Sei sulla pagina 1di 15

Electrical Power and Energy Systems 49 (2013) 1933

Contents lists available at SciVerse ScienceDirect

Electrical Power and Energy Systems


journal homepage: www.elsevier.com/locate/ijepes

DE optimized parallel 2-DOF PID controller for load frequency control


of power system with governor dead-band nonlinearity
Rabindra Kumar Sahu , Sidhartha Panda, Umesh Kumar Rout
Department of Electrical Engineering, Veer Surendra Sai University of Technology (VSSUT), Burla 768 018, Odisha, India

a r t i c l e

i n f o

Article history:
Received 14 September 2012
Received in revised form 13 December 2012
Accepted 30 December 2012
Available online 1 February 2013
Keywords:
Load Frequency Control (LFC)
Governor dead-band nonlinearity
Generation Rate Constraint (GRC)
Parallel 2-Degree Freedom of ProportionalIntegral-Derivative (2-DOF PID) controller
Differential Evolution (DE) algorithm

a b s t r a c t
In this paper, design and performance analysis of Differential Evolution (DE) algorithm based parallel 2Degree Freedom of Proportional-Integral-Derivative (2-DOF PID) controller for Load Frequency Control
(LFC) of interconnected power system is presented. A two area thermal system with governor dead-band
nonlinearity is considered for the design and analysis purpose. The design problem is formulated as an
optimization problem and DE is employed to search for optimal controller parameters. Conventional
and modied objective functions are used for the design purpose. Conventional objective functions
employed in the paper are Integral of Time multiplied by Squared Error (ITSE) and Integral of Squared
Error (ISE). In order to further increase the performance of the controller, a modied objective function
is derived using Integral Time multiply Absolute Error (ITAE), damping ratio of dominant eigenvalues,
settling times of frequency and peak overshoots with appropriate weight coefcients. The superiority
of the proposed approach has been demonstrated by comparing the results with a recently published
technique, i.e. Craziness based Particle Swarm Optimization (CPSO) for the same interconnected power
system. Further, sensitivity analysis is performed by varying the system parameters and operating load
conditions from their nominal values. It is observed that the proposed controllers are quite robust for
a wide range of the system parameters and operating load conditions from their nominal values. Finally,
the proposed approach is extended to a more realistic power system model by considering the physical
constraints such as time delay, reheat turbine, Generation Rate Constraint (GRC) and governor dead band.
2013 Elsevier Ltd. All rights reserved.

1. Introduction
The growth in size and complexity of modern electric power
systems along with increase in power demand has necessitated
the use of intelligent systems that combine knowledge, techniques
and methodologies from various sources for the real-time control
of power systems. A modern power system network consists of a
number of utilities interconnected together and power is exchanged between utilities over tie-line by which they are interconnected. For the stable operation of power systems, both constant
frequency and constant tie-line power exchange should be maintained. In each area, a Load Frequency Controller (LFC) monitors
the system frequency and tie-line ows, computes the net change
in the generation required (generally referred to as Area Control
Error-ACE) and changes the set position of the generators within
the area so as to keep the time average of the ACE at a low value
[1]. Therefore ACE, which is dened as a linear combination of
power net-interchange and frequency deviations, is generally
taken as the controlled output of LFC. As the ACE is driven to zero
Corresponding author. Tel.: +91 9439702316.
E-mail addresses: rksahu123@gmail.com (R.K. Sahu), panda_sidhartha@rediffmail.
com (S. Panda), umesh6400@gmail.com (U.K. Rout).
0142-0615/$ - see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.ijepes.2012.12.009

by the LFC, both frequency and tie-line power errors will be forced
to zeros [2].
The researchers in the world over are trying to employ several
strategies for LFC of power systems in order to maintain the system
frequency and tie line ow at their scheduled values during normal
operation and also during disturbance conditions. In [3], a critical
literature review on the LFC of power systems has been presented,
where various control aspects concerning LFC problem have been
studied. There has been considerable research work attempting
to propose better LFC control systems based on modern control
theory [4], neural network [5,6], fuzzy system theory [7], reinforcement learning [8] and ANFIS approach [9]. From the literature survey, it may be concluded that there is still scope of work on the
optimization of controller parameters to further improve the system performance. For this, various novel controller structures
and evolutionary optimization techniques can be proposed and
tested for comparative optimization performance study. New articial intelligence-based approaches have been proposed recently
to design a controller. Nanda et al. [10] have shown that Bacterial
Foraging Optimization Algorithm (BFOA) optimized controller provides better performance as compared to classical and Genetic
Algorithm (GA) based controllers for an interconnected power system. Ali and Abd-Elazim [11] have recently presented a BFOA

20

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

based Proportional Integral (PI) controller which provides better


performance as compared to that with GA based PI controller in
two area non-reheat thermal system. Gozde and Taplamacioglu
[12] proposed a gain scheduling PI controller for an Automatic
Generation Control (AGC) system consisting of two area thermal
power system with governor dead-band nonlinearity. The authors
have employed a Craziness based Particle Swarm Optimization
(CPSO) with different objective functions to minimize the settling
times and standard error criteria. Several classical controllers
structures such as Integral (I), Proportional-Integral (PI), IntegralDerivative (ID), Proportional-Integral-Derivative (PID) and IntegralDouble Derivative (IDD) have been used in LFC [13]. It clear
from literature survey that the performance of the system not only
depends on the articial intelligent techniques employed but also
on controller structure and selection of objective function. Two-degree-of-freedom (2-DOF) controllers have an advantage over the
classical single degree of freedom ones from the point of view of
achieving high performance in set-point tracking and the regulation in the presence of disturbance inputs. Surprisingly, in spite
of these advantages, 2-DOF controllers structures are not attempted for the LFC problems. In view of the above, a maiden attempt has been made in this paper for the application of parallel
2-Degree Freedom of Proportional-Integral-Derivative (2-DOF
PID) controller for Load Frequency Control of interconnected
power system.
Differential Evolution (DE) is a population-based direct search
algorithm for global optimization capable of handling non-differentiable, non-linear and multi-modal objective functions, with
few, easily chosen, control parameters [14]. DE uses weighted differences between solution vectors to change the population,
whereas in other stochastic techniques such as GA and Expert Systems (ES), perturbation occurs in accordance with a random quantity. DE employs a greedy selection process with inherent elitist
features. Also it has a minimum number of control parameters,
which can be tuned effectively [15]. In view of the above, an attempt has been made in this paper for the optimal design of DE
based 2-DOF PID controller for LFC in two area interconnected
power system considering the governor dead-band nonlinearity.
The design problem of the proposed controller is formulated as
an optimization problem and DE is employed to search for optimal
controller parameters. By minimizing the proposed objective functions, in which Integral Time multiply Absolute Error (ITAE), damping ratio of dominant eigenvalues, settling times and peak
overshoots of frequency and tie line power deviations are involved;
dynamic performance of the system is improved. Simulations results are presented to show the effectiveness of the proposed controller in providing good damping characteristic to system
oscillations over a wide range of loading conditions, and system
parameters. Further, the superiority of the proposed design approach is illustrated by comparing the proposed approach with recently published CPSO approach [12] for the same interconnected
power system.

2. Power system under study


A two area interconnected power system of thermal plant as
shown in Fig. 1 is considered for design and analysis purpose. Each
area has a rating of 2000 MW with a nominal load of 1000 MW.
The system is widely used in literature for the design and analysis
of automatic load frequency control of interconnected areas [12].
In Fig. 1, B1 and B2 represent the frequency bias parameters; ACE1
and ACE2 stands for area control errors; u1 and u2 are the control
outputs form the controller; R1 and R2 represent the governor
speed regulation parameters in p.u. Hz; TG1 and TG2 are the speed
governor time constants in sec; DPG1 and DPG2 are the change in

governor valve positions (p.u.); TT1 and TT2 are the turbine time
constant in sec; DPT1 and DPT2 are the change in turbine output
powers; DPD1 and DPD2 are the load demand changes; DPTie is
the incremental change in tie line power (p.u); KPS1 and KPS2 stands
for the power system gains; TPS1 and TPS2 represent the power system time constant in sec; T12 is the synchronizing coefcient and
Df1 and Df2 are the system frequency deviations in Hz. The relevant parameters are given in Appendix A.
The speed governor dead band has a great effect on the dynamic
performance of electric energy system. For more realistic analysis
the governor dead band has to be included which makes the system non-linear. Governor dead band is dened as the total amount
of a continued speed change within which there is no change in
valve position. Steam turbine dead band is due to the backlash in
the linkage connecting the servo piston to the camshaft. Much of
this appears to occur in the rack and pinion used to rotate the camshaft that operates the control valves. Due to the governor dead
band, an increase/decrease in speed can occur before the position
of the valve changes. The speed-governor dead band has makes
the system oscillatory. A describing function approach is used to
include the governor dead band nonlinearity.
The governor dead-band nonlinearity causes sustained sinusoidal oscillation of natural period of about T0 = 2 s, i.e. the frequency
of oscillation f0 = 0.5 Hz. The governor dead-band nonlinearity
equations are linearized in terms of change and rate of change in
the speed. The governor dead-band nonlinearity is dened as:

_
e Fg; g

In the above equation, g is taken as sinusoidal oscillation with a


frequency of oscillation of 0.5 Hz, given by:

g A sin2 p f 0 t

The function F is computed as a Fourier series as:

_ F0 K1g
Fg; g

K 2 dg

2 p f 0 dt

In the above equation, F0 is equal to zero as the governor dead


band nonlinearity is symmetrical about the origin. Neglecting the
fourth and higher order terms and assuming a backlash of 0.5%,
the Fourier coefcients are obtained as: K1 = 0.8 and K2 = 0.2.
So, the transfer function of governor with nonlinearity can be expressed as [12]:

Gg

0:8  0:2
p s
1 sT g

3. The 2-DOF PID controller


The degree of freedom of a control system is dened as the
number of closed-loop transfer functions that can be adjusted
independently. While designing a control system various performance criteria are to be satised so a two-degree-of-freedom
control system naturally has advantages over the conventional
single degree of freedom control system [16]. The 2-DOF controller produces an output signal based on the difference between a
reference signal and a measured system output. It computes a
weighted difference signal for each of the proportional, integral,
and derivative actions according to the specied set point
weights. The controller output is the sum of the proportional,
integral, and derivative actions on the respective difference signals, where each action is weighted according to the chosen gain
parameters [17].
Proportional Integral (PI) controllers are the most often type
used today in industry. A control without Derivative (D) mode is
used when: fast response of the system is not required, large dis-

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

21

Fig. 1. Transfer function model of two-area thermal system with governor dead band.

turbances and noises are present during operation of the process


and there are large transport delays in the system. Proportional
Integral Derivative (PID) controllers are used when stability and
fast response are required. Derivative mode improves stability of
the system and enables increase in proportional gain and decrease
in integral gain which in turn increases speed of the controller response. Owing to the above advantages, use of PID controller [17]
and an Integral Double Derivative (IDD) controller [13] for AGC has
also been reported in literature. However, when the input signal
has sharp corners, the derivative term will produce unreasonable
size control inputs to the plant. Also, any noise in the control input
signal will result in large plant input signals. These reasons often
lead to complications in practical applications. The practical solution to the these problems is to put a rst lter on the derivative
term and tune its pole so that the chattering due to the noise does
not occur since it attenuates high frequency noise. To improve performance when there is noise or random error in the measured
process variable, a derivative lter is employed. The derivative lter works to limit the large controller output shifts that derivative
action causes as a result of measurement noise. The derivative lter can also help to reduce the constant controller output uctuations that can lead to wear in the nal control element. In view of
the above a lter is used for the derivative term in the present
paper.
The structure of proposed parallel 2-DOF PID controller is
shown in Fig. 2, where R(s) represents the reference signal, Y(s)
represents the feedback from measured system output and U(s)
represents the output signal, KP, KI and KD are the proportional,
integral and derivative gains respectively, PW and DW are the proportional and derivative set point weights respectively, and N is the
derivative lter coefcient. A 2-DOF PID control system is shown in
Fig. 3. In this gure, C(s) is a single degree-of-freedom controller,
D(s) is the load disturbance and F(s) acts as a pre lter on the reference signal. For a parallel two-degree-of-freedom PID controller,
C(s) and F(s) are given by:

F s

PWK P DWK D s2 PWK P N K I s K I N


K P K D N s2 K P N K I s K I N

C s

K P K D N s2 K P N K I s K I N
s s N

For the design of a modern heuristic optimization technique


based PID controller, the objective function is rst dened based
on the desired specications and constraints. The selection of
objective function to optimize the controller parameters is usually
based on performance criteria which depend on system response.
In time domain systems the desired specications are peak overshooting, rise time, settling time, and steady-state error. Performance criteria that are usually reported in literature are the
Integral of Time multiplied Absolute Error (ITAE), Integral of
Squared Error (ISE), Integral of Time multiplied Squared Error
(ITSE) and Integral of Absolute Error (IAE). It has been shown that
the ITAE provides better responses as compared to other criteria
[18]. Also, the eigenvalues and modal analysis provides an extension of analytical methods to study the stability of a system. Eigen
value analysis investigates the dynamic behavior of the system
under different characteristic frequencies (modes). In a power system, it is required that all modes be stable and all electromechanical oscillations be damped out as quickly as possible. In other
words, the damping ratios of dominant eigenvalues should be
maximized as much as possible [1,18].
In order to convergence to optimal solution, three different
objective functions are considered in the present paper. The rst
and second objective functions consider only ISE and ITSE criteria
given by Eqs. (7) and (8), respectively. The third objective function
J3 tries to minimize the settling times and overshoots of Df1, Df2
and DPTie, maximize the damping ratio (f) and minimize an error
criteria (ITAE) as given by

J 1 ISE

t sim

Df1 2 Df2 2 DP Tie 2  dt

22

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

Fig. 2. Two degree of freedom (2-DOF) PID control structure.

Fig. 3. 2-DOF control system.

J 2 ITSE

t sim

i
Df1 2 Df2 2 DPTie 2  t  dt

J3

t sim

x1  jDf1 j jDf2 j jDPTie j  t  dt x2

min

Pn

i1 1

 fi

 x3  T S x4  OS

where Df1 and Df2 are the system frequency deviations; DPTie is the
incremental change in tie line power; tsim is the time range of simulation; TS is the sum of the settling times of frequency and tie line
power deviations; OS is the sum of the peak overshoot of frequency
and tie line power deviations; x1x4 are weighting factors. Inclusion of appropriate weighting factors to the right hand individual
terms helps to make each term competitive during the optimization
process. Wrong choice of the weighting factors leads to incompatible numerical values of each term involved in the denition of tness function which gives misleading result. The weights are so
chosen that numerical value of all the terms in the right hand side
of Eq. (9) lie in the same range. Repetitive trial run of the optimizing
algorithms reveals that numerical value of ITAE lie in the range 0.6
0.05, minimum damping ratio lies in the range 0.050.25, total settling times of Df1, Df2 and DPTie lies in the range 1550 and overshoot value lie in the range 0.0080.0016. To make each term
competitive during the optimization process the weights are chosen
as: x1 = 400, x2 = 1.5, x3 = 1.0 and x4 = 15,000.
The problem constraints are the controller parameter bounds.
Therefore, the design problem can be formulated as the following
optimization problem.

Minimize J

10

Subject to K P min 6 K P 6 K P max ;


K Imin 6 K I 6 K Imax and K D min 6 K D 6 K D max
PW min 6 PW 6 PW max ;
DW min 6 DW 6 DW max and Nmin 6 N 6 Nmax

11

where J is the objective function (J1, J2, J3) and KPmin, K Imin ; K P max , K Imax
and K D min , K Dmax are the minimum and maximum value of the

control parameters, PW min , DWmin and PWmax, DWmax are the minimum and maximum values of proportional and derivative set point
weights respectively, Nmin and Nmax are the minimum and maximum values of derivative lter coefcient.
The controller gains must be chosen small enough so that the
area generators do not chase load offsets of short durations [2].
To satisfy the above requirement, the minimum and maximum values of PID controller parameters are chosen as 0 and 1.0. The same
range of PID controller parameters is generally used for an AGC
system [10]. The set points weights PW and DW control the control
action when a reference change takes place and lie in the range of
5. The derivative lter time constant N is generally a value greater
than one [19]. In view of the above, the ranges of PW, DW and N are
taken as 0 and 1.0, 0 and 5 and 10 and 300 respectively.
4. Differential Evolution
Differential Evolution (DE) algorithm is a search heuristic algorithm introduced by Stron and Price [14]. It is a simple, efcient,
reliable algorithm with easy coding. The main advantage of DE
over GA is that GA uses crossover operator for evolution while
DE relies on mutation operation. The mutation operation in DE is
based on the difference of randomly sampled pairs of solutions
in the population. An optimization task consisting of D variables
can be represented by a D-dimensional vector. A population of NP
solution vectors is randomly initialized within the parameter
bounds at the beginning. The population is modied by applying
mutation, crossover and selection operators. DE algorithm uses
two generations; old generation and new generation of the same
population size. Individuals of the current population become target vectors for the next generation. The mutation operation produces a mutant vector for each target vector, by adding the
weighted difference between two randomly chosen vectors to a
third vector. A trial vector is generated by the crossover operation
by mixing the parameters of the mutant vector with those of the
target vector. The trial vector substitutes the target vector in the
next generation if it obtains a better tness value than the target
vector. The evolutionary operators are described below [15,20,21].
4.1. Initialization of parameter
DE begins with a randomly initiated population of size NP of D
dimensional real-valued parameter vectors. Each parameter j lies
within a range and the initial population should spread over this
range as much as possible by uniformly randomizing individuals
within the search space constrained by the prescribed lower bound
X Lj and upper bound X Uj .

23

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

4.2. Mutation operation

Table 1
Tuned 2-DOF-PID controller parameters for different objective functions.

For the mutation operation, a parent vector from the current


generation is selected (known as target vector), a mutant vector
is obtained by the differential mutation operation (known as donor
vector) and nally an offspring is produced by combining the donor with the target vector (known as trial vector). Mathematically
it can be expressed as:

V i;G1 X r1;G F  X r2;G  X r3;G

12

where Xi,G is the given parameter vector, Xr1,G Xr2,G Xr3,G are randomly selected vector with distinct indices i, r1, r2 and r3, Vi,G+1is
the donor vector and F is a constant from (0, 2).
4.3. Crossover operation
After generating the donor vector through mutation the crossover operation is employed to enhance the potential diversity of
the population. For crossover operation three parents are selected
and the child is obtained by means of perturbation of one of them.
In crossover operation a trial vector Ui,G+1 is obtained from target
vector (Xi,G) and donor vector (Vi,G). The donor vector enters the
trial vector with probability CR given by:


U j;i;G1

V j;i;G1

if

randj;i 6 CR or

j Irand

X j;i;G1

if

randj;i > CR or

jIrand

13

with randj,i  U(0, 1), Irand is a random integer from (1, 2, . . . , D),
where D is the solutions dimension, i.e. number of control variables. Irand ensures that V i;G1 X i;G .
4.4. Selection operation
To keep the population size constant over subsequent generations, selection operation is performed. In this operation the target
vector Xi,G is compared with the trial vector V i;G1 and the one with
the better tness value is admitted to the next generation. The
selection operation in DE can be represented by:


X i;G1

U i;G1
X i;G

if f U i;G1 < f X i;G


otherwise:

14

where i e [1, NP].


5. Results and discussions
5.1. Implementation of DE
The effectiveness, efciency, and robustness of the DE algorithm
are sensitive to the settings of the control parameters. The control
parameters in DE are step size function also called scaling factor
(F), crossover probability (CR), the number of population (NP), initialization, termination and evaluation function. F controls the
amount of perturbation in the mutation process and generally lies
in the range (0, 1). CR constants are generally chosen from the
interval (0.5, 1). Several strategies can be employed in DE optimization algorithm. The strategy in a DE algorithm is denoted by DE/x/
y/z, where x represents the mutant vectors, y represents the number of difference vectors used in the mutation process and z represents the crossover scheme used in the crossover operation. The
suggested choice of control parameters are [14] population size
of NP = 60 (NP = 5D, where D = dimensionality of the problem), step
size F = 0.9 and crossover probability of CR = 0.9 and these values
are selected in the present paper. The strategy employed is: DE/
best/1/exp. Optimization is terminated by the prespecied number
of generations which is set to 100. The upper and lower bounds of
the parameters are chosen as (1, 1). Simulations were conducted

Objective function/controller
parameters

J1 (ISE)

J2 (ITSE)

J3
(Proposed)

Proportional gain (KP)


Integral gain (KI)
Derivative gain (KD)
N
PW
DW

0.5409
0.9708
0.5144
180.6983
2.0832
0.6462

0.4935
0.7619
0.3007
177.4023
0.5997
2.5641

0.4299
0.7103
0.4193
169.4177
1.0881
0.1892

on an Intel, core 2 Duo CPU of 2.4 GHz and 2 GB RAM computer in


the MATLAB 7.10.0.499 (R2010a) environment. The optimization
was repeated 20 times for each objective functions (J1, J2, J3) and
the best nal solution among the 20 runs is given in Table 1.
6. Results and discussions
To compare the performance of the proposed controller optimized employing each objective functions (J1, J2, J3), a step load increase of 1 % is considered for area-1. Table 2 shows the system
modes, minimum damping ratio and errors with the controller
optimized using each objective function. Also, the settling times
and peak overshoots in Df1, Df2 and DPtie are shown in Figs. 4
and 5 respectively. To show the effectiveness of the proposed approach results are compared with a recently published CPSO technique for the same interconnected power system and for the same
disturbance [12] in Table 2 and Figs. 4 and 5. It can be seen from
Table 2 that minimum ISE and ITSE values are obtained with the
proposed DE optimized 2DOF- PID controller. The ISE value is reduced by 96.01% and ITSE value is reduced by 97.20% with proposed controller compared recently published CPSO results [12].
Also, it is clear from Fig. 4 that the settling times of Df1, Df2 and
DPtie are improved by 66.19%, 66.17% and 54.29% respectively compared to the results presented in literature [12] for the same investigated system with similar objective function ISE. The overshoots
in Df1 and Df2 are reduced by 20% and 52.38% respectively compared to the recently published CPSO results [12] as evident from
Fig. 5 with the same ISE objective function. With ITSE objective
function, the improvements in settling times are 69.05%, 68.67%
and 55.13% for Df1, Df2 and DPtie respectively and overshoots in
Df1 and Df2 are reduced by 69.13% and 78% respectively compared
to the recently reported results [12] as shown in Figs. 4 and 5 for
the same objective function ITSE.
It is clear from Table 2 and Figs. 4 and 5 that the performance of
the system greatly depends on the selection of objective function.
Hence, to further improve the performance of the controller a modied objective function J3 is proposed as given by Eq. (9). It can be
seen from Table 2 and Figs. 4 and 5 that best performance obtained
with proposed objective function J3 when all criterions (error, minimum damping ratios, settling times and peak overshoots) are considered simultaneously. For comparison, the corresponding results
with the best claimed objective function optimized using CPSO
[12] are also provided in Table 2 and Figs. 4 and 5. For this case,
the improvements in settling times are 53.49%, 54.22% and 22.5%
for Df1, Df2 and DPtie respectively and overshoots in Df1 and Df2
are reduced by 69.56% and 94.25% respectively compared to the
best claimed objective function [12] . In all the cases the peak overshoots in DPtie are zero (except two cases, i.e. J1: CPSO PI and J2:
CPSO PI) and not included in the analysis.
The above modal analysis-based small-signal response characteristics shows that the system performance is greatly improved
by applying the proposed controllers. Further, time domain simulations are performed for step load change at different locations.
A step increase in demand of 1% is applied at t = 0 s in area-1.

24

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

Table 2
System modes, minimum damping ratio and errors.
Objective function

System modes

Errors

DE optimized 2-DOF-PID controller (ISE objective function: J1)

185.25
185.24
0.47 7.37i
0.97 7.23i
1.30 1.55i
0.94 1.07i
0.30

0.0631

ISE

0.8940  104

CPSO optimized PI controller [12] (ISE objective function: J1)

6.6544
6.6456
0.1747 2.8449i
0.6568 1.6041i
1.0751
0.3045
0.4242

0.0613

ISE

22.4086  104

DE optimized 2-DOF-PID controller (ITSE objective function: J2)

181.78
181.75
0.47 7.30i
1.35 7.24i
0.42
0.65 1.07i
1.34 0.76i

0.0640

ITSE

1.0125  104

CPSO optimized PI controller [12] (ITSE objective function: J2)

6.6314
6.6080
0.1650 2.8129i
0.5705 1.5379i
0.9516
0.4703
0.6343

0.0586

ITSE

36.2505  104

DE optimized 2-DOF-PID controller (Proposed objective function: J3)

172.61
172.61
1.18 6.13i
1.67 5.89i
1.26 1.61i
0.93 1.12i
0.31

0.1896

ISE
ITSE
ITAE
IAE

1.2229  104
0.8173  104
5.9727  102
2.6784  102

CPSO optimized PI controller [12] (Objective function: 1)

6.3575
6.1150
0.4664 2.6254i
0.4947 0.7799i
1.2789
0.5465 0.0984i

0.1749

ISE
ITSE
ITAE
IAE

43.8016  104
83.1849  104
58.2969  102
21.875  102

40
(a) - Proposed J3 : DE 2DOF PID

35

(b) - J1 : DE 2DOF PID


(c) - J2 : DE 2DOF PID

Settling Time (Sec)

30

25

(d) - Obj. function 1 : CPSO PI [12]

(e) - J1 : CPSO PI [12]


(f) - J2 : CPSO PI [12]

20

15

d
10
5

b c

d
b

f2

f1

Ptie

Fig. 4. Comparison of settling times.

The system dynamic responses with controller parameters obtained using ISE and ITSE objective functions are shown in Figs.
68. For comparison, the simulation results with CPSO optimized

PI controller using the same ISE and ITSE objective functions


are also shown in Figs. 68. It is clear from Figs. 68 that the
performance of the system is greatly improved by the proposed

25

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

x 10-3

(a) - Proposed J3 : DE 2DOF PID

Peak Overshoot

(b) - J1 : DE 2DOF PID


7

(c) - J2 : DE 2DOF PID

(d) - J1 : CPSO PI [12]


(f) - J2 : CPSO PI [12]

e
d

c
3
2

(e) - Obj. function 1 : CPSO PI [12]

f
d

a
0

f1

f2

Ptie

Fig. 5. Comparison of peak overshoots.

0.005
0

f1 (H z)

-0.005
-0.01
-0.015

J1 - DE : 2DOF PID

-0.02

J2 - DE : 2DOF PID

-0.025

J1 - CPSO : PI [12]

-0.03

J2 - CPSO : PI [12]
0

10

12

14

16

18

20

Time (S)
Fig. 6. Frequency deviation of area-1 for 1% change in area-1 with ISE and ITSE objective function.

0.005
0

f2 (H z)

-0.005
-0.01
-0.015

J1 - DE : 2DOF PID

-0.02

J2 - DE : 2DOF PID

-0.025

J1 - CPSO : PI [12]

-0.03

J2 - CPSO : PI [12]
0

10

12

14

16

18

20

Time (S)
Fig. 7. Frequency deviation of area-2 for 1% change in area-1 with ISE and ITSE objective function.

controller compared to the recently reported results [12]. The system responses with controller parameters obtained using proposed
objective function J3 are shown in Figs. 911. Here the results are
compared with claimed improved results of recently reported

results [12], where modied objective functions are employed to


optimize the controller parameters. Critical analysis of the dynamic responses clearly reveals that dynamic performance of the
system is signicantly improved with proposed controller. Fig. 12

26

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

x 10

-3

Ptie (P.U.)

-1
-2
-3
-4
J1 - DE : 2DOF PID

-5

J2 - DE : 2DOF PID

-6

J1 - CPSO : PI [12]

-7
-8

J2 - CPSO : PI [12]
0

10

12

14

16

18

20

Time (S)
Fig. 8. Tie line power deviation for 1% change in area-1 with ISE and ITSE objective function.

0.005
0

f1 (H z)

-0.005
-0.01
-0.015
-0.02
Proposed J3 : DE 2DOF PID

-0.025

Obj. function 1 : CPSO PI [12]


Obj. function 2 : CPSO PI [12]

-0.03
-0.035
0

10

12

14

16

18

20

Time (S)
Fig. 9. Frequency deviation of area-1 for 1% change in area-1 with proposed objective function.

0.005
0
-0.005

f2 (H z)

-0.01
-0.015
-0.02
-0.025
Proposed J3 : DE 2DOF PID

-0.03

Obj. function 1 : CPSO PI [12]


Obj. function 2 : CPSO PI [12]

-0.035
0

10

12

14

16

18

20

Time (S)
Fig. 10. Frequency deviation of area-2 for 1% change in area-1 with proposed objective function.

shows the comparison of control efforts u1 and u2 versus time for


the controller parameters obtained using proposed objective function J3 and best claimed controller parameters of CPSO [12] approach. It is evident from Fig. 12 that proposed approach
requires less control efforts compared to the recently reported
CPSO [12] approach. As 1% step load disturbance is applied in
area-1, the reference power setting of that area, i.e. u1 settles at

1% (0.01 p.u.) in Fig. 12. The performance of the proposed controller is further investigated for simultaneous load disturbance at
both areas. A simultaneous step increase in demand of 1% in
area-1 and 3% in area-2 is considered at t = 0.0 s and the system responses are shown in Figs. 1315. It is evident from Figs. 1315
that the designed controllers are robust and perform satisfactorily
when the location of the disturbance changes.

27

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933
-3

x 10
0

Ptie (P.U.)

-2
-4
-6
Proposed J3 : DE 2DOF PID

-8
-10

Obj. function 1 : CPSO PI [12]


Obj. function 2 : CPSO PI [12]
0

10

12

14

16

18

20

Time (S)
Fig. 11. Tie line power deviation for 1% change in area-1 with proposed objective function.

u1 : Obj. function 1 CPSO : PI

0.03

u1 : J3 - DE : 2DOF PID
u2 : Obj. function 1 CPSO : PI

u1, u2

0.02

u2 : J3 - DE : 2DOF PID

0.01

-0.01
0

10

12

14

16

18

20

Time (S)
Fig. 12. Comparison of controll efforts for 1% step increase in demand in area-1.

f1 (H z)

-0.02
-0.04
-0.06
-0.08
-0.1

Proposed J3 : DE 2DOF PID


Obj. function 1 : CPSO PI [12]
Obj. function 2 : CPSO PI [12]

-0.12
-0.14

10

12

14

16

18

20

Time (S)
Fig. 13. Frequency deviation of area-1 for 1% change in area-1 and 3% change in area-2 with proposed objective function.

Finally, sensitivity analysis is done to study the robustness the


system to wide changes in the operating conditions and system
parameters [10,13,22,23]. Taking one at a time, the operating load
condition and time constants of speed governor, turbine, tie-line
power are changed from their nominal values in the range of
+50% to 50% in steps of 25%. The proposed objective function J3
is used due to its superior performance in all the cases for analysis.
The various performance indexes, settling times, overshoots and

minimum damping ratios with the above varied system conditions


are given in Table 3. The system modes under these parameter variation conditions are provided in Table 4. It is clear from Table 3
that the system performances hardly change when the operating
load condition and system parameters are changed by 50% from
their nominal values. It is also clear from Table 4 that for all the
cases the system eigen values lie in the left half of s-plane. Hence
it can be concluded that the closed loop system is stable when

28

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

0.02
0

f2 (H z)

-0.02
-0.04
-0.06
-0.08
Proposed J3 : DE 2DOF PID

-0.1

Obj. function 1 : CPSO PI [12]


Obj. function 2 : CPSO PI [12]

-0.12
0

10

12

14

16

18

20

Time (S)
Fig. 14. Frequency deviation of area-2 for 1% change in area-1 and 3% change in area-2 with proposed objective function.

x 10

-3

20
Proposed J3 : DE 2DOF PID
Obj. function 1 : CPSO PI [12]
Obj. function 2 : CPSO PI [12]

Ptie (P.U.)

15

10

0
0

10

12

14

16

18

20

Time (S)
Fig. 15. Tie line power deviation for 1% change in area-1 and 3% change in area-2 with proposed objective function.

Table 3
Sensitivity analysis.
Parameter variation

% Change

Performance index
ITAE  10

2

ISE  10

4

4

2

Settling time Ts (sec)

Peak overshoot (OS)

ITSE  10

IAE  10

Df1

Df2

Df1

Df2

Loading condition

+50
+25
25
50

5.9697
5.9712
5.9742
5.9757

1.2121
1.2174
1.2284
1.2339

0.8121
0.8146
0.8199
0.8227

2.6739
2.6762
2.6807
2.6829

5.20
5.20
5.19
5.19

5.50
5.51
5.55
5.57

8.69
8.68
8.68
8.69

0.0014
0.0014
0.0015
0.0015

0.0001
0.0002
0.0002
0.0002

0.1914
0.1905
0.1887
0.1877

TG

+50
+25
25
50

6.2171
6.0215
6.0148
6.1150

1.6497
1.4152
1.0680
0.9468

1.1068
0.9275
0.7569
0.7325

2.9712
2.7842
2.6360
2.6166

6.14
4.88
5.23
5.33

6.36
5.91
5.77
5.74

8.62
8.67
8.74
8.81

0.0031
0.0021
0.0010
0.0008

0.0009
0.0004
0.0001
0.0000

0.1515
0.1697
0.2131
0.2465

TT

+50
+25
25
50

6.0430
5.9379
6.0389
6.2023

1.7239
1.4482
1.0426
0.9027

1.1304
0.9361
0.7535
0.7294

2.9593
2.7746
2.6189
2.6017

5.27
4.93
5.31
5.47

5.50
5.97
2.81
2.99

8.52
8.62
8.77
8.89

0.0029
0.0019
0.0011
0.0009

0.0007
0.0004
0.0001
0.0000

0.1880
0.1925
0.1787
0.1595

T12

+50
+25
25
50

4.8897
5.2908
7.1371
9.4416

1.2123
1.2141
1.2408
1.2716

0.8019
0.8037
0.8463
0.8980

2.5891
2.6132
2.7807
2.9415

5.12
5.12
5.47
7.14

3.66
5.11
6.49
7.41

7.94
8.23
9.47
10.86

0.0014
0.0014
0.0015
0.0014

0.0004
0.0003
0.0000
0.0000

0.1508
0.1697
0.2101
0.2311

the operating conditions and system parameters varies. The frequency deviation response for 1% change in area-1 with these varied conditions is shown in Figs. 1619, respectively. It can be
observed from Figs. 1619 that the effect of the variation of operating loading conditions and system time constants on the system

DPTie

responses is negligible. So it can be concluded that, the proposed


control strategy provides a robust and stable control and the controller parameters obtained at the nominal loading with nominal
parameters, need not be reset for wide changes in the system loading or system parameters.

29

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933
Table 4
System modes for each case.
Loading condition (%)

System modes

TG (%)

System modes

TT (%)

System modes

T12 (%)

System modes

+50

172.61
172.61
1.20 6.14i
1.8 0.05.90i
1.26 1.60i
0.93 1.12i
0.31

+50

171.54
171.54
0.77 5.05i
1.33 4.65i
0.31
1.37 1.49i
0.97 1.16i

+50

1.7155
1.7154
0.93 4.85i
1.58 4.43i
0.31
1.49 1.46i
0.99 1.20i

+50

1.7261
1.7260
0.96 6.29i
1.67 5.89i
1.46 1.73i
0.93 1.12i
0.36

+25

172.61
172.61
1.19 6.13i
1.67 5.90i
1.26 1.60i
0.93 1.12i
0.31

+25

171.97
171.96
0.95 5.50i
1.47 5.19i
1.32 1.55i
0.95 1.14i
0.31

+25

171.97
171.97
1.06 5.39i
1.63 05.07i
1.37 1.54i
0.31
0.95 1.16i

+25

172.61
172.61
1.07 6.21i
1.67 5.89i
1.36 1.67i
0.93 1.12i
0.34

25

172.61
172.61
1.18 6.12i
1.66 5.89i
1.26 1.61i
0.93 1.13i
0.30

25

173.69
173.68
1.54 7.05i
1.98 6.89i
1.21 1.65i
0.91 1.11i
0.30

25

173.67
173.67
1.31 7.20i
1.72 7.03i
1.16 1.65i
0.90 1.09i
0.30

25

172.61
172.61
1.30 6.06i
1.67 5.89i
1.16 1.52i
0.93 01.12i
0.26

50

172.61
172.61
1.17 6.12i
1.66 5.88i
1.26 1.61i
0.93 1.13i
0.30

50

175.86
175.85
2.17 8.55i
2.58 8.48i
1.15 1.69i
0.89 1.09i
0.30

50

175.80
175.79
1.79 8.85i
1.45 8.95i
1.08 1.68i
0.88 1.07i
0.30

50

172.61
172.61
1.42 5.99i
1.67 5.89i
0.21
1.07 14.1i
0.93 1.12i

-3

x 10

f1 (Hz)

-5

-10

+50% of nomonal load


+25% of nomonal load
-25% of nomonal load
-50% of nomonal load

-15
0

10

12

14

16

18

20

Time (S)
Fig. 16. Frequency deviation of area-1 for 1% change in area-1 with varying load conditions.

6.1. Extension to realistic power system


To get an accurate insight of the AGC topic, it is essential to include the important inherent requirement and the basic physical
constraints and include them model. The important constraints
which affect the power system performance are time delay, Generation Rate Constraint (GRC), and governor dead band nonlinearity
[24]. In view of the above, the study is further extended to a more
realistic power system by considering the effect of time delay, reheat turbine and GRC in addition to governor dead band nonlinearity. Owing to the growing complexity of power systems in
deregulated environment, communication delays become a signicant challenge in the AGC analysis. Time delays can degrade a systems performance and even cause system instability. In a power
system having steam plants, power generation can change only at
a specied maximum rate. A typical value of the GRC for thermal
unit is 3%/min [25]. As most of the thermal plants are of reheat type,
a reheat turbine is also considered in the proposed realistic power

system model. In the present study, a time delay of 2 s and a GRC


of 3%/min are considered [24]. It is observed that the system becomes unstable when time delay, reheat turbine and GRC are included in the system model with the controller parameters
obtained considering only governor dead band nonlinearity. Hence
the controller parameters are retuned in the presence of above
physical constraints. In this case, minimization of control efforts is
also included in the objective function in addition to the earlier proposed objective function (J3). The objective function is dened as:

J4

tsim

x1  jDf1 j jDf2 j jDPTie j  t  dt x2

min

1
Pn

i1 1

 fi

 x3  T S x4  OS

 ju1 j ju2 j  t  dt

t sim

x5

15

The weights x1x5 are so chosen that numerical value of all the
terms in the right hand side of Eq. (15) lie in the same range. Repet-

30

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933
-3

x 10

f1 (Hz)

-5
+50% of T12

-10

+25% of T12
-25% of T12
-50% of T12

-15
0

10

12

14

16

18

20

Time (S)
Fig. 17. Frequency deviation of area-1 for 1% change in area-1 with varying tie line time constant.

x 10

-3

f1 (Hz)

-5
+50% of TT

-10

+25% of TT
-25% of TT

-15

-50% of TT
0

10

12

14

16

18

20

Time (S)
Fig. 18. Frequency deviation of area-1 for 1% change in area-1 with varying turbine time constant.

-3

x 10

f1 (Hz)

-5
+50% of TG

-10

+25% of TG
-25% of TG

-15

-50% of TG
0

10

12

14

16

18

20

Time (S)
Fig. 19. Frequency deviation of area-1 for 1% change in area-1 with varying speed governor time constant.

itive trial runs of the optimizing algorithms are done to nd out the
range of each term of objective function and selection of appropriate weights. The weights are chosen as: x1 = 1.0, x2 = 0.2, x3 =
0.02, x4 = 150, x5 = 0.03. The optimum parameters are obtained
as explained in Section 5.1. The optimal parameters are:
KP = 0.08, KI = 0.0794, KD = 0.0724, N = 271.0989; PW = 2.9079,
DW = 1.1739
A step increase in demand of 1% is applied at t = 0 s in area-1
and the system dynamic responses is shown in Fig. 20. It is evident

from Fig 20 that the system is stable with retuned controller


parameters but the dynamics of the power system is affected with
increased over shoot, performance errors and settling times. Finally, sensitivity analysis is done to study the robustness the system
to wide changes in the operating conditions and system parameters as before. The various performance indexes (errors), settling
times, overshoots, minimum damping ratios and control efforts
under normal and parameter variation cases are given in Table 5.
It can be noticed from Table 5 that when physical constraints are

31

f1 (Hz) f2 (Hz) Ptie (P.U.)

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

-0.01

-0.02
f1

-0.03

f2
Ptie

-0.04
0

10

20

30

40

50

60

Time (S)
Fig. 20. System response for 1 % change in area-1 with physical constraints (time delay, reheat turbine, GRC and governor dead band).

Table 5
Sensitivity analysis with time delay, reheat turbine, GRC and governor dead band.
Parameter
variation

%
Change

Performance index
ITAE  102

ISE  104

ITSE  104

IAE  102

Settling time Ts (sec)

Peak Overshoot
(OS)

ITAE of control efforts


(u1 + u2)

Df1

Df2

DPTie

Df1

Df2

0
+50

205.53
203.69

47.730
45.627

112.40
106.85

30.402
29.769

26.18
26.28

20.31
20.47

43.70
43.69

0.0066
0.0058

0.0058
0.0049

0.0887
0.0936

63.0491
63.0531

+25
25
50

204.60
206.49
207.52

46.654
48.858
50.043

109.53
115.47
118.76

30.081
30.732
31.076

26.22
26.16
26.16

20.39
20.23
20.18

43.69
43.71
43.73

0.0062
0.0070
0.0074

0.0053
0.0063
0.0069

0.0911
0.0863
0.0838

63.0511
63.0471
63.0451

TG

+50
+25
25
50

213.56
208.24
204.00
203.23

54.559
50.908
44.974
42.600

133.39
121.48
105.52
100.38

32.491
31.278
29.746
29.263

26.32
26.29
26.32
26.24

20.22
20.04
20.30
20.59

43.54
43.65
43.79
43.86

0.0096
0.0079
0.0056
0.0048

0.0113
0.0085
0.0055
0.0052

0.0439
0.0632
0.1215
0.1602

63.0549
63.0519
63.0463
63.0437

TT

+50
+25
25
50

221.19
210.082
203.171
202.605

57.990
52.401
43.842
40.625

146.36
126.41
102.71
96.13

33.894
31.765
29.427
28.847

26.76
26.34
26.32
26.40

20.57
20.17
20.23
20.67

43.47
43.59
43.85
43.97

0.0108
0.0084
0.0054
0.0043

0.0132
0.0095
0.0053
0.0046

0.0411
0.0603
0.1320
0.1955

63.0582
63.0535
63.0448
63.0407

T12

+50
+25
25
50

Unstable
223.302
211.914
264.350

48.732
48.979
53.290

122.20
116.60
139.73

31.837
30.595
34.203

33.17
27.54
38.56

31.02
19.85
35.35

44.52
44.88
47.98

0.0106
0.0097
0.0126

0.0086
0.0070
0.0089

0.0631
0.1276
0.1944

63.0787
63.0037
62.9268

Normal
Loading
condition

0.005
0

f1 (Hz)

-0.005
-0.01
-0.015
-0.02

+50% of nomonal load

-0.025

+25% of nomonal load


-25% of nomonal load

-0.03

-50% of nomonal load


0

10

15

20

25

30

35

40

Time (S)
Fig. 21. Frequency deviation of area-1 for 1% change in area-1 with physical constraints (time delay, reheat turbine, GRC and governor dead band) and varying load
conditions.

32

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

0.01

f1 (Hz)

-0.01

+50% of TG

-0.02

+25% of TG
-25% of TG
-0.03

-50% of TG
0

10

15

20

25

30

35

40

Time (S)
Fig. 22. Frequency deviation of area-1 for 1% change in area-1 with physical constraints (time delay, reheat turbine, GRC and governor dead band) and varying speed governor
time constant.

0.01

f1 (Hz)

-0.01

+50% of TT

-0.02

+25% of TT
-25% of TT

-0.03

-50% of TT
0

10

15

20

25

30

35

40

Time (Hz)
Fig. 23. Frequency deviation of area-1 for 1% change in area-1 with physical constraints (time delay, reheat turbine, GRC and governor dead band) and varying turbine time
constant.

0.02
0.01

f1 (Hz)

-0.01

+50% of T12

-0.02

+25% of T12
-25% of T12

-0.03

-50% of T12
-0.04

10

15

20

25

30

35

40

Time (S)
Fig. 24. Frequency deviation of area-1 for 1% change in area-1 with physical constraints (time delay, reheat turbine, GRC and governor dead band) and varying tie line time
constant.

introduced, the variations in performance index are more prominent. So it can be concluded that in the presence of time delay,
GRC and reheat turbine, the system becomes highly non-linear

(even for small load perturbation) and hence the performance of


the designed controller is degraded. To complete the analysis, a
1% step load increase in area-1 at t = 0 s is applied and the

R.K. Sahu et al. / Electrical Power and Energy Systems 49 (2013) 1933

frequency deviation response of area-1 for the above varied conditions are shown in Figs. 2124. From Table 5 and Figs. 2124 it
can be seen that when nonlinearity is introduced, the effect of
change in loading conditions and governor time constant is minimum on system dynamic performance. It is also evident from Table
5 that the changes in performance index are more prominent to
change in T12 and system becomes unstable when T12 is increased
by +50%.
7. Conclusion
This study presents the design and performance analysis of parallel 2-Degree Freedom of Proportional-Integral-Derivative (2-DOF
PID) controller optimized employing Differential Evolution (DE)
algorithm for Load Frequency Control (LFC) of interconnected
power system. A widely used standard test system which is a
two area thermal system with governor dead-band nonlinearity
is considered for the design and analysis purpose. First it has been
demonstrated that objective function plays an important role on
the performance of the controller by comparing the results with
controller parameters obtained using conventional ISE and ITSE
objective functions. Further, a modied objective function is proposed which includes a conventional error criterion, damping ratio
of dominant eigenvalues, settling times of frequency and peak
overshoots with appropriate weight coefcients. The results obtained from the simulations demonstrate that the proposed control
strategy optimized with modied objective function achieves better dynamic performances than the standard objective functions.
The superiority of the proposed approach has been shown by
comparing the results with a recently published Craziness based
Particle Swarm Optimization (CPSO) technique for the same interconnected power system. Additionally, sensitivity analysis is performed by varying the system parameters and operating load
conditions from their nominal values. It is observed that the proposed controllers are quite robust for a wide range of the system
parameters and operating load conditions from their nominal values. Finally, the proposed approach is further extended to a more
realistic power system model by considering the physical constraints such as time delay, reheat turbine, GRC and governor dead
band and the objective function is further modied to include to
control efforts minimization. It is observed that when physical constraints are introduced, the variations in performance index are
prominent and the system becomes unstable when T12 increases
by +50%.
Acknowledgement
The authors wish to thank the reviewers and the editor for providing critical comments to improve the quality of the paper.
Appendix A
Nominal parameters of the system investigated are: PR = 2000
MW (rating), PL = 1000 MW (nominal loading); f = 60 Hz, B1, B2 =

33

0.425 p.u. MW/Hz; R1 = R2 = 2.4 Hz/p.u.; TG1 = TG2 = 0.2 s; TT1 = TT2 =
0.3 s; KPS1 = KPS2 = 120 Hz/p.u. MW; TPS1 = TPS2 = 20 s; T12 = 0.0707 p.u.;
a12 = 1, Kr1 = Kr2 = 0.5, Tr1 = Tr2 = 10.
References
[1] Kundur P. Power system stability and control. 8th ed. New Delhi: Tata
McGraw-Hill; 2009.
[2] Elgerd OI. Electric energy systems theory. An introduction. 2nd ed. New
Delhi: Tata McGraw-Hill; 2007.
[3] Ibraheem, Kumar P, Kothari DP. Recent philosophies of automatic generation
control strategies in power systems. IEEE Trans Power Syst 2005;20:34657.
[4] Shoults RR, Jativa Ibarra JA. Multi area adaptive LFC developed for a
comprehensive AGC simulation. IEEE Trans Power Syst 1993;8:5417.
[5] Demiroren A, Zeynelgil HL, Sengor NS. Application of ANN technique to load
frequency control for three area power system. In: IEEE power tech. conf.,
Porto; 2001.
[6] Chaturvedi DK, Satsangi PS, Kalra PK. Load frequency control: a generalized
neural network approach. Int J Elect Power Energy Syst 1999;21:40515.
[7] Talaq J, Al-Basri F. Adaptive fuzzy gain scheduling for load frequency control.
IEEE Trans Power Syst 1999;14:14550.
[8] Ahamed TPI, Rao PSN, Sastry PS. A reinforcement learning approach to
automatic generation control. Elect Power Syst Res 2002;63:926.
[9] Khuntia SR, Panda S. Simulation study for automatic generation control of a
multi-area power system by ANFIS approach. Appl Soft Comput 2012;12:
33341.
[10] Nanda J, Mishra S, Saikia LC. Maiden application of bacterial foraging based
optimization technique in multiarea automatic generation control. IEEE Trans
Power Syst 2009;24(2):6029.
[11] Ali ES, Abd-Elazim SM. Bacteria foraging optimization algorithm based load
frequency controller for interconnected power system. Int J Elect Power
Energy Syst 2011;33:6338.
[12] Gozde H, Taplamacioglu MC. Automatic generation control application with
craziness based particle swarm optimization in a thermal power system. Int J
Elect Power Energy Syst 2011;33:816.
[13] Saikia LC, Nanda J, Mishra S. Performance comparison of several classical
controllers in AGC for multi-area interconnected thermal system. Int J Elect
Power Energy Syst 2011;33:394411.
[14] Stron R, Price K. Differential evolution a simple and efcient adaptive scheme
for global optimization over continuous spaces. J Global Optim 1995;11:
34159.
[15] Das S, Suganthan PN. Differential evolution: a survey of the state-of-the-art.
IEEE Trans Evol Comput 2011;15:431.
[16] Sncheza J, Visioli A, Dormido S. A two-degree-of-freedom PI controller based
on events. J Process Control 2011;21:63951.
[17] Zhao YM, Xie WF, Tu XW. Performance-based parameter tuning method of
model-driven PID control systems. ISA Trans 2012;51:3939.
[18] Ogatta K. Modern control engineering. NJ (USA): Prentice Hall; 1990.
[19] Tan Wen. Unied tuning of PID load frequency controller for power systems
via IMC. IEEE Trans Power Syst 2010;25(1):34150.
[20] Panda S. Differential evolution algorithm for SSSC-based damping controller
design considering time delay. J Franklin Inst 2011;348(8):190326.
[21] Panda S. Robust coordinated design of multiple and multi-type damping
controller using differential evolution algorithm. Int J Elect Power Energy Syst
2011;33:101830.
[22] Gozde H et al. Comparative performance analysis of Articial Bee Colony
algorithm in automatic generation control for interconnected reheat thermal
power system. Int J Elect Power Energy Systs 2012;42:16778.
[23] Parmar KPS et al. Load frequency control of a realistic power system with
multi-source power generation. Int J Elect Power Energy Syst 2012;42:42633.
[24] Golpra H, Bevrani H, Golpra H. Application of GA optimization for automatic
generation control design in an interconnected power system. Energy Convers
Manage 2011;52:224755.
[25] Sudha KR, Raju YB, Sekhar AC. Fuzzy C-means clustering for robust
decentralized load frequency control of interconnected power system with
generation rate constraint. Int J Elect Power Energy Syst 2012;37:5866.

Potrebbero piacerti anche