2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia.

3-4 June 2013

A Hybrid Differential Evolution and Harmony Search

for Nonconvex Economic Dispatch Problems
Le Dinh Luong Vo Ngoc Dieu, Nguyen Thanh Hop, and Le Anh Dung
Faculty of Mechanical - Electrical - Electronic Department of Power Systems
HCMC University of Technology, Vietnam HCMC University of Technology, Vietnam
Email: [email protected] Email: [email protected]

Abstract- In this paper, we presents a novel approach for solving the ED problem consider valve-point effects was
solving economic dispatch (ED) problems with nonconvex cost proposed in [5], simulated annealing (SA) and the hybrid
functions using a hybrid differential evolution and harmony GA/SA for dealing with classical ED problems [6,7], particle
search (DEHS). The DEHS method is an improved differential swarm optimization (PSO) with improvements and a new PSO
evolution method based on the harmony search scheme. The
DEHS method has the flexible adjustment of the parameters to
hybrid with local search [8], etc. However, these methods
get a better optimal solution. Moreover, an effective constraint have large number of iterations to achieve solution and easily
handling framework in the method is employed for properly affected by the relevant control parameters. One powerful
handling equality and inequality constraints of the problem. The algorithm from evolutionary computation due to it’s excellent
proposed DEHS has been tested on three systems with 40, 15 and convergence characteristics and few control parameters is
140 units and the obtained results from DEHS algorithm have differential evolution. There have been many applications of
been compared to those from other methods in the literature. The DE for solving ED problems such as an improved DE (IDE)
result comparison has indicated that the proposed DEHS method based on cultural algorithm and diversity measure has been
is more effective than many other methods for obtaining the discussed in [9] to solve two problems with valve point
optimal solution for the test systems. Therefore, the proposed
effects, a DE algorithm with a specially designed repair
DEHS is a very favorable method for solving the nonconvex ED
problems. operation was proposed in [10] to solve the ED problem with
different constraints, a new algorithm by combining a chaotic
I. INTRODUCTION differential evolution (CDE) and quadratic programming has
been discussed in [11] to solve the ELD problem with valve
In power system operation, the operation cost needs to be point effects. Harmony search (HS) is a novel meta-heuristic
minimized at each time satisfying constraints via economic algorithm. It can be conceptualized from the observation that
dispatch (ED) problem. Previous efforts on solving ED the aim of music is to search for a perfect state of harmony.
problems have employed various mathematical programming Since its inception, HS method has been successfully applied
methods and optimization techniques such as linear to a wide variety of practical optimization problems such as
programming, quadratic programming, gradient based steel frame design was proposed in [12], combined heat and
method, Lagrange relaxation, etc. Many algorithms have power economic dispatch problem has been discussed in [13],
bean used to solve the ED problems by approximating the scheduling problem was proposed in [14], etc. However, the
cost function of each generator by a single quadratic number of article about this HS algorithm for solving ED
function. In practical power system operation conditions, problems are still limited such as an improved HS algorithm to
many thermal generating units, especially those units have solve the 13-unit ED problem with valve point effects [15] and
valve-point effects, prohibited operating zones, have formed a an improved HS algorithm was proposed by tuning one
nonlinear ED problem with many local optima and multiple control parameter depending on the variance of the population
constraints in nature, which prevents the classical methods for the 6-unit problem [16].
from obtaining the global optima. This is a complicated, real- This paper proposes a hybrid differential evolution and
world and non-convex optimization problem since it contains harmony search (DEHS) algorithm by combining the
the discontinuous values at each boundary forming multiple mechanisms of both HS and DE. We had some experiments of
local optimal. Therefore, the classical solution methods have the classical HS meta-heuristics with slow convergence and
difficulty in dealing with this problem. the growth of search space dimensionality. In the present
In recent years, many methods have been applied to solve work, we replace the pitch adjustment operation in original HS
the non-convex ED problem such as evolutionary with a mutation strategy borrowed from the realm of the DE
programming (EP) for solving the ED problem with multiple algorithms. After that, the memory consideration and the
fuel cost function has been discussed in [1, 2], genetic enhanced pitch adjustment operation are both employed to
algorithm (GA) for soling the ED problem with many types of strengthen the exploration ability. Compared with the classical
fuel cost function, GA for solving the ED problem with valve- HS meta-heuristics, the use of differential mutation and
point effects was proposed in [3], tabu search algorithm (TSA) crossover can enhance the exploitation in the DEHS. DEHS
for solving the ED problem with multiple minima [4], TSA for algorithm may inherit elements from as many individuals as

978-1-4673-5074-7/13/$31.00 ©2013 IEEE 238

 2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June 2013

its number of dimensions when generating a new individual to To combine the ramp rate limits into power output limits
enhance the exploration ability. The proposed DEHS has been constraints, the generator limits (5) can be rewritten as:
tested on three systems including 40 unit system with valve max { Pi ,min , Pi 0 − DRi } ≤ Pi ≤ min {Pi ,max , Pi 0 + URi } (8)
effects, 15 unit system with prohibited operating zones, ramp
rate constraints, and power loss, and 140-unit system with B. ED Problem with Valve-Point Effects
valve point effects, prohibited operating zones, and ramp rate The ED with valve point loading effects (VPE) is a non-
limits. The numerical results from the proposed method are smooth and nonconvex problem with the cost function include
compared to those from many other methods in the literature. the ripple curve. The generating units with valve point effects
steam turbines exhibit a greater variation in the fuel-cost
II. NONCONVEX ECONOMIC DISPATCH functions. Since the valve point results in the ripples, the cost
PROBLEM function becomes more nonlinear. Therefore, the equation (2)
A. Classical ED Problem should be replaced by (9) for considering the valve-point
The main purpose of the economic dispatch problem is to effects. The sinusoidal functions are thus added to the
minimize the total fuel cost subject to the constraints of a quadratic cost functions as following.
power system. This problem is formulated as following
N
Fi ( Pi ) = ai + bi Pi + ci Pi 2 + ei × sin( fi × ( Pi ,min − Pi )) (9)
C = ∑ Fi ( Pi ) (1)
i =1
NG where ei and fi are the fuel cost coefficients of the ith unit with
F ( x ) = ∑ ( ai + bi PGi + c P ) 2
i Gi (2) valve point effects.
i −1 C. ED Problem Considering Prohibited Operating Zones
In some cases, the operating range of a generating unit is not
where Fi(Pi) is the fuel cost function of the ith unit, ai, bi, ci are always available due to limited physical operation. Units may
the fuel cost coefficients of the ith unit, NG is the number of have prohibited operating zones due to generator have some
generators, and Pi is the power generated by the ith unit. faults in operation or associated auxiliaries. Such faults may
lead to instability in certain ranges of generator power output
subject to [6]. Thus, for generating units with prohibited operating zones,
- Power balance constraints there are additional constraints on the unit operating range as
N follows:
∑P = P i D + PLoss (3) ⎧ Pi ,min ≤ Pi ≤ Pi l,1
i =1 ⎪⎪
Pi ∈ ⎨ Pi ,uk −1 ≤ Pi ≤ Pi ,lk (10)
where PD is the system load demand and PLoss is the ⎪ u
transmission loss. The power loss can be approximately ⎪⎩ Pi , pzi ≤ Pi ≤ Pi ,max
l u
calculated by using Kron’s loss formula: where k = 2, 3,…, pzi; i = 1, 2,…, npz; Pi , k and Pi , k are the
NG NG NG
lower and upper bounds of prohibited operating zone of unit i
PL = ∑∑
i =1 j =1
PGi Bij PGi + ∑P
j =1
Gi Bi 0 +B00 (4) respectively. Here, pzi is the number of prohibited zones of
unit i and npz is the number of units which have prohibited
operating zones.
where Bij, Bi0, B00 are power loss coefficients or B-coefficients.
- Generating capacity limits III. HYBRID DIFFERENTIAL EVOLUTION AND
Pi ,min ≤ Pi ≤ Pi ,max (5) HARMONY SEARCH
A. Differential Evolution
In 1995, Price and Storn proposed a new evolutionary
where Pi,min and Pi,max are the minimum and maximum power algorithm for global optimization and named it as differential
outputs of the ith unit. evolution (DE). This method has few parameters for tuning
- Ramp rate constraints make the algorithm quite popular in the literature. DE obtains
The actual operating range of all the online units is restricted solutions to optimization problems using three basic
by their corresponding ramp rate limits. The ramp-up and operations including mutation, crossover, and selection. The
ramp-down constraints can be written as follows: steps of DE including the operations are described as follows:
Pi – Pi0 ≤ URi as power increases (6) Step 1: mutation
Choose the target vector xi,g (= x0,g) and basic vector xr0,g (=
Pi0 - Pi ≤ DRi as power decreases (7) x2,g).
Randomly select two vector components xr1,g (= x3,g) and
where Pi0 is the previous power output of the ith xr2,g (= xNp-2,g.).
generating unit; URi and DRi are the up-ramp and down-ramp Calculate the value for a mutant vector:
limits of generator i, respectively. vi , g = xr 0, g + F .( xr1, g − xr 2, g ) (11)

239
 2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June 2013

Step 2: crossover Step 2: Harmony memory (HM) initialization.

Select new vector from the target vector and mutant vector Step 3: Improvise a new harmony from HM. A new harmony
according to the following rules: vector is generated from HM based on based on three rules:
⎪⎧v j ,i , g if (rand j (0,1) ≤ Cr or j = jrand ) memory considerations, pitch adjustments, and randomization.
ui , g = u j ,i , g = ⎨ (12) The HMCR, which varies between 0 and 1, is the probability
⎪⎩ x j ,i , g otherwise of choosing one value from the historic values stored in the
HM, and (1-HMCR) is the probability of randomly choosing
Step 3: selection one feasible value not limited to those stored in the HM.
Calculate the value of the objective function with the new ⎧⎪ x ji , g if ( rand (0,1) ≤ HMCM ,
vector is created, compare this value to the objective function xi , g +1 = ⎨ (14)
value of the first selected target vector to decide whether to ⎪⎩li + rand (0,1)(ui − li ) with probability (1- HMCR)
select a new vector or not.
⎪⎧ui , g if f ( ui , g ) ≤ f ( xi , g )
The pitch adjusting process is performed only after a value is
xi , g +1 = ⎨ (13) chosen from the HM. The value PAR sets the rate of doing
⎪⎩ xi , g otherwise pitch adjusting and the value (1-PAR) sets the rate of doing
nothing.
The proceudre of the DE method for solving an optimization ⎧⎪ xi , g +1 - rand (0,1) * bw if (rand (0,1) ≤ 0.5,
problem is given in Fig. 1. xi , g +1 = ⎨ (15)
⎪⎩ xi , g +1 + rand (0,1) * bw if (rand (0,1) > 0.5.

where bw is an arbitrary distance bandwidth (a scalar number)

and rand() is a uniformly distributed random number between
0 and 1.

Fig. 1. The procedure of DE

B. Harmony Search
Harmony search (HS) algorithm was recently developed in
an analogy of music improvisation process where music
players improvise the pitches of their instruments to obtain
better harmony [14]. First of all, HS initializes harmony
memory (HM) which stores the feasible vectors that are all in
the feasible space. The harmony memory size (HMS)
determines the number of vectors to be stored. Then, through
the harmony memory considering rate (HMCR), HS choose Fig. 2. Flow-chart of the DEHS algorithm
any one value from the HM, utilize the pitch adjusting rate
(PAR), choose an adjacent value of one value from the HM, Step 4: Update HM. If the new harmony vector is better than
and choose totally random value from the possible value range the worst harmony in HM, judged in terms of the objective
[25]. The steps in the procedure of HS are as follows: function value, the new harmony is included in HM and the
Step 1: Initialize the algorithm parameters and optimization existing worst harmony is excluded from HM.
operators such as HMS, HMCR, and PAR.

240
 2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June 2013

Step 5: Repeat Steps 3 and 4 until the terminating criterion is proposed DEHS is better than many other methods for this
satisfied. system.
TABLE II
C. Hybrid Differential Evolution and Harmony Search
RESULT COMPARISON OF 40-UNIT SYSTEM CONSIDERING
The overall procedure of the proposed algorithm can be VALVE-POINT EFFECTS
summarized as in Fig. 2. Method Total cost ($)
ABC [18] 121441.03
IV. NUMERICAL RESULTS SA-PSO [24] 121430.00
GA-PS-SQP [31] 121458.14
ACO [19] 121532.41
The proposed DEHS algorithm has been applied to ED SOH-PSO [26] 121501.14
problems in three different power systems including 40-unit FAPSO-NM [32] 121418.30
system with valve-point effects, 15-unit system with BSA-ES [20] 121686.00
prohibited operating zones, ramp rate limits, and transmission ICA-PSO [27] 121413.20
network losses, and 140-unit Korean power system with CBPSO-RVM [33] 121555.32
DE [22] 121416.29
valve-point effects, prohibited operating zones, and ramp DEC-SQP [28] 121741.98
rate limits. For each system, the proposed DEHS method is DE-BBO [34] 121420.89
run 100 independent trials and the obtained optimal results are APSO [23] 121663.52
compared to those from other methods. Self-tuning HDE [29] 121698.51
TSARGA [35] 121463.07
PSO-RDL [24] 121468.82
Case study 1: 40-unit system with valve point loading effects UHGA [30] 121424.48
In this example, the test system consists of 40 generating DEHS 121412.54
units with valve-point effects and the characteristics of the
40 unit are described in [26]. The total demand of the system Case study 2: 15-unit system with prohibited operating zones,
is 10500 MW neglecting power loss and ramp rate constraints. ramp rate limits, and transmission network losses
The DEHS parameters are set as follows: HMS = 50, MaxImp This test system is from [37] and the load demand for this
= 8000, HCMR = 0.95, PAR = 0.8, cr = 0.1, and w = 0.9. system is 2630 MW. The power loss and ramp rate constraints
are included in this system. The DEHS parameters for this
TABLE I
OPTIMAL DISPATCH OF 40-UNIT SYSTEM
system are set to HMS = 60, MaxImp = 4000, HCMR = 0.95,
Unit Pi,min Pi,max Output (MW) Unit Pi,min Pi,max Output (MW) PAR = 0.8, cr = 0.2, and w = 0.9.
1 36 114 110.79983 21 254 550 523.27937 The best solution, total power loss, and total cost obtained
2 36 114 110.79983 22 254 550 523.27937 by the DEHS method for the 15-unit system with prohibited
3 60 120 97.39991 23 254 550 523.27937 operating zones, ramp rate limits, and transmission network
4 80 190 179.73310 24 254 550 523.27937 losses are compared to those from genetic algorithm (GA) [37]
5 47 97 87.79990 25 254 550 523.27937 and particle swarm optimization (PSO) [37] as shown in Table
6 68 140 140.00000 26 254 550 523.27937 III. Obviously, the minimum cost obtained by the proposed
7 110 300 259.59965 27 10 150 10.00000
DEHS method is better than that from both PSO and GA.
8 135 300 284.59965 28 10 150 10.00000
9 135 300 284.59965 29 10 150 10.00000
TABLE III
10 130 300 130.00000 30 47 97 87.79990 RESULT COMPARISON OF 15-UNIT SYSTEM CONSIDERING PROHIBITED
11 94 375 94.00000 31 60 190 190.00000 OPERATING ZONES AND TRANSMISSION NETWORK LOSSES
12 94 375 94.00000 32 60 190 190.00000 Unit DEHS PSO [37] GA [37]
13 125 500 214.75979 33 60 190 190.00000
1 455.00088 439.12 415.31
14 125 500 394.27938 34 90 200 164.79983
2 380.00000 407.97 359.72
15 125 500 394.27937 35 90 200 194.39777
3 130.00000 119.63 104.42
16 125 500 394.27937 36 90 200 200.00000
17 220 500 489.27937 37 25 110 110.00000 4 130.00000 129.99 74.98
18 220 500 489.27937 38 25 110 110.00000 5 170.00000 151.07 380.28
19 242 550 511.27937 39 25 110 110.00000 6 460.00000 459.99 426.79
20 242 550 511.27937 40 242 550 511.27937 7 430.00000 425.56 341.32
Total generation (MW) 10500.0000 8 71.78281 98.56 124.79
Min cost ($/h) 121412.5357 9 58.87773 113.49 133.14
Average cost ($/h) 121414.9488 10 160.00000 101.11 89.26
Max cost ($/h) 121426.7608 11 80.00000 33.91 60.06
Standard deviation ($/h) 3.385839746 12 80.00000 79.96 50
13 25.00000 25 38.77
Table I shows the best optimal solution with total power 14 15.00011 41.41 41.94
generation, the minimum, mean, maximum cost achieved by 15 15.00000 35.61 22.64
Total power (MW) 2660.661538 2662.41 2668.44
the proposed method for the system. The best total cost from
Total loss (MW) 30.66153766 32.42 38.28
the DEHS method is compared to that from other methods as Total cost ($/h) 32704.4496 32858.00 33113.00
given in Table II. The result comparison has shown that the

241
 2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June 2013

TABLE IV TABLE V
OPTIMAL DISPATCH AND CORRESPONDING COST OF 140-UNIT SYSTEM WITH OPTIMAL DISPATCH AND CORRESPONDING COST OF 140-UNIT SYSTEM WITH
SMOOTH FUNCTION VALVE-POINT EFFECTS AND PROHIBITED OPERATING ZONES
Unit Pi (MW) Unit Pi (MW) Unit Pi (MW) Unit Pi (MW) Uni Power Unit Power Unit Power Unit Power
1 118.99999 36 500.00000 71 141.42707 106 880.90000 t (MW) (MW) (MW) (MW)
2 164.00000 37 241.00000 72 365.84082 107 873.70000 1 118.99979 36 500.00000 71 141.75588 106 880.90000
3 190.00000 38 241.00000 73 195.00000 108 877.40000 2 164.00000 37 241.00000 72 388.32741 107 873.70000
4 190.00000 39 774.00000 74 239.98337 109 871.70000 3 190.00000 38 241.00000 73 195.00000 108 877.40000
5 189.99959 40 769.00000 75 203.23577 110 864.80000 4 190.00000 39 774.00000 74 194.32658 109 871.70000
6 190.00000 41 3.00000 76 257.22273 111 882.00000 5 168.53982 40 769.00000 75 207.51328 110 864.80000
7 490.00000 42 3.00000 77 397.17964 112 94.00000 6 190.00000 41 3.00000 76 257.52256 111 882.00000
8 490.00000 43 250.00000 78 330.00000 113 94.00000 7 490.00000 42 3.00000 77 393.43812 112 94.00000
9 496.00000 44 250.00000 79 531.00000 114 94.00000 8 490.00000 43 250.00000 78 330.00000 113 94.00000
10 496.00000 45 250.00000 80 531.00000 115 244.00000 9 496.00000 44 250.00000 79 531.00000 114 94.00000
11 496.00000 46 250.00000 81 541.99996 116 244.00000 10 496.00000 45 250.00000 80 531.00000 115 244.00000
12 496.00000 47 249.99590 82 56.00000 117 244.00000 11 496.00000 46 250.00000 81 541.99673 116 244.00000
13 506.00000 48 250.00000 83 115.00000 118 95.00000 12 496.00000 47 250.00000 82 56.00000 117 244.00000
14 509.00000 49 250.00000 84 115.00000 119 95.00000 13 506.00000 48 250.00000 83 115.00000 118 95.00000
15 506.00000 50 250.00000 85 115.00000 120 116.00000 14 509.00000 49 250.00000 84 115.00000 119 95.00000
16 505.00000 51 165.00000 86 207.00000 121 175.00000 15 506.00000 50 250.00000 85 115.00000 120 116.00000
17 506.00000 52 165.00000 87 207.00000 122 2.00000 16 505.00000 51 165.00000 86 207.00000 121 175.00000
18 506.00000 53 165.00000 88 175.00000 123 4.00000 17 506.00000 52 165.00000 87 207.00000 122 2.00000
19 505.00000 54 165.00000 89 175.00000 124 15.00000 18 506.00000 53 165.00000 88 175.00000 123 4.00000
20 505.00000 55 180.00000 90 180.50629 125 9.00000 19 505.00000 54 165.00000 89 175.00000 124 15.00000
21 505.00000 56 180.00000 91 175.00000 126 12.00000 20 505.00000 55 180.00000 90 179.69848 125 9.00000
22 505.00000 57 103.00000 92 575.40000 127 10.00000 21 505.00000 56 180.00000 91 175.00000 126 12.00000
23 505.00000 58 198.00000 93 547.50000 128 112.00000 22 505.00000 57 103.00000 92 575.40000 127 10.00000
24 505.00000 59 312.00000 94 836.80000 129 4.00000 23 505.00000 58 198.00000 93 547.50000 128 112.00000
25 537.00000 60 308.40638 95 837.50000 130 5.00000 24 505.00000 59 312.00000 94 836.80000 129 4.00000
26 537.00000 61 163.00000 96 682.00000 131 5.00000 25 537.00000 60 308.21078 95 837.50000 130 5.00000
27 549.00000 62 95.00000 97 720.00000 132 50.00000 26 537.00000 61 163.00000 96 682.00000 131 5.00000
28 549.00000 63 510.99951 98 718.00000 133 5.00000 27 549.00000 62 95.00000 97 720.00000 132 50.00000
29 501.00000 64 510.99988 99 720.00000 134 42.00000 28 549.00000 63 510.99880 98 718.00000 133 5.00000
30 499.00000 65 490.00000 100 964.00000 135 42.00000 29 501.00000 64 511.00000 99 720.00000 134 42.00000
31 506.00000 66 256.51255 101 958.00000 136 41.00000 30 499.00000 65 490.00000 100 964.00000 135 42.00000
32 506.00000 67 490.00000 102 947.90000 137 17.00000 31 506.00000 66 256.13225 101 958.00000 136 41.00000
33 506.00000 68 490.00000 103 934.00000 138 7.00000 32 506.00000 67 490.00000 102 947.90000 137 17.00000
34 506.00000 69 130.00000 104 935.00000 139 7.00000 33 506.00000 68 490.00000 103 934.00000 138 7.00000
35 500.00000 70 294.59056 105 876.50000 140 26.00000 34 506.00000 69 130.00000 104 935.00000 139 7.00000
Total power (MW) 49342.000 35 500.00000 70 339.43951 105 876.50000 140 26.00000
Min cost ($/h) 1655679.485 Total power (MW) 49342
Max cost ($/h) 1655780.272 Min cost ($/h) 1657962.724
Mean cost ($/h) 1655689.721 Max cost ($/h) 1658166.802
Standard deviation ($/h) 27.07533 Mean cost ($/h) 1658001.450
Standard deviation ($/h) 47.51134397
Case study 3: The Korean power system 140-unit with valve- TABLE VI
point effects, prohibited operating zones, and ramp rate RESULT COMPARISON OF 140-UNIT SYSTEM
limits Method
ED with smooth ED with valve-point effects,
The test system is from [40] with the total demand is 49342 function prohibited operating zones
MW. The power los is neglected in this system. The CTPSO[36] 1655685 1657962.73
CSPSO[36] 1655685 1657962.73
parameters of DEHS for this system are set to HMS = 50, COPSO[36] 1655685 1657962.73
MaxImp = 20000, HCMR = 0.95, PAR = 0.45, cr = 0.3, and w CCPSO[36] 1655685 1657962.73
= 0.55. DEHS 1655679 1657962.72
Table IV and Table V list the optimum dispatch of
generators, total power, minimum, mean, maximum cost V. CONCLUSION
obtained by the proposed DEHS method for two cases
including smooth function and nonsmooth function with In this paper, the hybrid differential evolution and harmony
valve-point effects, prohibited operating zones, and ramp rate search (DEHS) algorithm has been presented to solve the non-
limits. The best total costs obtained by the proposed DEHS convex economic dispatch problems. In the new improved
method are better than those obtained by CTPSO [36], CSPSO method, the conventional DEHS algorithm is used with the
[36], COPSO [36] and CCPSO [36] as shown in Table VI. variance coefficients to speed up the convergence to the global
Therefore, the proposed DEHS is very effective for solving solution in a fast manner regardless of the shape of the cost
nonconvex ED problem for large-scale systems. function. Numerical results have indicated that the DEHS
algorithm has dominates many other methods in the literature

242
 2013 IEEE 7th International Power Engineering and Optimization Conference (PEOCO2013), Langkawi, Malaysia. 3-4 June 2013

in solving the nonconvex ED problems for all the test cases in Electrical Power and Energy Systems, vol. 32, no. 5, pp. 478-487, Jun.
2010.
terms of total costs. Moreover, the proposed method is very [20] J. S. Dhillon, J. S. Dhillon, and D. P. Kothari, “Economic-emission load
effective for solving large-scale systems. Therefore, the dispatch using binary successive approximation-based evolutionary
proposed DEHS method is favorable for solving nonconvex search,” IET Gener. Transm. Distrib., vol. 3, no. 1, pp. 1-16, Jan. 2009.
ED problems. [21] Xiaobo Liu, Zhihua Cai, Chao Yu, “A Hybrid Harmony Search
Approach Based on Differential Evolution,” Journal of Information &
Computational Science, vol. 8, pp. 1889–1900, 2011.
REFERENCES [22] N. Noman and H. Iba, “Differential evolution for economic load dis-
patch problems,” Electric Power Systems Research, vol. 78, no. 8, pp.
[1] H.T. Yang, P.C. Yang and C.L. Huang. “Evolutionary programming 1322-1331, Aug. 2008.
based economic dispatch for units with non-smooth fuel cost functions,” [23] A. I. Selvakumar and K. Thanushkodi, “Anti-predatory particle swarm
IEEE Trans. Power Syst, vol. 11, no. 1, pp. 112-118, 1996. optimization: Solution to nonconvex economic dispatch problems,”
[2] N. Sinha, R. Chakrabarti and P.K. Chattopadhyay “Evolutionary Electric Power Systems Research, vol. 78, no. 1, pp. 2-10, Jan. 2008.
programming techniques for economic load dispatch,” IEEE Trans. [24] Y.-P. Chen, W.-C. Peng, and M.-C. Jian, “Particle swarm optimization
Evol.Comput, vol. 7, no. 1, pp. 83-94, 2003. with recombination and dynamic linkage discovery,” IEEE Trans. Sys-
[3] D.C. Walters and G. B. Sheble “Genetic algorithm solution of economic tems, Man, and Cybernetics - Part B: Cybernetics, vol. 37, no. 6, pp.
dispatch with valve point loading,” IEEE Trans. Power Syst, vol. 8, no. 1460-1470, Dec. 2007.
3, 1993. [25] C.-C. Kuo, “A novel coding scheme for practical economic dispatch by
[4] W.M. Lin, F.S. Cheng and M.T. Tsay, “Improved tabu search for modified particle swarm approach,” IEEE Trans. Power Systems, vol.
economic dispatch with multiple minima,” IEEE Trans. Power Syst, vol. 23, no. 4, pp. 1825-1835, Nov. 2008.
17, no. 1, pp. 108-112, 2002. [26] K. T. Chaturvedi, M. Pandit, and L. Srivastava, “Self-organizing hierar-
[5] S. Khamsawang, C. Boonseng ans S. Pothiya, “Solving the economic chical particle swarm optimization for nonconvex economic dispatch,”
dispatch problem with tabu search algorithm,” In: Proceeding of the IEEE Trans. Power Systems, vol. 23, no. 3, pp. 1079-1087, Aug. 2008.
IEEE International Conference on Industrial Technology 2002, [27] J. G. Vlachogiannis and K. Y. Lee, “Economic load dispatch - A
Bangkok, Thialand, pp. 274-278, 2002. comparative study on heuristic optimization techniques with an
[6] K.P. Wong, Y.W. Wong, “Genetic and Genetic/Simulated – Annealing improved coordinated aggregation-based PSO,” IEEE Trans. Power
approaches to economic dispatch,” IEE Proc.Gener Transm. Distrib, Systems, vol. 24, no. 2, pp. 991-1001, May 2009.
vol. 141, no. 5, pp. 507-513, 1994. [28] L. dos Santos Coelho and V. C. Mariani, “Combining of chaotic differ-
[7] K.P. Wong 'Solving power system optimization problems using ential evolution and quadratic programming for economic dispatch op-
simulated annealing', Engng Applic. Artif.Intell. vol. 8, no. 6, pp. 665- timization with valve-point effect,” IEEE Trans. Power Systems, vol.
670, 1996. 21, no. 2, pp. 989-996, May 2006.
[8] A. Immanuel Selvakumar and K. Thanushkodi, “A new particle swarm [29] S.-K. Wang, J.-P. Chiou, and C.-W. Liu, “Non-smooth/non-convex
optimization solution to nonconvex economic dispatch problem,” IEEE economic dispatch by a novel hybrid differential evolution algorithm,”
Trans. Power Syst, vol. 22, no. 1, pp. 42-51, 2007. IET Gener. Transm. Distrib., vol. 1, no. 5, pp. 793–803, Jan. 2007.
[9] L.D.S. Coelho, R.C.T. Souza, V.C. Mariani, “Improved differential [30] D.-K. He, F.-L. Wang, and Z.-Z. Mao, “Hybrid genetic algorithm for
evolution approach based on cultural algorithm and diversity measure economic dispatch with valve-point effect,” Electric Power Systems
applied to solve economic load dispatch problems.” Math Comput Simul, Research, vol. 78, no. 4, pp. 626-633, Apr. 2008.
vol. 79, pp. 3136–47, 2009. [31] J. S. Alsumait, J. K. Sykulski, and A. K. Al-Othman, “A hybrid GA–
[10] N. Noman and H. Iba, “Differential evolution for economic load PS–SQP method to solve power system valve-point economic dispatch
dispatch problems.” Electr Power Syst Res, vol. 78, pp. 1322–31, 2008. problems,” Applied Energy, vol. 87, no. 5, pp. 1773-1781, May 2010.
[11] Coelho LDS, Mariani VC. Combining of chaotic differential evolution [32] T. Niknam, “A new fuzzy adaptive hybrid particle swarm optimization
and quadratic programming for economic dispatch optimization with algorithm for non-linear, non-smooth and non-convexeconomic dis-
valve-point effect. IEEE Trans Power Syst., vol. 21, pp. 989–96, 2006. patch problem,” Applied Energy, vol. 87, no. 1, pp. 327-339, Jan. 2010.
[12] S.O. Degertekin, “Optimum design of steel frames using harmony [33] H. Lu, P. Sriyanyong, Y. H. Song, and T. Dillon, “Experimental study of
search algorithm,” Struct Multidiscip Optim, vol. 36, pp. 393–401, 2008. a new hybrid PSO with mutation for economic dispatch with non-
[13] A. Vasebi, M. Fesanghary, and S. M. T. Bathaeea, “Combined heat and smooth cost function,” Electrical Power and Energy Systems, vol. 32,
power economic dispatch by harmony search algorithm”, International no. 9, pp. 921-935, Nov. 2010.
Journal of Electrical Power and Energy Systems, Vol. 29, No. 10, 713- [34] A. Bhattacharya, and P. K. Chattopadhyay, “Hybrid differential evolu-
719, Elsevier, 2007. tion with biogeography-based optimization for solution of economic
[14] L. Wang, Q.K. Pan, M.F. Tasgetiren, “A hybrid harmony search load dispatch,” IEEE Trans. Power Systems, vol. 25, no. 4, pp. 1955-
algorithm for the blocking permutation flow shop scheduling problem,” 1964, Nov. 2010.
Comput Ind Eng, vol. 61, pp. 76–83, 2011. [35] P. Subbaraj, R. Rengaraj, and S. Salivahanan, “Enhancement of self-
[15] L.D.S. Coelho, V.C. Mariani, “An improved harmony search algorithm adaptive real-coded genetic algorithm using Taguchimethod for eco-
for power economic load dispatch,” Energy Convers Manage, vol. 50, nomic dispatch problem,” Applied Soft Computing, vol. 11, no. 1, pp.
pp. 2522–2526. 83-92, Jan. 2011.
[16] V.R. Pandi, B.K. Panigrahi, M.K. Mallick, A. Abraham, S. Das [36] Jong-Bae Park, Yun-Won Jeong, Joong-Rin Shin, and Kwang Y. Lee,
“Improved harmony search for economic power dispatch.” In: 2009 “An Improved Particle Swarm Optimization for Nonconvex Economic
Ninth international conference on hybrid intelligent systems; 2009. p. Dispatch Problems,” IEEE Trans. Power Systems, vol. 25, No. 1,
403–8. February 2010.
[17] D. K. Thanushkodi, S. Muthu Vijaya Pandian, R.S.Dhivy Apragash, M. [37] Z. L. Gaing, “Particle swarm optimization to solving the economic
Jothikumar, S.sriramnivas and K.Vindoh, “An Efficient Particle Swarm dispatch considering the generator constraints,” IEEE Trans. Power
Optimization for Economic Dispatch Problems With Non-smooth cost Syst., vol. 18, no. 3, pp. 1187–1195, Aug. 2003.
functions,” WSEAS Transactions on Power Systems, Issue 4, Volume 3,
pp. 257-266, April 2008.
[18] S. Hemamalini and S. P. Simon, “Artificial bee colony algorithm for
economic load dispatch problem with non-smooth costfunctions,”
Electric Power Components and Systems, vol. 38, no. 7, pp. 786-803,
Jul. 2010.
[19] S. Pothiya, I. Ngamroo, and W. Kongprawechnon, “Antcolony optimi-
sation for economic dispatch problem with non-smooth cost functions,”

243