You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/255811211

Comparison of PSO variants with traditional solvers for large scale multi-area
economic dispatch

Conference Paper · July 2011


DOI: 10.1049/cp.2011.0379

CITATIONS READS
5 182

3 authors, including:

Manjaree Pandit
Madhav Institute of Technology & Science Gwalior
117 PUBLICATIONS   1,649 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Bio-inspired algorithms for power system optimisation View project

Optimization of benchmark function using bird swarm algorithm for economic load dispatch View project

All content following this page was uploaded by Manjaree Pandit on 02 June 2014.

The user has requested enhancement of the downloaded file.


Chennai and Dr.MGR University Second International Conference on Sustainable Energy and Intelligent System (SEISCON 2011) ,
Dr. M.G.R. University, Maduravoyal, Chennai, Tamil Nadu, India. July. 20-22, 2011.

Comparison of PSO variants with traditional solvers for


large scale multi-area economic dispatch
Rameshwar Singh*, Kalpana Jain† , Manjaree Pandit†
* Department of Electrical Engineering, M.I.T.S., Gwalior, India.

Keywords: Constrained non-linear optimization, particle swarm port constraints was presented in reference[1]. Desell et al.[2]
optimization, economic load dispatch, transmission constraints. proposed an application of linear programming to transmission
constrained production cost analysis. Farmer et al.[3] presented
Abstract a probabilistic method. Hopfield neural network based approach
was also proposed to solve the MAED problem[4]. Doty and Mc-
Economic load dispatch (ELD) has the objective of generation al- Intyre[5] solved multi-area economic dispatch problem by using
location to the power generators such that the total fuel cost is mini- spatial dynamic programming with linear losses. Linear program-
mized and all operating constraints are satisfied. Generally ELD is ming application is proposed in[6] to production cost analysis with
solved without accounting for transmission constraints, however, transmission constraint. Wang and Shahidehpour[7] proposed a
in deregulated power system environment it is essential to include decomposition approach using expert systems.
practical multi-area cases with tie line constraints. A number of Soft computing based approaches are also becoming very popu-
traditional methods are used for solving ELD and other power lar. Although these methods do not always guarantee global best
system problems. During the last decade soft computing methods solutions, they often achieve a fast and near global optimal solu-
like particle swarm optimization (PSO), have been increasingly tion. Recently covariance matrix adapted evolutionary strategy
proposed for complex optimization problems. This paper proposes has been proposed for MAED problems where a Karush Kuhun
some improved PSO technique in which premature convergence is Tucker (KKT) optimality based stopping criterion is applied to
avoided by tuning the PSO parameters for enhanced global and lo- guarantee optimal convergence[8]. Ref.[9] gives a comparison of
cal search. The paper reviews and compares the performance of the
DE strategies for the MAED problem. Large dimension problems
proposed PSO variants with traditional solver GAMS for multi-area
are difficult to optimize using soft computing methods, as these
economic dispatch (MAED) on four standard test systems having
techniques take a long time to converge; on the other hand, tra-
different sizes and complexity levels. A large 120-unit power system
is included for validating the results. ditional methods like the GAMS solver computes the best result
almost instantaneously. This paper proposes some modified PSO
variants where a proper tuning of cognitive and social coefficient
1 Introduction is carried out to escape local minima and enhance global search.
There has been phenomenal growth in mathematical program-
Power utilities try to achieve high operating efficiency to pro- ming techniques and development of computer codes to solve
duce cheap electricity. Competition exists in the electricity sup- large scale optimization models over the past four to five decades.
ply industry in generation and in the marketing of electricity. The There has also been noteworthy development in relational data-
operating cost of a power pool can be reduced if the areas with base for improved data organization and transformation capabili-
more economic units generate larger power than their load, and ties. The tremendous effort that goes in developing an optimiza-
export the surplus power to other areas than their load, and ex- tion model has a significant part in debugging lengthy computer
port the surplus power to other areas with more expensive units. codes, data preparation, data transformation and report genera-
The benefits thus gained will depend on several factors like the tion. A number of efficient modeling languages have been devel-
characteristics of a pool, the policies adopted by utilities, types of oped which makes use of both the development in improved da-
interconnections, tie-line limits and load distribution in different tabase management and mathematical programming techniques.
areas. Therefore, transmission capacity constraints in production One of the most popular and flexible languages among these is
cost analysis are important issues in the operation and planning the General Algebraic Modeling System (GAMS)[10]. GAMS
of electric power systems. The economic dispatch problem is fre- module was originally developed through a World Bank funded
quently solved without considering transmission constraint. Com- study in 1988.
pared to the classical ED problem the MAED problem is more This paper reviews and compares the performance of proposed
complex due to the additional tie-line constraints and area power PSO based techniques with traditional solver GAMS for multi-
balance requirements. However, some researchers have taken area economic dispatch (MAED) on a large power system. Per-
transmission capacity constraints into consideration. A complete formance comparison is carried out for four standard test systems
formulation of multi-area generation scheduling with import/ex- having different sizes and complexity levels.

304
Chennai and Dr.MGR University Second International Conference on Sustainable Energy and Intelligent System

Section II of the paper gives an overview of the GAMS language, N


in section III, problem formulation for the MAED is given. Sec- MinFT = ∑ Fi ( Pi ) (1)
i=1
tion IV presents an overview of some PSO variants; Section V
summarizes the results on four test cases. where Fi is the total fuel cost for the ith generator (in $/h) which
is defined by,
2 Formulation using GAMS solvers Fi ( Pi ) = (ai Pi 2 + bi Pi + ci ) Rs / MWh (2)
The General Algebraic Modeling System (GAMS) is specifi-
cally designed for modeling linear, nonlinear and mixed integer where ai bi and ci are the fuel-cost coefficients of the ith unit.
optimization problems. The system is particularly very advanta- N
geous with large, complex problems. GAMS allows the user to ∑ Pi − ( PD + PL ) = 0 (3)
concentrate on the modeling problem by making the setup simple. i=1
GAMS is especially useful for handling large, complex, one-of-a- (4)
kind problems which may require many revisions to establish an Pi min ≤ Pi ≤ Pi max i =1, 2,..., N
accurate model. The user can change the formulation quickly and
easily, and can even change from one solver to another. Similarly For a given total real load PD the system loss PL is a function
the use can easily convert from linear to nonlinear optimization of active power generation at each generating unit. To calculate
option with little trouble. The basic structure of a mathematical system losses, methods based on penalty factors and constant loss
model coded in GAMS has the components: sets, data, variable, formula coefficients or B-coefficients[12] are in use. The latter
equation, model and output[10]. The data presentation in GAMS is adopted in this paper as per which transmission losses are ex-
can be done in its most elemental form using Table 1, columns etc. pressed as
There are standard IF-ELSE, WHILE, LOOP, exception handling N N N
PL = ∑ ∑ Pi Bij Pj + ∑ Boi Pi + Boo (5)
logic available which give the inherent flexibility to use GAMS
i=1 j=1 i=1
almost like any programming language while retaining the basic
advantages. Excellent debugging features exist for quick and ef- Tie-line power flow between areas plays a very important role in
fective identification of errors[11]. The available solvers provide deciding the operating cost in multi-area power systems. Taking
access to most of the state of the art optimization tools but new into consideration the cost of transmission though each tie-line,
solution algorithms for specific applications can be developed us- the objective function of MAED is given as
ing the existing ones and the basic features of GAMS with much
N M
lower efforts than needed in a conventional programming lan- MinFT = ∑ Fi ( Pi ) + ∑ f j (T j ) (6)
guage like FORTRAN. i=1 j

Here fj is the cost function associated with jth tie line power low
Table-1: The available solvers in GAMS10,11 Tj. There are N number of generating units and M number of tie-
Problem Type Solver Sub-System
lines.
In MAED problem the power balance constraints need to be sat-
LP BDMLP, MINOS5, ZOOM, MPSX, SCICONIC, APEX isfied for each area. The power balance constraints for area q
IV, LAMSP, OSL, XA, CPLEX
neglecting losses can be given as
NLP MINOS 5, CONOPT, GRG 2, NPSOL
MILP ZOOM, MPSX, SCICONIC, APEX IV, XA, OSL,
Nq ⎛ Mq ⎞
∑ Piq =⎜
⎜ PDq + ∑ T jq ⎟
⎟= 0 (7)
LAMPS
i=1 ⎝ j ⎠
MINLP DICOPT++
for q = 1,2……..M (areas). For the qth area , PDq is the load de-
The tool kit in GAMS gives algorithms for each category of prob- mand , Tjq is the tie- line flows from other areas, Nq are number
lem. GAMS also has the unique feature of providing a common of generating units and Mq represents the number of tie-lines con-
language that can make use of a variety of solvers. Table-1gives nected to the qth area.
the list of currently available GMS solvers. The choice of solver is
thus dependent on the understanding of the special structure of the
problem and some experimentation In this paper the performance 4 Overview of PSO variants
of GAMS is compared with PSO variants for a large 120-unit
multi-area ELD problem with changing loads and tie-line limits. A number of different PSO strategies are being applied by re-
searchers for solving the ELD and other power system problems.
Here, a short review of the significant developments is presented
3 Multi-area economic load dispatch to serve as a performance measure for the GAMS technique ap-
plied in this paper.
The objective of the economic dispatch problem is to determine
the generated powers Pi of units for a total load of PD so that the 4.1 Classical PSO
total fuel cost, FT for the N number of generating units is mini-
The PSO[13] is a population based modern heuristic search meth-
mized subject to the power balance constraint and unit upper and
od inspired by the movement of a flock of birds searching for
lower operating limits. The objective is
food. It is a simple and powerful optimization tool which scatters

305
Comparison of PSO variants with traditional solvers for large scale multi-area economic dispatch

random particles into the problem space. In the proposed PSO_CAC approach the cognitive coefficient c1
The position and velocity vectors of the ith particle of a d-dimen- is reduced from an initial value c1i to a final value c1f while the
sional search space can be represented as Xi = (xi1,xi2 ,............ social coefficient c2 is increased chaotically from an initial value
xid) and Vi = (vi1,vi2,........v_id) respectively. On the basis of c2i to c2f using the following dynamics:
the value of the evaluation function, the best previous position
of a particle is recorded and represented as pbes ti = (pi1,pi2........
pid) .If the gth particle is the best among all particles in the group
so far, it is represented as pbes tg = gbes t = (pg1,pg2,..........pgd)
. The modified velocity and position of each particle for fitness
evaluation in the next i.e. (k +1)th iteration are calculated using
the following equations:
vidk+1 = C[ w × vidk + c1 × rand1 × ( pbestid − xid ) + c2 ×
(8)
rand 2 × ( gbest gd − xid )}

xidk+1 = xid + vidk+1 (9)

Here w is the inertia weight parameter which controls the global


and local exploration capabilities of the particle. Constant C is
constriction factor, c1, c2 are cognitive and social coefficients re- Figure 1. Inertia weight in PSO_CIW ; μ =4, w(t=0) =0.9
spectively, and rand1},rand2 are random numbers between 0 and (14)
1. As per the concept of time-varying inertial weight w is given cx1 (t ) = μ × cx1 (t −1) ×[1− cx1 (t −1)]
by ⎡ iter ⎤
c1 (t ) =⎢ (c1 f − c1i ) + c1i ⎥cx1 (t ) (15)
(itermax − iter ) ⎣ itermax ⎦
w = ( wmax − wmin ) × + wmin (10)
itermax
cx2 (t ) = μ × cx2 (t −1) ×[1− cx2 (t −1)] (16)
Where itermax is the maximum number of iterations. Constant c1
pulls the particles towards local best position whereas c2 pulls it ⎡ iter ⎤
c2 (t ) =⎢ (c2 f − c2i ) + c2i ⎥cx2 (t ) (17)
towards the global best position. To improve the convergence the ⎣ itermax ⎦
constriction factor is also in use.
The variation of Cognitive and social coefficients has been shown
2 for μ =3 in Figure.2 respectively.
C= where 4.1 ≤ φ ≤ 4.2 (11)
2 − φ − φ2 − 4 φ
4.4 Time varying PSO (PSO_TVAC)
4.2 PSO with chaotic inertia weight (PSO_CIW) In population-based optimization methods, the policy is to en-
courage the individuals to roam through the entire search space,
The weight w in (8) is changed chaotically making use of One of during the initial part of the search, without clustering around lo-
the simplest dynamic systems evidencing chaotic behavior is the cal optima. In PSO_TVAC, this is achieved by changing the ac-
iterator called the logistic map14, whose equation is given by celeration coefficients c1 and c2 with time in such a manner that
the cognitive component is reduced while the social component is
y (t ) = μ × y (t −1) ×[1− y (t −1)] (12) increased as the search proceeds. During the latter stage in optimi-
zation, a small cognitive component and a large social component
where t is the iteration count and μ is a control parameter, 0 ≤ allow the particles to converge to the global optima. The accelera-
μ ≤ 4 . The value of μ controls the variation of the chaotic se- tion coefficients are expressed as[15]:
quence. The weight is iteratively varied as
iter
(13) c1 = (c1 f − c1i ) + c1i (18)
w(t ) = μ × w(t −1) ×[1− w(t −1)] itermax
A very small difference in w(0) causes significant difference in
iter
its variation pattern. The system at (13) displays chaotic behavior c2 = (c2 f − c2i ) + c2i (19)
when μ = 4 and w(0) ∉ {0,0.25,0.5,0.75,1.0}. The variation of w itermax
has been shown for μ =4 in Figure.1 respectively. where c1i, c1f , c2i and c2f are initial and final values of cognitive
and social acceleration factors respectively.
4.3 PSO with chaotic acceleration coefficients (PSO_
CAC) 5 Results and Discussions
Kennedy and Eberhart[13] stated that a relatively higher value of The performance of traditional optimization approach using the
the cognitive component, compared with the social component, NLP minimization module of GAMS has been compared with
results in roaming of individuals through a wide search space. PSO and its variants for four test cases having different sizes and

306
Chennai and Dr.MGR University Second International Conference on Sustainable Energy and Intelligent System

complexity levels as described below. Simulations were carried 5.2 Parameter setup for PSO variants
out using MATLAB 7.0.1 on a Pentium IV processor, 2.8 GHz.
with 1 GB RAM. For all PSO variants the population size was taken as 100 and
number of iterations was set at 100 for Case I and Case II and 700
for the remaining two test cases. For classical PSO both accelera-
tion coefficients were fixed at 2. The initial and final acceleration
coefficients in PSO_TVAC and PSO_CAC were taken as 2.5 and
0.5 respectively. As PSO type techniques converge to different
solutions in every run, results were computed out of 50 trials with
different initial populations. Fig. 3 shows the convergence behav-
iour of the PSO variants for Test case III. The PSO_TVAC was
found to me most consistent followed by PSO_CIW and PSO_
CAC.

5.3 Simple ELD cases


Figure 2 Cognitive and social coefficients in PSO_CAC for μ=3, Table 5 presents the results of GAMS for the simple 3-unit ELD
c1,c2(t=0) =0.48 case for different values of demand. The results match with the
results given in[17]. To increase the complexity level, losses are
also included. The results for different load demands for test case
5.1 Description of the test cases
II are given in Table 6. It can be seen from both the Tables that the
The performance of traditional optimization approach using the power balance violations and unit operating constraints are satis-
NLP minimization module of GAMS has been compared with fied for all load cases.
PSO and its variants for four test cases having different sizes and
complexity levels as described below. 5.4 Effect of tie-line limits in MAED
(1) Test case I: This system is taken from[17]. It has 3-generat- In multi-area economic dispatch the operating cost is heavily in-
ing units supplying a total load of 800 MW. Transmission fluenced by the tie-line limits. For the Test case III, i.e. 2-area,
losses are neglected while minimizing cost function given 120-unit large system the tie-line limits are changed from 500
by eq. (1) subject to constraints given by (3) and (4).The MW to 700 MW and the cost is found to reduce by $80/h for
fuel-cost characteristics are given in Table 2. PD1=18200 and PD2=9800.The results are given in Table 7. It
(2) Test case II: Economic dispatch is carried out including can be seen that the area power balance constraints are met sat-
power losses. The cost coefficients are given in Table 3 and isfactorily.
loss coefficients are given in Table 4[17].
(3) Test case III: This is a large multi-area economic dispatch Table 4: B-loss coefficients for test case II
problem having 120 generating units divided into two ar-
Bij 0.0218 0.0093 0.0028
eas. This system is created by expanding the 40-unit system
taken from18.The total load demand is 28,000MW and the 0.0093 0.0228 0.0017
transmission maximum capacity limit is 3000MW. The first 0.0028 0.0017 0.0179
area has 70 units and 65% of the total load demand (i.e. Bio 0.003 0.0031 0.0015
PD1=18200) and the second area has 50 units and 35% of
Boo 0.00030523
total load demand (PD2=9800).
(4) Test case IV: The above 120-unit system is formulated for
single area operation without tie-line.
Table 2: Fuel cost coefficient of Test case I

UNIT Pmin(MW) Pmax(MW) ai bi ci


1 200 450 0.004 5.3 500
2 150 350 0.006 5.5 400
3 100 225 0.009 5.8 200

Table 3: Fuel cost coefficient of Test case II

UNIT Pmin(MW) Pmax(MW) ai bi ci


1 10 85 0.008 7.0 200
2 10 85 0.009 6.3 180
3 10 70 0.007 6.8 140

Figure 3 Convergence characteristics of PSO variants; case III

307
Chennai and Dr.MGR University Second International Conference on Sustainable Energy and Intelligent System

Table 5. Results of ELD for Test case I; different loads 5.6 Comparison of PSO variants and GAMS for Large
Load(MW) P1(MW) P2(MW) P4(MW) Cost($/h) Power
single and multi-area ELD
balance Table 9 and Table 10 show comparison of single and multi area op-
violation eration. The cost reduced from$362451.4854/h to $362433.2648/
800 400.000 250.000 150.000 6682.500 0.0000 h for 28000 MW and from $315116.17452/h to $315114.4120/h
850 423.684 265.789 160.526 7112.237 0.0010 for 24000 MW when tie-line limits were released.
975 450.000 325.000 200.000 8236.000 0.0000 Table 10 Cost comparison of GAMS with PSO variants
1000 450.000 340.000 210.000 8473.000 0.0000 TEST PSO_ PSO_ PSO_ GAMS($/h)
CASE CIW($/h) CAC($/h) TVAC($/h)
Table 6: Results of ELD for Test case II; different loads
Test case I 6682.5158 6682.5018 6682.5000 6682.5000
Load P1 P2 P4 Loss Cost($/h) Power Test case II 1592.6515 1592.6505 1592.5600 1592.6500
(MW) (MW) (MW) (MW) (MW) balance Test case III 364648.3785 364458.2457 362451.4856 362451.4854
violation
Test case IV 364489.7457 362449.4107 362433.8457 362433.2648
100 18.341 49.280 33.204 0.824 1216.899 0.0000
150 35.091 64.132 52.477 13699 1592.650 The performance of PSO_TVAC was observed to be the best,
0.0010
200 52.958 79.989 70.000 2.947 1984.960 0.0000 matching in accuracy with GAMS but with almost 100 times
more computational time as given in Table 11. However, the ad-
225 78.686 80.000 70.000 3.686 2192.234 0.0000
vantage of PSO technique is its ability to work efficiently for non-
Table7: Effect of tie- limits in 120-unit MAED Test case II convex, multimodal and discontinuous functions for which tradi-
tional NLP solvers like the GAMS can not find solution [16].For
PD1 PD2 Tie- Tie-line Total Areal Areal Time quadratic cost functions GAMS is superior to all PSO variants.
(MW) (MW) line flow cost viola- viola- (sec)
Table 11 Time comparisons of GAMS with PSO variants
con- ($/h) tion tion
straints TEST PSO_ PSO_ PSO_ GAMS(sec)
18200 9800 700 -694.129 362433 -0.0010 -0.0010 0.016 CASE CIW(sec) CAC(sec) TVAC(sec)
18200 9800 600 -600.000 362451 -0.0020 0.0020 0.016 Test case I 0.03672 0.03672 0.03671 0.016
18200 9800 500 -500.000 362513 -0.0020 -0.0010 0.016 Test case II 0.14682 0.14681 0.14680 0.047
Test case III 1.6015 1.6015 1.6010 0.016
Table 8: Effect of load in 120-unit multi-area
Test case IV 1.4505 1.4503 1.4502 0.015
PD1 PD2 Tie- Tie-line Total cost Area 1 Area 2
(MW) (MW) linecon- flow ($/h) viola- viola-
straints tion tion
6 Conclusion
18200 9800 600MW -600MW 362451.4854 -0.0020 0.0020 The performance of PSO variants was compared with traditional
15600 8400 600MW -600MW 315116.17452 -0.0020 0.0000 NLP solver GAMS for economic dispatch problem of four test
cases. The following conclusions were drawn.
Table 9 Effect of load; single area 120-unit system; Case IV
• Soft computing techniques like the PSO use random opera-
Load (MW) Total cost ($/h) Power balance viloation Time(sec) tors for achieving the optimal result therefore in every fresh
28000 362433.2648 -0.0020 0.016 trial, these methods converge to different solutions near the
global best solution. The traditional NLP algorithm like the
24000 315114.4120 -0.0010 0.015
GAMS uses mathematical operations to achieve the best
solution so they are always consistent and converge to the
5.5 Effect of load variation unique global minimum solution.
The performance of GAMs was tested for a total load of 24,000MW • The time taken by soft computing techniques is quite large
for test case III, with PD1=15600 and PD2=8400. The GAMS as compared to GAMS. The time requirement increases tre-
module converged within 0.0016 seconds for this large system. mendously with problem complexity (like the inclusion of
The results are given in Table 8. When load was decreased from losses) and with increase in problem size.No such issue is
28000 MW (PD1=18000 MW; there with GAMS.
• Both GAMS and PSO variants are able to handle complex
PD2=9800MW) to 24000 MW (PD1=15600 MW; PD2=8400MW)
constraints like generation limits, area-wise power balance
the optimal cost was reduced from $362451.4854/h to
and losses effectively.
$315116.17452/h for tie-line limit 600 MW between the two ar-
• Soft computing techniques however are becoming popular
eas.
for non-convex, multimodal, discontinuous optimization
problem for which traditional methods cannot provide solu-
tion.

308
Chennai and Dr.MGR University Second International Conference on Sustainable Energy and Intelligent System

Acknowledgement [11] Richard E. Rosenthal, GAMS, A User’s Guide, Tutorial,


GAMS Development Corporation. Washington, 2010.
The authors sincerely acknowledge the financial support provided [12] Wood and B.F. Wollenberg, “Power generation, operation
by UGC under major research project entitled Power System Op- and control”, (Book), New York, Wiley, 1984.
timization and Security Assessment Using Soft Computing Tech- [13] J. Kennedy and R. Eberhart, “Particle swarm optimization”,
niques, vide F No.34-399/2008 (SR) dated, 24th December 2008. proc. IEEE Conf. on Neural Networks (ICNN’95), vol. IV,
The authors also thank the Director, M.I.T.S. Gwalior for provid- Perth, Australia, 1995 pp. 1942-48.
ing facilities for carrying out this work. The third author acknowl- [14] R. Caponetto, L. Fortuna, S. Fazzino, and M. G. Xibilia,
edges UGC research award for post doctoral work sanctioned by “Chaotic sequences to improve the performance of evolu-
UGC, New Delhi vide letter no. F-30-120(SC)/2009 (SA-II). tionary algorithms,” IEEE Trans. Evol. Computat., vol. 7,
no. 3, Jun. 2003 pp. 289–304.
References [15] A. Ratnaweera, S.K. Halgamuge and H.C. Watson, “Self-
organizing hierarchical Particle swarm optimizer with time
[1] R.R.Shoults, S.K. Chang, S. Helmick and W.M. Grady, “A varying acceleration coefficients”, IEEE Trans Evol Com-
practical approach to unit commitment, economic dispatch putation, vol. 8, No. 3, Aug. 2004 pp. 240-255.
and savings allocation for multiple area pool operation with [16] J.B.Park, Y.W.Jeong and J.R.Shin, “An improved particle
import/export constraints”, IEEE Trans Power Apparatus swarm optimization for nonconvex economic dispatch
and systems, vol. 99, No.2, 1980 pp. 625-35. problems” IEEE Trans. Power system,. vol. 25, no. 1, 2010
[2] A.L. Desell, K. Tammar, E.C. McClelland and P.R. Van pp.156-16.
Home, “Transmission constrained production cost analysis [17] Hadi Sadat, Power System Analysis, Tata McGraw Hill,
in power system planning”, IEEE Trans. on Power Appa- “Economic Dispatch Neglecting losses and including Gen-
ratus and Systems, vol. PAS-103, NO. 8, 1984 pp. 2192- erator limits, pg. no. 277, International Edition,1999.
2198. [18] P.H.Chen and H.C.Chang, “Large scale economic dispatch
[3] E.D. Farmer, M.J. Grubb and K. Vlahos, “Probabilistic approach by genetic algorithm”,IEEE Transactions on Pow-
production costing of transmission-constrained power sys- er Systems, vol. 10, No.4, November 1995 pp. 1919-1926.
tems”, 10th PSCC Power System Computation Conference, [19] J.B. Park et al., “An improved PSO for non-convex economic
1990pp. 663-669. dispatch problems”, IEEE Transactions on Power Systems,
[4] T. Yalcinoz and M. J. Short:, “Neural Networks Approach vol. 25, No. 1, February, 2011.
for Solving Economic Dispatch Problem with Transmission [20] Krishna Teerth Chaturvedi, Manjaree Pandit and Laxmi
Capacity Constraints”, IEEE Trans. on Power Systems, vol. Srivastava, “Particle Swarm Optimization with Time Vary-
13, No. 2, May 1998 pp. 307-313. ing Acceleration Coefficients for Nonconvex Economic
[5] Doty and P.L. McEntire, “An analysis of electrical power Power Dispatch”, Electrical Power and Energy systems,
brokerage systems”, IEEE Trans Power Apparatus System, vol. 31, No.6, July 2009 pp. 249-257.
vol. 101, No.2, 1982 pp. 389-396.
[6] Hemick SD,Shoults RR, A practical approach to an interim
multi-area economic dispatch using limited computer re- Biographies
sources. IEEE Trans Power Apparatus System 1985;104(6):
1400-4 Rameshwar Singh obtained his B.E. degree in Electrical Engi-
[7] Wang C, Shahidehpour SM. A decomposition approach to neering from Madhav Institute of Technology & Science Gwalior
non -linear multi-area generation scheduling with tie-line (India) in 2009. He is doing M.E.(ISD) from Madhav Institute of
constraints using expert system. IEEE Trans Power Appa- Technology & Science (India).
ratus System 1992;7(4): 1409-18 Kalpana Jain obtained her M.E. degree in Electrical Engi-
[8] P.S. Manoharan , P.S. Kannan, S. Baskar, M. Willjuice neering from Madhav Institute of Technology & Science Gwalior
Iruthayarajan, “Evolutionary algorithm solution and KKT (India) in 2010. She is currently working as Project Fellow in the
based optimality verification to multi-area economic dis- department of Electrical Engineering, M.I.T.S.,Gwalior (India)
patch”, Electrical Power and Energy Systems vol. 31, pp. Her areas of interest are Power System Security Analysis, Opti-
365–373, 2009A.J. mization and soft computing/evolutionary methods.
[9] Manisha Sharma, Manjaree Pandit and Laxmi Srivastava,
“Multi-Area Economic Dispatch with Tie-Line Constraints M. Pandit obtained her M.Tech degree in Electrical Engineer-
Employing Evolutionary Approach”, International Journal ing from Maulana Azad College of Technology, Bhopal, (India)
of Engineering Science and Technology (IJEST), (Spe- in 1989 and Ph.D. degree from Jiwaji University Gwalior (India)
cial Issue on Application of Computational Intelligence in in 2001. She is currently working as Professor in Department of
Emerging Power Systems), Vol. 2, No. 3, 2010 pp. 133- Electrical Engineering, M.I.T.S., Gwalior, (India). Her areas of
150. interest are Power System Security Analysis, Optimization and
[10] Debabrata Chattopadhyay “Application of general algebra- soft computing/ evolutionary methods, ANN and Fuzzy neural
ic modeling system to power system optimization” IEEE applications to Power System.
Trans. on Power Systems, vol. 14, No. 1, February, 1999.

309

View publication stats

You might also like