Vous êtes sur la page 1sur 8

602 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 24, NO.

2, MAY 2009

Maiden Application of Bacterial Foraging-Based


Optimization Technique in Multiarea
Automatic Generation Control
Janardan Nanda, Fellow, IEEE, S. Mishra, Senior Member, IEEE, and Lalit Chandra Saikia

Abstract—A maiden attempt is made to examine and highlight Steam turbine reheat constant of area .
the effective application of bacterial foraging (BF) to optimize sev-
eral important parameters in automatic generation control (AGC) Steam turbine reheat time constant of area (s).
of interconnected three unequal area thermal systems, such as in- Steam turbine time constant of area (s).
tegral controller gains (KIi ) for the secondary control, governor
speed regulation parameters (Ri ) for the primary control and fre- Frequency bias of area .
quency bias parameters (Bi ), and compare its performance to es- Nominal system frequency (Hz).
tablish its superiority over genetic algorithm (GA) and classical
methods. Comparison of convergence characteristics of BF, GA, .
and classical approach reveals that the BF algorithm is quite faster (Hz/p.u.).
in optimization, leading to reduction in computational burden and
giving rise to minimal computer resource utilization. Simultaneous Gain of integral controller in area .
optimization of KIi , Ri , and Bi parameters which surprisingly has (i.e., frequency response
never been attempted in the past, provides not only best dynamic
response for the system but also allows use of much higher values of
characteristics of area ).
Ri (than used in practice), that will appeal to the power industries
for easier and cheaper realization of governor. Sensitivity analysis .
is carried out which demonstrates the robustness of the optimized
KIi , Ri , and Bi to wide changes in inertia constant (H), reheat time Simulation time (s).
constant (Tr ), reheat coefficient (Kr ), system loading condition,
and size and position of step load perturbation. Incremental change in frequency of area (Hz).
Index Terms—Automatic generation control, bacterial foraging Incremental generation of area (p.u.).
technique, genetic algorithm, sensitivity analysis, speed regulation Incremental change in tie power of tie 1–2
parameter. (p.u.).
Incremental change in tie power of tie 1–3
NOMENCLATURE: (p.u.).
, , Synchronizing coefficients.
Subscript referred to area (1, 2, 3).
Superscript denotes optimum value. Number of parameters to be optimized.
Rated power of area (MW). Number of bacteria.
Inertia constant of area (s). Swimming length after which tumbling of
Incremental load change in area (p.u.). bacteria will be undertaken in a chemotactic
loop.
(p.u./Hz).
Number of iterations to be undertaken in a
Incremental generation change in area (p.u.). chemotactic loop .
Governor speed regulation parameter of area Maximum number of reproduction to be
. (Hz/p.u.MW). undertaken.
Steam governor time constant of area (s). Maximum number of elimination and dispersal
events.
Manuscript received May 30, 2008; revised December 10, 2008. First
Probability with which elimination and
published April 14, 2009; current version published April 22, 2009. Paper no. dispersal will continue.
TPWRS-00421-2008.
J. Nanda and S. Mishra are with the Department of Electrical Engineering,
Indian Institute of Technology Delhi, New Delhi, India (e-mail: janardan- I. INTRODUCTION
nanda@yahoo.co.in; sukumar@ee.iitd.ac.in).
L. C. Saikia is with the National Institute of Technology Silchar, Assam, India
(e-mail: lcsaikia@yahoo.com).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
M ODERN power system consists of number of utilities
interconnected together and power is exchanged be-
tween utilities over tie-lines by which they are connected. In
Digital Object Identifier 10.1109/TPWRS.2009.2016588 order to achieve interconnected operation of a power system,
0885-8950/$25.00 © 2009 IEEE
NANDA et al.: MAIDEN APPLICATION OF BACTERIAL FORAGING-BASED OPTIMIZATION TECHNIQUE 603

an electric energy system must be maintained at a desired oper- ters and frequency bias parameters simultaneously for an AGC
ating level characterized by nominal frequency, voltage profile system in order to explore and examine their optimum values
and load flow configuration. This is achieved by close control for each area and their specific impact on the overall dynamics
of real and reactive powers generated through the controllable of the system as compared to when only secondary control gains
source of the system. Automatic generation control (AGC) are optimized, considering values of governor speed regulation
plays a significant role in the power system by maintaining parameters and frequency bias parameters same as
scheduled system frequency and tie line flow during normal normally used in practice. Simultaneous optimization of param-
operating condition and also during small perturbations. Many eters may also throw new findings for governor operation and
investigations in the area of isolated and interconnected power design.
systems have been reported in the past. Almost all such past GA can effectively explore many region of search space
works pertain to the design of governor secondary controllers simultaneously rather than a single region. Hence, GA is less
or supplementary controllers and surprisingly little attention sensitive to local minimum as compared to the conventional
has been paid to the design of governor primary control, i.e., approach. GA manipulates the representation of potential
selection of proper governor droop or governor speed reg- solution, rather than the solutions itself. To overcome the pos-
ulation parameter (R) in Hz/p.u.MW. It is known that with sibility of being trapped into local minima only two operations
only primary control (i.e., secondary or supplementary control crossover and mutation are performed. Recent research has
absent) the smaller the governor droop the smaller the steady identified some of the deficiencies in GA performance [13].
state error in frequency but in the presence of supplementary The premature convergence of GA degrades its efficiency and
control there is nothing to be sacrosanct to use a small governor reduces the search capability. To overcome this problem a
droop (of the order of 4% to 6% used in practice) as any large more recent and powerful computational intelligence technique
but credible value of R can also guarantee zero steady state bacterial foraging (BF) is available in which the number of
error in frequency. However, what is the best value of R and parameters that are used for searching the total solution space is
its procedure for selection remains unexplored. A few works much higher compared to those in GA [14], [15] and hence the
[5], [9], [19] have been reported to an extent for selection of possibility of overriding local minimum in BF is much higher
governor speed regulation parameter R but no comprehensive then in GA. BF technique meanwhile has been successfully
approach or optimization procedure has been provided for se- applied in some of the areas of electrical engineering [18],
lection of suitable value of R. Further, no valuable information [20] where they have shown the superiority of BF over GA.
is available regarding selection of suitable value of frequency However, it has not been applied to AGC problem yet.
bias setting (B) in the governor supplementary control loop, In view of the above, the present work dwells in the applica-
except that the value B should not be less than the area fre- tion of BF technique for simultaneous optimization of several
quency response characteristics parameter [3], [4]. Thus parameters such as integral controller gains for secondary
for multiarea AGC the problem of selection of optimum value control, governor speed regulation parameters for primary
of R and B for each area remained unsolved. control loops and frequency bias parameters for AGC of a
Several approaches for control and optimization such as clas- three unequal area thermal power system, provides critical anal-
sical, optimal, genetic algorithm, fuzzy logic, particle swarm ysis of the dynamic responses and compares them with those ob-
optimization, artificial neural network, etc. for the design of sup- tained with simultaneous optimization of only (keeping
plementary controller have been reported in literature [1]–[20]. and at values used in practice) or simultaneous optimization
For any optimization technique both the convergence and op- of and only (keeping at values used in practice) so as
timal value achieved are important. Classical integral or propor- to explore what valuable findings are lost when all the parame-
tional integral controllers considering integral square error (ISE) ters ( , , and ) are not optimized simultaneously. Further,
criterion for optimization of their gains are quite in vogue. The the performance of BF technique is duly compared to the per-
optimization of gains for such classical controller, however, is formance of GA and classical techniques to highlight its merits.
quite involved and time consuming when traditional approach The robustness of the optimum , , and parameters ob-
is used and the result may be far from optimal. tained at the nominal operating condition is examined through
In the traditional approach sequential optimization is used sensitivity analysis by changing the loading pattern and several
where one parameter is optimized at a time using ISE criterion system parameters by a wide margin from their nominal values
keeping the other parameters fixed and then repeating this op- and also for changes in size and position of step load perturba-
eration for every other parameter in turn to complete one itera- tion (SLP).
tion of optimization. Several iterations are made using the said
procedure so as to determine the overall minimum of the per- II. SYSTEM INVESTIGATED
formance index for deciding the optimum parameters. When Investigations have been carried out on three unequal area
the number of parameters to be optimized is large, classical (area 1: 2000 MW, area 2: 4000 MW, and area 3: 8000 MW)
technique for optimization is certainly not the preferred one. systems. The systems are provided with single reheat tur-
Some authors [16], [17] have used genetic algorithm (GA) to bines, integral controllers and have realistic generation rate
optimize controller gains of a multiarea AGC system simulta- constraints (GRC) of 3% per minute. The nominal parame-
neously more effectively than is possible with traditional ap- ters of the systems are taken from [8] and presented in the
proach. Surprisingly, till date no work has been reported to op- Appendix . MATLAB version 7.01 has been used to obtain
timize secondary control parameters, primary control parame- dynamic responses. Per unit values of different parameters of
604 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 24, NO. 2, MAY 2009

Fig. 1. Transfer function model of a three-area thermal system considering generation rate constraints.

the unequal areas are considered to be same on their respective direction, say , is generated; this will be used to define the
MW capacity bases. Hence, while modeling interconnected direction of movement after a tumble. In particular
areas of unequal capacities, the quantities ,
, and are considered. The
idea presented by Elgerd and Fosha [3] has been used for mod- (1)
eling the system. The transfer function model of a three-area
system is shown in Fig. 1. where represents the th bacterium at th chemotactic,
th reproductive and th elimination and dispersal step. is
III. BACTERIAL FORAGING OPTIMIZATION TECHNIQUE the size of the step taken in the random direction specified by
the tumble.
A recent evolutionary computation technique, called BF
scheme has been proposed by Passino [14], [15] in which the
number of parameters that are used for searching the total B. Swarming
solution space is much higher compared to those of GA. To During the process of reaching towards the best food loca-
overcome the possibility of being trapped in local minima, in tion it is always desired that the bacterium which has searched
GA only two operations (crossover and mutation) between the the optimum path should try to provide an attraction signal to
chromosomes are adopted. In BF, the foraging (methods of other bacteria so that they swarm together to reach the desired
locating, handling, ingesting food) behavior of E.coli bacteria location. In this process, the bacteria congregate into groups and
present in our intestine is mimicked. The control system of hence move as concentric patterns of groups with high bacterial
these bacteria that dictates how foraging should proceed can be density. The mathematical representation for swarming can be
subdivided into four sections namely Chemotaxis, Swarming, represented by
Reproduction, and Elimination and Dispersal. These operations
among the bacteria are used for searching the total solution
space. Thus the possibility of avoiding local minimum and
achieving faster convergence is higher than GA. The brief
descriptions of these operations are as follows.

A. Chemotaxis
This process is achieved through swimming and tumbling via
Flagella. Depending upon the rotation of Flagella in each bac-
terium, it decides whether it should move in a predefined direc-
tion (swimming) or altogether in different directions (tumbling),
in the entire lifetime. To represent a tumble, a unit length random (2)
NANDA et al.: MAIDEN APPLICATION OF BACTERIAL FORAGING-BASED OPTIMIZATION TECHNIQUE 605

where is the cost function value to be added h) The location of each bacterium P (1-p, 1-S, 1) which is
to the actual cost function to be minimized to present a time specified by random numbers within .
varying cost function. “S” is the total number of bacteria and i) The value of C(i), which is assumed to be constant in our
“p” the number of parameters to be optimized which are present case for all bacteria to simplify the design strategy.
in each bacterium. are j) The value of , , and . It
different coefficients that are to be chosen properly. is to be noted here that the value of and
must be same so that the penalty imposed on the cost func-
C. Reproduction tion through “ ” of (2) will be “0” when all the bacteria
The least healthy bacteria die and the other healthiest bacteria will have same value, i.e., they have converged.
each split into two bacteria, which are placed in the same loca- After initialization of all the above variables, keeping one
tion. This makes the population of bacteria constant. variable changing and others fixed the value of “J” proposed
in (3) is obtained by running the simulation of Fig. 1 using
D. Elimination and Dispersal the parameters contained in each bacterium. Corresponding to
It is possible that in the local environment the life of a popula- the minimum cost, the magnitude of the changing variable is
tion of bacteria changes either gradually (e.g., via consumption selected. Similar procedure is carried out for other variables
of nutrients) or suddenly due to some other influence. Events keeping the already optimized one unchanged. In this way all
can occur such that all the bacteria in a region are killed or a the variables of Step I-Initialization are obtained and are pre-
group is dispersed into a new part of the environment. They have sented below.
the effect of possibly destroying the chemotactic progress, but , , , , ,
they also have the effect of assisting in chemotaxis, since dis- , , , , and
persal may place bacteria near good food sources. From a broad . The value of p is equal to the number of param-
perspective, elimination and dispersal are parts of the popula- eters to be optimized.
tion-level long-distance motile behavior. 2) Step-2 Iterative Algorithm for Optimization: This section
models the bacterial population chemotaxis, swarming, repro-
duction, elimination, and dispersal (initially, ).
IV. BACTERIAL FORAGING ALGORITHM
For the algorithm updating automatically results in updating
In case of BF technique we assign each bacterium with a set of of “P”.
variables to be optimized and are assigned with random values 1) Elimination-dispersal loop:
within the universe of discourse defined through upper and 2) Reproduction loop:
lower limit between which the optimum value is likely to fall. In 3) Chemotaxis loop:
our case they are , , and . Each bacterium a) For , calculate cost for each bacterium
is allowed to take all possible values within the range and the as follows.
objective function which is ISE defined by the following is min- • Compute value of cost .
imized: Let
[i.e., add on the
cell-to-cell attractant effect obtained through (2)
(3)
for swarming behavior to the cost value obtained
through (3)].
where is the simulation time. In this paper, the BF algorithm • Let to save this value since
reported in [20] having better convergence characteristics is im- we may find a better cost via a run.
plemented as follows. • End of For loop.
1) Step 1-Initialization: b) For take the tumbling/swimming de-
a) Number of parameters (p) to be optimized. In this paper it cision
is either or ( and ) or ( , , and ) depending • Tumble: Generate a random vector
on the case. with each element , a random
b) Number of bacteria (S) to be used for searching the total number on .
region. It is equivalent to number of chromosomes in GA. • Move: let
c) Swimming length after which tumbling of bacteria will
be undertaken in a chemotactic loop.
(4)
d) , the number of iteration to be undertaken in a chemo-
tactic loop .
e) , the maximum number of reproduction to be under- Fixed step size in the direction of tumble for bac-
taken. terium is considered.
f) , the maximum number of elimination and dispersal • Compute and then let
events to be imposed over bacteria.
g) , the probability with which the elimination and dis-
persal will continue. This step is equivalent to mutation in
GA. (5)
606 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 24, NO. 2, MAY 2009

Fig. 2. Flow chart for bacterial foraging algorithm.

• Swim:
i) Let ; (counter for swim length).
ii) While (have not climbed down too
long)
• Let .
• If (if doing
better), let and
let
Fig. 3. Comparison of responses of Case A ( , K R = 24 B = : ; =
0 425
: K R B =
), Case B ( , , ) and Case C (K R B )
; ; . (a) Frequency
deviation in area 1 versus time. (b) Frequency deviation in area 2 versus time.
(6) (c) Frequency deviation in area 3 versus time. (d) Deviation in the tie line power
connecting area 1 and area 2 versus time. (e) Deviation in the tie line power con-
necting area 1 and area 3 vs. time, (f) Deviation in the tie line power connecting
areas 2 and 3 versus time.
and use this to compute the
new
• Else, let . This is the end of the
while statement.
c) Go to next bacterium if (i.e., go to b) to
process the next bacterium.
4) If , go to step 3. In this case, continue chemotaxis
since the life of the bacteria is not over.
5) Reproduction
a) For the given and , and for each ,
let be the
health of the bacterium (a measure of how many
nutrients it got over its life time and how successful
it was at avoiding noxious substance). Sort bacteria
in order of ascending cost (higher cost means
lower health). Fig. 4. Convergence characteristics of classical, GA, and BF algorithm.
b) The bacteria with highest values die
and other bacteria with the best value split (and the
copies that are made are placed at the same location The flow chart of the bacterial foraging algorithm is shown in
as their parent). Fig. 2.
6) If , go to 2; in this case, we have not reached the
number of specified reproduction steps, so we start the next
V. RESULT AND ANALYSIS
generation in the chemotactic loop.
7) Elimination-dispersal: For , with probability The three unequal area system described in Section II is con-
, eliminates and disperses each bacterium (this keeps sidered for the studies. Dynamic responses are obtained consid-
the number of bacteria in the population constant) to a ering 1% SLP in area 1. The cost J obtained using (3) is given
random location on the optimization domain. to the BF technique. Three cases are examined.
NANDA et al.: MAIDEN APPLICATION OF BACTERIAL FORAGING-BASED OPTIMIZATION TECHNIQUE 607

Fig. 5. (a) Comparison of frequency deviation in area 1 as a function of time for Fig. 5 (continued) (d) Comparison of frequency deviation n area 1 as a function
K R B
70% loading with , , corresponding to 70% loading and 50% loading. of time for 3% SLP in area 1 withK R B, , corresponding to 1% SLP in area
(b) Comparison of frequency deviation in area 1 as a function of time for 30% 1 and 3% SLP in area 1. (e) Comparison of frequency deviation in area 1 as a
K R B
loading with , , corresponding to 30% loading and 50% loading. (c) K R B
function of time for 1% SLP in area 2 with , , corresponding to 1%
Comparison of deviation in tie line power connecting areas 1 and 2 as a function SLP in area 1 and 1% SLP in area 2. (f) Comparison of frequency deviation in
K = 0 625 K R B
of time for : with , , corresponding to nominal K =05 : K R B
area 1 as a function of time for 1% SLP in area 3 with , , corresponding
and K = 0 625: . to 1% SLP in area 1 and 1% SLP in area 3.

A) The three integral gains for the three B) Six variables (parameters) consisting of
areas are first optimized simultaneously by BF technique, and are simultaneously optimized by BF
keeping their speed regulation parameters and keeping for all of the three areas. The six op-
frequency bias settings as used in practice. The timum value of variables (parameters) obtained are
three optimum values of integral gains found are , , ,
, , and . Dynamic , , and .
responses are shown in Fig. 3(a)-(f). Dynamic responses are given in Fig. 3(a)-(f).
608 IEEE TRANSACTIONS ON POWER SYSTEMS, VOL. 24, NO. 2, MAY 2009

TABLE I
OPTIMUM VALUES OF VARIABLES AT DIFFERENT SYSTEM CONDITIONS AND SYSTEM PARAMETERS

C) Nine variables (parameters) , and area 3 needs nearly equal to 1.7 times of . BF tech-
, and for the three areas are simul- nique helped significantly in handling efficiently simultaneous
taneously optimized by BF technique satisfying the con- optimization of several parameters and provided valuable infor-
straint . To evaluate the merits of BF technique the mation to suitable selection of governor droops and frequency
optimization of , and is also tried both through bias settings for different areas in a multiarea system for AGC.
the classical and GA techniques. In case of GA a popu- Sensitivity analysis is carried out to study the robustness of
lation size of 30 with 900 number of generation having the optimum , , and obtained at nominal conditions to
mutation probability 0.07 and crossover probability 0.83 wide changes in the system loading condition by from
is considered. The convergence characteristics of BF, GA, its nominal loading of 50%, wide changes in system parame-
and classical techniques are depicted in Fig. 4. From Fig. 4 ters such as H, , and by from their nominal values
it is quite clear that from among the three techniques the (given in the Appendix), and change in position and size of SLP.
convergence of BF is the fastest though the final value of The optimum value of , , and at changed loading con-
cost function is almost same. Here we have considered ditions, changed system parameters and changed positions and
“number of J evaluation” as the abscissa to obtain a plat- sizes of SLP are provided in Table I. Dynamic responses are
form for comparison of different techniques. The “number depicted in Fig. 5 for each changed condition with their corre-
of J evaluation” will be nearly equal to the scaled value sponding optimum values of , , and and compared to
of computational time since for our problem the calcula- the responses corresponding to optimized , , and at the
tion of “J” takes almost all the computational time. Hence, nominal condition. Critical examination of all frequency and tie
a faster converging algorithm will reduce computational responses clearly reveals that responses are more or less same.
burden. Only six sample responses are provided in Fig. 5 to justify the
The nine optimum values of variables (parame- said statement. Thus the optimum values of , , and ob-
ters) obtained through BF are , tained at the nominal parameters and nominal loading of 50%
, , , need not be reset for wide changes in the system loading or
, , system parameters or size and location of SLP.
,
and ). The dynamic re-
VI. CONCLUSION
sponses are given in Fig. 3(a)-(f).
Critical analysis of the dynamic responses clearly reveals that An attempt has been made for the first time to apply a pow-
for cases A and B the responses are practically same. However, it erful computational intelligence technique like BF to simultane-
is observed that when and are simultaneously optimized ously optimize effectively several important parameters such as
we get higher values of (even to the tune of 15% in area 1) , , and for AGC of a three unequal area thermal system
which is up to four times higher than 4% normally used in prac- with reheat turbines and generation rate constraint. Results re-
tice. High values of speed regulation parameters are preferred veal that BF technique performs much better then GA and clas-
as they are supposed to facilitate simpler and cheaper design of sical techniques.
governors [5]. Simultaneous optimization of , , and provides not
Comparing the responses for cases A, B, and C it is evident only the best dynamic response for the system but also reveals
that the responses for case C are the best from the view point new knowledge that different areas can have different optimum
of peak deviation, settling time and magnitude of oscillations. values of R and several areas may have much higher values of
It is clear that simultaneous optimization of along with R, with some area even having a value close to four times the
and has proved valuable in providing the best dynamic re- value of 4% used in practice. Such high values of R revealed
sponse, thereby highlighting the significance for optimum se- through simultaneous optimization are recommended for adop-
lection of for different areas rather than the usual practice tion in practice for easy and cheaper realization of governors.
of keeping for each area. For the specific three un- Further, for best dynamic response, it is to be appreciated that
equal area system studied area 1 needs nearly equal to 3.5 different areas may have different optimum frequency bias pa-
times of , area 2 needs nearly equal to 1.6 times of rameters B (observing always ) and some area may even
NANDA et al.: MAIDEN APPLICATION OF BACTERIAL FORAGING-BASED OPTIMIZATION TECHNIQUE 609

have value of B to the tune of three times of rather than usual [14] K. M. Passino, “Biomimicry of bacterial foraging for distributed op-
selection of for all areas. timization and control,” IEEE Control Syst. Mag., vol. 22, no. 3, pp.
52–67, Jun. 2002.
Sensitivity analysis reveals that the optimum values of , [15] Y. Liu and K. M. Passino, “Biominicry of social foraging for dis-
, and at the nominal loading condition are quite robust and tributed optimization: Models, principles and emergent behaviors,” J.
need not be reset for wide changes in system loading condition Optim. Theory Appl., vol. 115, no. 3, pp. 603–623, Dec. 2002.
[16] S. P. Ghoshal and S. K. Goswami, “Application of GA based optimal
or in system parameters such as , , and H from their nom- integral gains in fuzzy based active power-frequency control of non-
inal values or to changes in the size and position of step load reheat and reheat thermal generating systems,” Elect. Power Syst. Res.,
perturbation. vol. 67, pp. 79–88, 2003.
[17] S. P. Ghoshal, “Application of GA/GA-SA based fuzzy automatic
generation control of a multi-area thermal generating system,” Elect.
APPENDIX Power Syst. Res., vol. 70, pp. 115–127, 2004.
[18] S. Mishra, “A hybrid least square-fuzzy bacterial foraging strategy for
Nominal system parameters of the multiarea thermal system harmonic estimation,” IEEE Trans. Evol. Comput., vol. 9, no. 1, pp.
investigated: 61–73, Feb. 2005.
[19] J. Nanda, A. Mangla, and S. Suri, “Some new findings on automatic
, ; generation control of an interconnected hydrothermal system with con-
; ; ventional controllers,” IEEE Trans. Energy Convers., vol. 21, no. 1, pp.
; ; ; 187–194, Mar. 2006.
[20] S. Mishra and C. N. Bhende, “Bacterial foraging technique-based op-
; timized active power filter for load compensation,” IEEE Trans. Power
; ; Del., vol. 22, no. 1, pp. 457–465, Jan. 2007.
; ; ;
; . Janardan Nanda (F’06) has been a Professor in
the Department of Electrical Engineering, Indian
Institute of Technology Delhi, New Delhi, India,
REFERENCES since 1973. He served as Head of the Electrical
Engineering Department, Dean (UGS), and member,
[1] C. Concordia and L. K. Kirchmayer, “Tie-line power frequency control Board of Governors at Indian Institute of Tech-
of electric power system,” AIEE Trans, III-A, vol. 72, pp. 562–572, Jun. nology, New Delhi, during 1984–90. His field of
1953. interest comprises power system analysis, dynamics,
[2] C. Concordia and L. K. Kirchmayer, “Tie-line power frequency con- control, optimization, and application of computa-
trol of electric power system: Part II,” AISE Trans, III-A, vol. 73, pp. tional intelligence to power system problems.
133–146, Apr. 1954. Prof. Nanda is a fellow of the Indian National Sci-
[3] O. I. Elgerd and C. E. Fosha, “Optimum megawatt-frequency control ence Academy, a fellow of the Indian National Academy of Engineering, and a
of multiarea electric energy systems,” IEEE Trans. Power App. Syst., fellow of the Third World Academy of Sciences. He has served as chairman or
vol. PAS-89, no. 4, pp. 556–563, Apr. 1970. member of many national expert committees and has many national awards and
[4] C. E. Fosha and O. I. Elgerd , “The megawatt-frequency control lifetime achievement awards to his credit.
problem-A new approach via optimal control theory,” IEEE Trans.
Power App. Syst., vol. PAS-89, no. 4, pp. 563–577, Apr. 1970.
[5] J. Nanda and B. L. Kaul, “Automatic generation control of an inter-
connected power system,” Proc. Inst. Elect. Eng., vol. 125, no. 5, pp.
385–390, May 1978. S. Mishra (M’97–SM’04) received the B.E. degree
[6] M. L. Kothari, P. S. Satsangi, and J. Nanda, “Sampled data automatic from University College of Engineering, Burla,
generation control of interconnected reheat thermal systems consid- Orissa, India, and the M.E. and Ph.D. degrees from
ering generation rate constraints,” IEEE Trans. Power App. Syst., vol. Regional Engineering College, Rourkela, Orissa,
PAS-100, no. 5, pp. 2334–2342, May 1981. India, in 1990, 1992, and 2000, respectively.
[7] J. Nanda, M. L. Kothari, and P. S. Satsangi, “Automatic generation con- In 1992, he joined the Department of Electrical En-
trol of an interconnected hydrothermal system in continuous and dis- gineering, University College of Engineering Burla
crete modes considering generation rate constraints,” Proc. Inst. Elect. as a Lecturer and subsequently became a Reader in
Eng., vol. 130, pt.D, no. 1, pp. 17–27, Jan. 1983. 2001. Presently, he is an Associate Professor with the
[8] O. I. Elgerd , Electric Energy Systems Theory an Introduction, 2nd Department of Electrical Engineering, Indian Insti-
ed. New Delhi, India: Tata McGraw-Hill, 1983, pp. 299–33. tute of Technology Delhi, New Delhi, India. His in-
[9] L. Hari, M. L. Kothari, and J. Nanda, “Optimum selection of speed reg- terests are in soft computing applications to power system control and power
ulation parameters for automatic generation control in discrete mode quality and renewable energy.
considering generation rate constraints,” Proc. Inst. Elect. Eng. C, vol. Dr. Mishra has been honored with many prestigious awards, such as the INSA
138, no. 5, pp. 401–406, Sep. 1991. Young Scientist Medal in 2002, the INAE Young Engineer’s Award in 2002,
[10] F. Beaufays, Y. Abdel-Magid, and B. Widrow, “Application of neural and recognition as the DST Young Scientist in 2001 to 2002. He is a fellow of
network to load frequency control in power systems,” Neural Netw., the Indian National Academy of Engineering and Institute of Electronics and
vol. 7, no. 1, pp. 183–194, 1994. Communication Engineering.
[11] M. Djukanovic, M. Novicevic, D. J. Sobajic, and Y. P. Pao, “Con-
ceptual development of optimal load frequency control using artificial
neural networks and fuzzy set theory,” Int. J. Eng. Intell. Syst. Elect.
Eng. Commun., vol. 3, no. 2, pp. 95–108, 1995. Lalit Chandra Saikia received the B.E. degree from Dibrugarh University,
[12] G. A. Chown and R. C. Hartman, “Design & experience of fuzzy logic Assam, India, in 1993 and the M.Tech. degree in power systems in 2007 from
controller for automatic generation control (AGC),” IEEE Trans. Power Indian Institute of Technology Delhi, New Delhi, India.
Syst., vol. 13, no. 3, pp. 965–970, Aug. 1998. Since, 2000, he has been working as a Lecturer at the National Institute of
[13] M. A. Abido, “Optimal design of power-system stabilizers using par- Technology (formerly known as Regional Engineering College), Silchar, India.
ticle swarm optimization,” IEEE Trans. Energy Convers., vol. 17, no. His interests include soft computing applications to power system control and
3, pp. 406–413, Sep. 2002. power quality.

Vous aimerez peut-être aussi