzification is the process of making crisp value to fuzzy quantity. The quantities that are
considered being crisp and deterministic are actually not deterministic at all. Fuzzification is related to the
vagueness and imprecision in a natural language. It could be defined as a mapping from an observed input
space to fuzzy sets in certain input universes of discourse. Fuzzification plays an important role in dealing
with uncertain information, which might be objective or subjective in nature.

 ISSN: 2252-8792
30
3.4. DETERMINATION OF MEMBERSHIP FUNCTION
The range of input and output variables is assigned with linguistic variables. These variables
transform the numerical values of the input to fuzzy quantities. These linguistic variables specify the quality
of the control. As the number of the linguistic variables increases, the computational time increases.
Therefore a compromise between the quality of control and computational time is needed to choose the
number of linguistic variables. The Gauss membership function is chosen in this work. The input and output
variables are assigned with 5 linguistic variables as follows:
1. The Area Control Error (ACE) is classified into:
Negative maximum (ACE -vemax); Negative medium (ACE -vemed); Zero (ACEzero); Positive medium
(ACE +vemed); Positive maximum (ACE +vemax)
2. The change in Area Control Error (∆ACE ) is classified into:
Negative maximum (∆ACE -vemax); Negative medium (∆ACE -vemed); Zero (∆ACEzero); Positive
medium (∆ACE +vemed); Positive maximum (∆ACE +vemax)
3. The output of fuzzy logic controller u is classified into:
Negative maximum (u-vemax); Negative medium (u -vemed); Zero (u zero); positive medium (u +vemed);
Positive maximum (u+vemax).
The input variable ACE and ∆ACE lies with in the range of [-0.0188 0.001] and [-0.009.0.0188].
The control output ‘u’ lies in the range of [0 0.0123]. These input and output ranges are used for designing the
FLC, in which each of input and output set is assigned with five linguistic variables and 25 rules are framed
in fuzzy inference engine.
3.5 KNOWLEDGE BASE
The knowledge base of an FLC comprises of two components, a database and fuzzy control base.
The concepts associated with a database are used to characterize fuzzy control rule. It should be noted that the
correct choice of the membership function plays an essential role in the successful application. A lookup table
based on discrete universes defines the output of a controller for all possible combinations of the input
signals. A fuzzy system is characterized by a set of linguistic statements. It is in the form of “IF-THEN”
rules; these rules are easily implemented by fuzzy conditional statements. In fuzzy logic the collection of
fuzzy control rules, that are expressed as fuzzy conditional statements forms the rule set of an FLC. The
decision table is given in Table. 2.
Table 2.: Rule base of the proposed FLC
Area Control Error
(ACE)
Change in Area Control Error (∆ACE )
ACE
–vemax
ACE
–vemed ACE Zero ACE +vemed ACE +vemax
ACE –vemax u +vemax u +vemax u+vemax u +vemed u zero
ACE –vemed u +vemax u +vemax u +vemed u zero u –vemed
ACE zero u+vemax u +vemed u zero u-vemed u-vemax
ACE +vemed
u
–vemed u zero u –vemed u -vemax u -vemax
ACE +vemax u zero u –vemed u -vemax u -vemax u -vemax
The associated gauss membership function for input variable ACE and its derivative ∆ACE are
shown in Fig.3 and Fig.4. The output variable u is shown in Fig .5.
Fig.3. Membership functions of input variable area1 (ACE)

IJAPE ISSN: 2252-8792 
31
Fig 4.Membership functions of input variable area (∆ACE)
Fig 5.Membership functions of output variable area (u)
3.6. Defuzzification
This process is used to convert a fuzzy value back to the actual crisp output. To get the control
output in fuzzy values, the max-min method is adopted in decision-making. The crisp control output (u) is
obtained by centroid method which is given by,
(5)
where μ is the membership value of the variable recommending the fuzzy controller action, Hi is precise
numerical value corresponding to that fuzzy controller action and i is the index of the fuzzy logic
rules(i=1,….N), N is the total number of rules to be fired.
4. DESIGN OF ANNC
In four-area power system, frequency perturbation in each area has to be brought back to their steady
state values. The ANN controller can be used to provide control input, which succeeds in this deed. ANN
controller is indeed an adaptive nonlinear controller with control strategy defined by the learning rule used in
changing the weights of the synaptic connections [10]. In this paper, a Multi Layer Perception network is
trained between inputs Area Control Error (ACE: the components of ACE are F and Ptie ) and change in
Area Control Error (∆ACE) and control output (u) as given in Fig.6. The training is performed by using
MATLAB 6.0 Neural Network Toolbox, wherein the learning rate and momentum are adjusted internally to
minimize the mean square error with in the prescribed epochs.
Input Hidden
ACE
Layer Layer
Output
Layer
Control action
∆ACE
‘u’
Fig 6. Structure of ANN
The structure of ANN is given as : {1x1 cell} of input, {2x1 cell} of layers, {1x2 cell} containing 1
output, {1x2 cell} target 1, {2x1 cell} containing 2 biases, {2x1 cell} containing 1 input weight and {2x2

 ISSN: 2252-8792
32
cell} containing 1 layer weight. After the network is trained, an ANN control block is created using the
command “gensim”. In ANNC, the training data and test data are formed in between ACE and ∆ACE and
control output ‘u’.
5. FLANNC
The combination of Fuzzy system and Artificial Neural Network has the advantages of each of them.
The output of FLC and ANNC are added together to get the required control output. The combination of this
controller is called FLANNC. The block diagram of FLANNC is shown in Fig .7
Fig. 7 Block diagram of FLANNC
6. HYBRID NEURO FUZZY CONTROLLER (HNFC)
A neuro-fuzzy system is a fuzzy system that uses a learning algorithm derived from neural network
theory to determine its parameters (fuzzy sets and fuzzy rules) by processing data samples. Hybrid neuro-
fuzzy results are obtained from fusion of neural network and fuzzy logic. Hybrid neuro-fuzzy system is
designed using simulink / MATLAB 6.0.
Steps to design HNF controller [16] as follows:
1. Draw the simulink model with FLC and simulate it with the given rule base.
2. The first step to design the HNF controller is collecting the training data while simulating with FLC.
3. The two inputs, i.e., ACE and d(ACE)/dt and the output signal gives the training data.
4. Use anfisedit to create HNF.fis file.
5. Load training data collected in step.1 and generate the Fuzzy Inference System (FIS) with gauss MF’s.
6. Train the collected data with generated FIS upto particular number of Epochs.
The proposed scheme utilizes sugeno-type fuzzy inference system controller, with the parameters
inside the fuzzy inference system decided by the neural-network back propagation method [17]. The ANFIS
is designed by taking ACE and rate of change of ACE as inputs. Another HNFC is designed using gauss
membership functions instead of gbell membership functions which is used in [16].
7. RESULT AND ANALYSIS
7.1. CONTROLLERS RESPONSE UNDER VARIOUS CONDITIONS
In this section, the performances of integral controller, FLC, ANNC and FLANNC are plotted
under sudden increase in load and different ramp rates. The model is simulated up to 1000 sec for all the
cases.
7.1.1. SUDDEN INCREASE IN LOAD
The model is tested for sudden increase in load by applying step input of 1%. The plot of ΔF and
ΔPtie of FLANNC is compared with FLC, ANNC and integral controller for four areas given in Fig 8. The
performances of the controllers are given in Figs 8-9. From the figures, it can be observed that the first

IJAPE ISSN: 2252-8792 
33
undershoot is same for all the controllers and the first overshoot is 0.04 for integral controller, 0.034 for
both FLC, 0.022 for ANNC and FLANNC. From Fig 8.a, it is observed that FLC settles at 450 sec with
2% steady state error, ANNC settles at 470 sec with 2% steady state error and FLANNC settles at 420 sec
with zero steady state error. Fig 8 shows the tie line error of the area 1.The performance of the FLANNC
is better than other controllers in terms of settling time and steady state error.
Fig 8.Performance of controllers for ΔF in Area 1 for increase in load
Fig. 8.a. Performance of controllers for ΔF in Area 1 for increase in load focused between 350sec -550 sec
Fig. 9 Performance of controllers for ∆Ptie in Area 1 for increase in load
7.1.2 EFFECT OF RAMP RATES
In this model, for getting different ramp rates the values of Tp1 and Tp3 are chosen as 50. Figs 10-12
display the effect of low ramp rate on F Area1and area 2. Fig 12 displays the response of ΔPtie to the effect
of low ramp rate in area 1.

 ISSN: 2252-8792
34
Fig 10 Response to lower ramp rate for Area 1 (Tp1=50)
Fig 10.a. Response to lower ramp rate for Area 1 (Tp1=50) Focused between 500 sec -1200 sec
Fig 11 Response to lower ramp rate for Area 2(Tp2 = 20)
Fig 12 Response of Ptie for lower ramp rate in Area1(Tp2 = 50)

IJAPE ISSN: 2252-8792 
35
The performances of the controllers are given in Figs 10-12. From the figures, it can be observed
that the first undershoot is same for all the controllers and the first overshoot is 0.04 Hz for integral controller,
0.017 Hz for FLC, 0.018Hz for ANNC and 0.012 for FLANNC. From Fig 10.a, it is observed that integral
controller settles at 1200 sec with 0.8% of steady state error, FLC settles at 800 sec with 2% steady state
error, ANNC settles at 850 sec with 1% steady state error and FLANNC settles at 790 sec with 1% steady
state error. The performance of the FLANNC is better than other controllers in terms of settling time and
steady state error. Fig.10 and Fig.11 also indicates that if the ramp rate is low, the response of generator will
be slow and settling time is higher. The model is simulated up to 2000 sec.
7.2. RESPONSE OF FLANNC AND HNFC
In this section, the performances of FLANNC is compared with HNFC with gbell membership
functions and HNFC with gauss membership functions, under sudden increase in load, sudden decrease in
load and different ramp rate. The model is simulated up to 1000 sec for all the cases. HNFC approach to AGC
for two area system given in [16],[17] used gbell membership functions. In this work, HNFC with gauss
membership functions is used to achieve AGC of four area system.
7.2.1. SUDDEN INCREASE IN LOAD
The model is tested for sudden increase in load by applying step input of 1%. The plot of ΔF and
ΔPtie of FLANNC, HNFC with gauss membership functions and HNFC with gbell membership functions are
compared for four areas given in Fig 13 and Fig 14 for increase in load. From the figures, it can be observed
that the first undershoot is same for all the controllers and the first overshoot is also same for all the
controllers. The second undershoot 0.02 Hz for FLANNC, 0.028 Hz for HNFC with gauss membership
functions 0.018Hz for HNFC with gbell membership functions. From Fig 13.a, it is observed that FLANNC
settles at 440 sec with zero steady state error, HNFC with gauss membership function settles at 460 sec with
1% steady state error and HNFC with gbell membership functions settles at 475 sec with 3% steady state
error. The performance of the FLANNC is better than HNFC with gauss membership functions controller and
gbell membership functions controller in terms of settling time and steady state error.
Fig. 13 Performance of controllers for ΔF in Area 1 for increase in load
Fig. 13.a Performance of controllers for ΔF in Area 1 for increase in load Focused between 300 sec-550 sec

 ISSN: 2252-8792
36
Fig. 14 Performance of controllers for ∆Ptie in Area 1 for increase in load
7.2.2. EFFECT OF RAMP RATES
In this model, for getting different ramp rates the values of Tp1 and Tp3 are chosen as 50. Figs 15-
16 display the effect of low ramp rate in area 1and 2. Fig 15 and Fig 16 also indicates that if the ramp rate
is low, the response of generator will be slow and settling time is higher. The model is simulated up to
2000 sec. Fig 15.a indicates the response to ramp rate for Area 1 (Tp1=50) focused between 500sec-
1000sec.
Fig 15 Response to ramp rate for Area 1 (Tp1=50)
Fig 15.a Response to ramp rate for Area 1 (Tp1=50) Focused between 500sec-1000sec
Fig 16 Response of ΔPtie for lower ramp rate in Area1(Tp2 = 50)

IJAPE ISSN: 2252-8792 
37
8. CONCLUSION
This paper presents new approaches for Automatic Generation Control using the FLANNC. The
simulation model is developed for four-area interconnected power system using simulink/MATLAB 6.0. The
FLANNC performance is compared with FLC, ANNC and conventional integral controller. From the result, it
is concluded that FLANNC settles faster than FLC, ANNC and integral controller. The magnitude of F with
FLANNC is also small compared to other controllers. Recently, HNFC application to AGC is presented in
[16],[17] in which gbell membership functions are used. In this work, the gbell membership functions are
replaced by gauss membership functions. The performance of FLANNC, HNFC with gauss membership
functions and HNFC with gbell membership are given in Figs. From the result, it can be concluded that
HNFC with gauss membership functions are better than HNFC with gbell membership functions. FLANNC is
better than FLC, ANNC, and HNFC with gauss membership functions and HNFC with gbell membership
functions.
Appendix A
An extended power system can be divided into a number of load frequency control areas
interconnected by means of tie lines as explained in [12]. Modern control theory is applied in this section to
design an optimal AGC for a four area system.
Thermal Systems (Area 1 to 3)
∆Prefi (k) = ∆Prefi (k-1) – KIiT (Bi∆fi (k-1) +∆Pi (k-1)) (1)
∆P1 (k-1) = ∆Ptie1 (k-1) + a31∆Ptie3 (k-1) + a41∆Ptie4 (k-1)
∆P2 (k-1) = ∆Ptie2 (k-1) + a12∆Ptie1 (k-1)
∆P3 (k-1) = ∆Ptie3 (k-1) + a23∆Ptie2 (k-1)
∆Ptie1(k) = ∆Ptie1(k-1) + 2πT( T12 + T13 +T14 )∆f1(k-1) – 2πT( T12∆f2(k-1) + T13∆f3(k-1) + T14∆f4(k-1) )
∆Ptie2(k) = ∆Ptie2(k-1) + 2πT( T21 + T23 )∆f2(k-1)– 2πT( T21∆f1(k-1) + T23∆f3(k-1) )
∆Ptie3(k) = ∆Ptie3(k-1) + 2πT( T31 + T32 )∆f3(k-1) – 2πT( T31∆f1(k-1) + T32∆f2(k-1) )
∆xEi (k) = (1-T/TGi) ∆xEi (k-1) + T/TGi ( ∆Prefi(k-1) – 1/Ri ∆fi(k-1) ) (2)
∆PRi (k) = (1 – T/TTi) ∆PRi (k-1) + T/TTi ∆xEi (k-1) (3)
∆PGi (k) = (1 – T/TRi) ∆PGi (k-1) + KRi∆PRi (k) + (T/Tri – KRi) ∆PRi (k-1) (4)
∆fi(k) = ( 1 – T/TPi )∆fi(k-1) + KPiT/TPi +( ∆PGi(k-1) – ∆PDi(k-1) – ∆Ptiei(k-1) ) (5)
Where, Subscript i represents each thermal area in power system (i = 1, 2, 3).
Hydro System (Area 4)
∆Pref4 (k) = ∆Pref4 (k-1) – KI4T (B4∆f4 (k-1) +∆P4 (k-1)) (6)
∆Ptie4 (k) = ∆Ptie4 (k-1) + 2πT41T (∆f4 (k-1) – ∆f1 (k-1))
∆xE4 (k) = (1 – T/T1) ∆xE4 (k-1) + T/T1 (∆Pref4 (k-1) – 1/R4 ∆f4 (k-1)) (7)
∆PR4 (k) = (1 – T/T3) ∆PR4 (k-1) + (T-T2)/T3 ∆xE4 (k-1) + T2/T3 ∆xE4 (k) (8)
∆PG4 (k) = (1 – 2T/Tw) ∆PG4 (k-1) + 2(1+ T/Tw) ∆PR4 (k-1) – 2∆PR4 (k) (9)
∆f4 (k) = (1 – T/TP4) ∆f4 (k-1) + KP4T/TP4 + ( ∆PG4(k-1) –∆PD4(k-1) – ∆Ptie4(k-1) ) (10)
Here ∆P4 = ∆Ptie4
REFERENCES
[1] J.Nanda, B.Kaul. “Automatic generation control of an interconnected power system”, IEE Proceedings, Vol/Issue:
125(5). Pp. 385-390, 1978.
[2] O.I.Elgerd, C.E. Fosha. “Optimum megawatt frequency control of multi area electric energy systems”, IEEE
Transcations on PAS, Vol. 89. Pp. 556-563, 1970.
[3] R.K.Green. “Transformed automatic generation control”, IEEE Transcations on Power Systems, Vol/Issue: 11(4).
Pp. 1799-1804, 1996.
[4] G.V. Hicks, B.Jayasurya, W.F.Snow. “An investigation of automatic generation control for an isolated power
system”, IEEE PES winter meeting. Pp. 31-34, 1997.
[5] Y.L.Karnavas, D.P.Papadopoulos. “AGC for autonomous power system using combined intelligent techniques”,
Electrical Power Systems Research, Vol. 62. Pp. 225-239, 2002.
[6] T. Hiyama. “Application of rule based stabilizing the controller to electric power systems”, IEE Proceedings,
Vol/Issue: 136(3). Pp. 175-181, 1989.
[7] J. Talaq and F. Al-Basri. “Adaptive fuzzy gain scheduling for load-frequency control”, IEEE Trans. Power Syst.,
Vol/Issue: 14(1). Pp.145-150, 1999.
[8] K.Sabahil, M.A.Nekouil, M.Teshnehlab, M.Aliyari and M.Mansouri. “Load frequency control in interconnected
power system using dynamic modified neural networks”, Mediterranean conference on control and automation. Pp.
1-5, 2007.
[9] D.M.V.Kumar. “Intelligent controllers for automatic generation control”, TENCON ’98. IEEE region 10th
International conference on global connectivity in Energy, computer, computer and control, Vol. 2. Pp. 557-574,

 ISSN: 2252-8792
38
1998.
[10] H.L.Zeynelgil, A.Demiroren, N.S.Sengor. “The application of ANN technique to Automatic generation control for
multi-area power system”, Electrical Power Energy Systems ,Vol. 24. Pp. 345-354, 2002.
[11] G.A.Chown and R.C.Hartman. “Design and experiences with a FLC for automatic generation control (AGC)”, IEEE
Transactions on Power systems. Pp. 965-970, 1998.
[12] Tzu-Ping Wu and Shyi-Ming Chen. "A new method for constructing Membership functions and fuzzy rules from
training examples", IEEE transactions on systems and cybernetics-part B cybernetics, Vol/Issue: 29(1). Pp. 25-40,
1999.
[13] D.Dubios and H.Prade. "What are fuzzy rules and how to use them" Fuzzy sets and system, Vol/Issue: 84(2). Pp.169-
185, 1996.
[14] T.P. Hong and Lee. “Induction of fuzzy rules and membership functions from training examples", Fuzzy sets and
system, Vol/Issue: 84(1). Pp.33-47, 1996.
[15] Stamatios V. Kartalopoulos. Understanding neural networks and fuzzy logic, IEEE Press, Prentice-Hall of India,
Indian Society of Technical Education, 2002.
[16] Gayadhar Panda, Sidhartha Panda and Cemel Ardil. “Automatic generation control of an interconnected power
system with generation rate constraints by hybrid neuro fuzzy approach”, International journal of electrical power
and energy systems engineering, Vol/Issue: 2(1). Pp. 13-18, 2009.
[17] Gayadhar Panda, Sidhartha Panda and C.Ardil. “Hybrid Neuro Fuzzy approach for Automatic generation control of
two-area inter connected power system”, International journal of computational intelligence , Vol/Issue: 5(1). Pp.
80-84, 2009.
BIOGRAPHY OF AUTHOR
K.Manickavasagam received the PhD from Madurai Kamaraj University, M.E.Degree from
Thiagarajar College of engineering, Madurai, Tamilnadu, India. Currently he is working in
Gopalan College of Engineering and Management, Bangalore, Karnataka, India. His reseach
interests includes Power generation and control and artificial intelligence applications in power
systems. He is a member of the Institute of engineers (India) and life member of Indian Society
of Technical Education.

Automatic Generation Control of Multi-Area Power System with Generating Rate Constraints Using Computational Intelligence Techniques

More Related Content

What's hot (20)

Similar to Automatic Generation Control of Multi-Area Power System with Generating Rate Constraints Using Computational Intelligence Techniques (20)

More from IJAPEJOURNAL (20)

Recently uploaded (20)

Automatic Generation Control of Multi-Area Power System with Generating Rate Constraints Using Computational Intelligence Techniques