Electrical Power and Energy Systems 77 (2016) 228–249

Contents lists available at ScienceDirect

Electrical Power and Energy Systems

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

Implementation of hybrid harmony search/random search

algorithm for single area unit commitment problem
Vikram Kumar Kamboj a,⇑, S.K. Bath b, J.S. Dhillon c
a
Department of Electrical Engineering, I.K. Gujral Punjab Technical University, Kapurthala, India
b
Department of Electrical Engineering, Giani Zail Singh Campus College of Engineering & Technology, Bathinda, Punjab, India
c
Department of Electrical & Instrumentation Engg., Sant Longowal Institute of Engineering & Technology, Longowal, Punjab, India

a r t i c l e i n f o a b s t r a c t

Article history: Harmony Search (HS) is a population based metaheuristics search algorithm inspired from the musical
Received 21 July 2015 process of searching for a perfect state of harmony and has ability to escape from local minima, does
Received in revised form 6 October 2015 not require differential gradients and initial value setting for the variables and free from divergence
Accepted 17 November 2015
and has strong ability on exploring the regions of solution space in a reasonable time. However, it has
lower exploitation ability in later period and it easily trapped into local optima and converges very
slowly. To improve the exploitation ability of HS algorithm in later stage and provide global optimal solu-
Keywords:
tion, a memetic algorithm approach considering Harmony Search and Random Search is presented in the
Hybrid Harmony Search (HHS)
Pattern Search (PS)
proposed research to solve unit commitment problem of electric power system. The proposed memetic
Random Search (RS) algorithm is tested for IEEE benchmarks consisting of 4, 10, 20 and 40 generating units. The effectiveness
Unit Commitment Problem (UCP) of proposed hybrid algorithm is tested for unimodal and multimodal benchmark functions and is com-
pared with others well known evolutionary, heuristics and meta-heuristics search algorithms and it
has been found that performance of proposed hybrid algorithm is much better than classical Harmony
Search Algorithm and Improved Harmony Search Algorithm as well as recently developed algorithms.
Ó 2015 Elsevier Ltd. All rights reserved.

Introduction as unit commitment (UC). The unit which is decided or scheduled

Today’s power system is characterized by large proportions,
high interconnections and high nonlinearities, as the size of the
power system is growing exponentially due to heavy demand of
power in all the sectors viz. agricultural, industrial, residential
and commercial ones. Increase in the electrical energy demand
and trends in privatization and deregulation result in overloading
impact on electrical grids. The situation necessitates the develop-
ment of electrical grid at the same pace as the demand increases,
but economical commitment and scheduling has the ability to
tackle the time-varying power demand, environmental constraints
and led to the full exploitation of accessible grid. In the modern
power system networks, there are various generating resources
like thermal, hydro, nuclear, etc. Also, the load demand varies dur-
ing a day and attains different peak values. Thus, it is required to
decide which generating unit to turn on and at what time it is
needed in the power system network and also the sequence in
which the units must be shut down keeping in mind the cost effec-
tiveness of turning on and shutting down of respective units. The
entire process of computing and making these decisions is known
⇑ Corresponding author.
to be connected to the power system network, as and when
required, is known to be committed unit. Unit commitment in
power systems refers to the problem of determining the on/off
states of generating units to minimize the operating cost for a
given time horizon [1]. Generators cannot be immediately turned
on to meet up power demand. So it is required that the planning
of generating units must be so prepared that there is enough gen-
eration available to fulfil the load demand along with an ample
reserve generation to avoid failures and malfunctions under
adverse conditions. Unit commitment knob the unit generation
schedule in electric power system for minimizing operational
and fuel cost and satisfying system and physical constraints such
as load demand and system reserve requirements over a set of time
periods [2]. Unit Commitment Problem (UCP) is basically about
finding the most suitable schedule to turn-on or turn-off the gen-
erating units to meet the electric power demand and at the same
time keep the cost of generation as much minimum as possible.
UCP is a non-linear, large scale, mixed integer constrained opti-
mization problem [3] and happens to belong to combinatorial opti-
mization problems. There are many constraints involved in UCP
and hence it is quite a complex and tedious task to compute or
to find the optimal solution for Unit Commitment Problem (UCP).

http://dx.doi.org/10.1016/j.ijepes.2015.11.045
0142-0615/Ó 2015 Elsevier Ltd. All rights reserved.
 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249
Unit commitment problem
The scheduling of the units together with the allocation of the
generation quantities which must be scheduled to meet the
demand for a specific period represents the Unit Commitment
Problem (UCP). The Unit Commitment Problem is to determine a
smallest cost turn-on and turn-off plan of a set of generating units
to meet a power demand while satisfying system operational and
physical constraints liked with various generating units. The pro-
duction cost includes fuel, startup, no load costs and shutdown
cost. The operational constraints that must be taken into consider-
ation comprises, 1. The total power generated must meet the
power demand plus system losses. 2. There must be an adequate
amount of spinning reserve to cover any shortfalls in power gener-
ation. 3. The loading of each unit must be within its minimum and
maximum permissible rating. 4. The minimum up and down times
of each unit must be pragmatic. The unit commitment is aimed to
formulate a proper generator commitment schedule for electric
power system over a period of one day to one week. The main
objective of unit commitment is to minimize the total production
cost over the study period & to satisfy the system and physical con-
straints imposed on the system such as spinning reserve, power
generation-load balance, operating constraints, and minimum up
time & minimum down time. Several conventional methods are
available to solve the unit commitment problem. But all these
methods need the exact mathematical model of the system & there
may be a chance of getting stuck at the local optimum [4]. Sriyany-
ong and Song [5] proposed Particle Swarm Optimization combined
with Lagrange Relaxation method for solving UCP. The proposed
approach employs PSO algorithm for optimal settings of Lagrange
multipliers. The feasibility of the proposed method is demon-
strated for 4 and 10 unit systems, respectively. Xiong et al. [6] have
applied multi particle swarm to parallel arithmetic to produce par-
ticle to enhance the convergence speed and found the more effi-
cient results than genetic algorithm. Jeong et al. [7] have
discussed binary Particle Swarm optimization-based approach for
solving the UC problems. Ge [8] has proposed a new approach to
solve ramp rate constrained unit commitment problem by improv-
ing the method of Particle Swarm Optimization. Borghetti et al. [9]
have suggested that there is no guarantee that the Tabu search will
yield the global optimal result for large systems. There is a similar
method named Particle Swarm Optimization proposed in [10].
Rajan et al. [11] proposed Neural based Tabu search algorithm
for the unit commitment problem and developed an improved ver-
sion of Neural based Tabu search approach [3]. Gaing [12] pro-
posed binary particle swarm optimization (BPSO). The BPSO is
used to solve the combinatorial unit on/off scheduling problem
for operating fuel and transition costs. The ED subproblem is
solved using the lambda iteration method for obtaining the total
production cost. Zhao et al. [13] presented an improved particle
swarm optimization algorithm (IPSO) for UC which utilizes more
particles information to control the process of mutation operation.
For proper selection of parameters some new rules are also pro-
posed. The proposed method combines LR technique to 0–1 vari-
able. Lee and Chen [14] presented a new approach for UCP
named the iteration particle swarm optimization (IPSO). The pro-
posed method improves the quality of solution in terms of total
production cost and also improves the computation efficiency. A
standard 48 unit system has been tested for validation. Samudi
et al. [15] have presents a new approach of particle swarm opti-
mization (PSO) algorithm for short term hydro thermal scheduling
(HTS) problems. The proposed algorithm is ideally suitable for
hydro-thermal co-ordination problems, hydro economic dispatch
problems with unit commitment, thermal economic dispatch with
unit commitment problems and scheduling of hydraulically cou-
229
pled plants. Yuan et al. [16] proposed a new improved binary
PSO (IBPSO). The standard PSO is improved the using the priority
list and heuristic search to improve the MUT and MDT constraints.
The 10–100 units have been tested to validate the proposed
approach. Numerical performance shows that the proposed
approach is superior in terms of low total production cost and
short computational time compared with other published results.
No optimization algorithm can perform general enough to solve
all optimizations problems, each optimization algorithm have their
own advantages and disadvantages. Particle swarm optimization
(PSO) has simple concept, easy implementation, relative robust-
ness to control parameters and computational efficiency [17,18],
although it has numerous advantages, it get trapped in a local min-
imum, when handling heavily constrained problems due to the
limited local/global searching capabilities [19,20]. The limitations
of the numerical techniques [21,22] and dynamic programming
method [23,24] are the size or dimensions of the problem, large
computational time and complexity in programming. The mixed
integer programming methods [25,26] for solving the economic
load dispatch problem fails when the participation of number of
units increases because they require a large memory and suffer
from great computational delay. Gradient Descent method [27] is
distracted for Non-Differentiable search spaces. The Lagrangian
Relaxation (LR) approach [28] fails to obtain solution feasibility
and solution quality of problems and becomes complex if the num-
ber of units are more. The Branch and Bound (BB) method [29]
employs a linear function to represent fuel cost, start-up cost and
obtains a lower and upper bounds. The difficulty of this method
is the exponential growth in the execution time for systems of a
large practical size. An Expert System (ES) algorithm [30] rectifies
the complexity in calculations and saving in computation time. But
it faces the problem if the new schedule is differing from schedule
in database. The fuzzy theory method [31] using fuzzy set solves
the forecasted load schedules error but it suffers from complexity.
The Hopfield neural network technique [32] considers more con-
straints but it may suffer from numerical convergence due to its
training process. The Simulated Annealing (SA) [33] and Tabu
Search (TS) [34] are powerful, general-purpose stochastic opti-
mization technique, which can theoretically converge asymptoti-
cally to a global optimum solution with probability one. But it
takes much time to reach the near-global minimum. Gravitational
Search algorithm has the advantages to explore better optimized
results, but due to the cumulative effect of the fitness function
on mass, masses get heavier and heavier over the course of itera-
tion. This causes masses to remain in close proximity and neu-
tralise the gravitational forces of each other in later iterations,
preventing them from rapidly exploiting the optimum [35]. There-
fore, increasing effect of the cost function on mass, masses get
greater over the course of iteration and search process and conver-
gence becomes slow.
Motivation for proposed hybrid harmony search algorithm
The Harmony Search (HS) algorithm proposed by Geem et al.
[36] is recently developed metaheuristics search algorithm
inspired from the musical process of searching for a perfect state
of harmony, HS has a novel stochastic derivative [37] applied to
discrete variables, which uses musician’s experiences as a search-
ing direction and is free from divergence. It can handle discrete
and continuous variables and do not require initial value setting
for the variables. Also, it does not require differential gradients
and has the ability to escape from local optima. HS has ability to
overcome the drawback of GA’s building block theory and explic-
itly considers the relationship using ensemble operation [38].
Geem et al. [39] proposed a Multi-pitch Adjusting Rate (multiple
230 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249

PAR) for Generalized Orienteering Problem. They proposed three

PAR’s that are the rates of moving to nearest, second nearest, and
third nearest cities, respectively. Geem [40] presented the use of
fixed parameter values, such as HMS, HMCR, PAR, and NI, while
bw was set to a range from 1% to 10% of the total value data range.
Mahdavi et al. [41] proposed Improved Harmony Search (IHS) algo-
rithm, which includes dynamic adaptation for both pitch adjust-
ment rate (PAR) and bandwidth (bw) values. But it faces the
difficulty of determining the lower and upper bound of automatic
bandwidth (bw), which was overcome by Global-best harmony
search (GHS) algorithm proposed by Omran and Mahdavi [42].
GHS algorithm incorporate the PSO concept, global best particle,
by replacing the bw parameter altogether and adding a randomly
selected decision variables from the best harmony vector in HM.
Mukhopadhyay et al. [43] suggested that bw will be the standard Fig. 3. Pitch adjustment rate.
deviation of the current population when HMCR is close to 1.
Degertekin [44] proposed a new HM initialization technique that
generated two times of HMS initial harmonies but placed only
the best HMS of these into the initial HM. Chakraborty et al. [45]
proposed Differential Harmony Search algorithm, a new improve-
ment to HS through inspiring the Differential Evolution (DE) muta-
tion operator, which replaces the pitch adjustment operation in
classical HS with a mutation strategy borrowed from the DE (DE/
rand/1/bin class) algorithm. Hasancebi et al. [46] and Saka and
Hasancebi [47] proposed a new adaptation for HS by making both
HMCR and PAR change dynamically during the improvisation pro-
cess of HS. This step is to make the selection of these parameter
values problem independent, therefore, improves the performance
of HS in finding an optimal solutions. Kattan et al. [48] used HS as a
new training technique for feed-forward artificial neural networks Fig. 4. Updation of worst harmony with best harmony.

Fig. 1. Random initialization of harmony memory. Fig. 5. Ensemble consideration.

Fig. 2. Harmony memory consideration. Fig. 6. Violated harmony consideration.

V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249 231

Fig. 7. PSEUDO code for harmony search algorithm.

Fig. 8a. Flow-chart for harmony search algorithm.

232 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249

Fig. 8b. PSEUDO code for proposed hybrid harmony search algorithm.

Fig. 8c. Flow chart for handling various UCP constraints.

 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249 233

Convergence of HS-Random Search

5 Algorithm for 10-Units System
Convergence of HS-Random x 10 Convergence of HS-Random
4 Search Algorithm 5.66 5 Search Algorithm(SR=5%)
x 10 x 10
HS-Random Search Algorithm
7.456 5.596
HS-Random Search Algorithm HS-Rand Search Algorithm
7.455 5.594
5.655
7.454 5.592

Generation Cost ($)

Generation Cost ($)

7.453 5.59
5.65

Cost ($)
7.452 5.588

7.451 5.645 5.586

7.45 5.584

7.449 5.64 5.582

7.448 5.58

7.447 5.635 5.578

0 200 400 600 800 1000 20 40 60 80 100 120 140 0 50 100 150 200 250
Iterations-----> Iterations------> Iterations------>
(a) 4-Unit Test System (b) 10-Unit Test System (SR=10%) (c) 10-Unit Test System (SR=5%)

Fig. 9. Convergence of DE-random search algorithm for 4 and 10-generating unit test system.

Convergence of HS-Random Search Convergence of HS-Random Search

Algorithm for 20-Units Test System Algorithm for 40-Units Test System
6 6
x 10 x 10
1.138 2.33
HS-Random Search Algorithm HS-Random Search Algorithm
2.32
1.136

2.31
1.134
Generation Cost ($)

Generation Cost ($)

2.3

1.132 2.29

1.13 2.28

2.27
1.128
2.26

1.126
2.25

1.124 2.24
0 50 100 150 0 50 100 150 200 250 300
Iterations-----> Iterations----->
(a) 20-Unit Test System (b) 40-Unit Test System
Fig. 10. Convergence of HS-random search algorithm for 20 and 40-generating unit system (for 10% spinning reserve).

Load Demand Pattern for 10, 20 and 40-Units Test System

6000 Table 1
20 Units Committed status and generation scheduling of 4-unit test system.
10 Units
40 Units Commitment
5000
Hour Status Generaon Schedule
U1 U2 U3 U4 U1 U2 U3 U4
Load Demand (MW)

4000 1 1 1 0 0 300 150 0 0

2 1 1 1 0 300 205 25 0
3000 3 1 1 1 1 300 250 30 20
4 1 1 1 0 300 215 25 0
5 1 0 1 1 300 0 80 20
2000
6 1 0 1 0 255 0 25 0
7 1 0 1 0 265 0 25 0
1000
8 1 1 0 0 300 200 0 0
Total Cost($) 74476.00
0 Time(Sec.) 22.863569
0 5 10 15 20 25
Hour-------> The bold results show the superiority of proposed Hybrid Harmony
search algorithm over others well known heuristics, meta-heuristics
Fig. 11. Load demand pattern for 10, 20 and 40-unit test system. and evolutionary algorithms.
234 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249

Table 2
Comparison of results for 4-generating unit system.

Method Generation cost ($) Iteration time (s)

Best Average Worst Best Average Worst
Improved Lagrangian Relaxation (ILR) [92] 75231.9 NA NA – – –
B. SMP [74] 74,812 74,877 75,166
A.SMP [74] 74,812 74,877 75,166
Lagrangian Relaxation and PSO (LRPSO) [92] 74,808 NA NA – – –
Binary Differential Evolution (BDE) [73] 74,676 NA NA – – –
Genetic Algorithm (GA) [60] 74,675 NA NA – – –
Hybrid HS and Random Search Algorithm [Proposed Algorithm] 74,476 74,476 74,476 20.68704 22.86357 22.9709

The bold results show the superiority of proposed Hybrid Harmony search algorithm over others well known heuristics, meta-heuristics and evolutionary algorithms.

(ANN). Wang and Huang [49] proposed a new variation of HS algo-
rithm that focuses on the dynamic selection of bw and PAR param-
eters. Al-Betar et al. [50] also proposed a Multi-pitch Adjusting
Rate strategy for enhancing the performance of HS in solving
course timetabling problem. They proposed eight procedures
instead of using one PAR value, each of which is controlled by its
PAR value range. Each pitch adjustment procedure is responsible
for a particular local change in the new harmony. Furthermore,
the acceptance rule for each pitch adjustment procedure is chan-
ged to accept the adjustment that leads to a better or equal objec-
tive function. Even though, HS has the ability to escape from local
minima, does not require differential gradients and initial value
setting for the variables and free from divergence and has strong
ability on exploring the regions of solution space in a reasonable
time. However, it has lower exploitation ability in later period
and it easily trapped into local optima and converges very slowly.
To improve the exploitation ability of HS algorithm in later stage
and provide global optimal solution. A memetic algorithm
approach combining Harmony Search (HS) and Random Search
(RS) algorithm is tested for unimodal and multimodal benchmark
functions and used to solve the unit commitment problem of elec-
tric power system.
Table 3
Committed status of 10-unit test system (with 10% spinning reserve).
Unit commitment problem formulation
The foremost objective of unit commitment is to find the opti-
mal schedule for operating the available generating units to regu-
late the total operating and generation cost of electric power
utilities. Total operating cost of power generation includes fuel
cost, shut down and start up costs. The fuel costs are calculated
using the data of generating unit characteristics such as fuel price
information, heat rate of generating utilities, turn-on, turn-off and
initial status of units, which is mathematically, a quadratic, non-
smooth and non-convex equation of power output of each genera-
tor at each hour and can be determined by Economic Load Dispatch
(ELD) [20], as represented below:
F cost ðPi Þ ¼ ai P2i þ bi Pi þ ci ; ð1Þ
where ai , bi and ci are the fuel cost coefficients of ith generating
units.
The total fuel cost over the given time horizon ‘H’ is
NG h
H X
X i
TFC ¼ ðai P 2i þ bi P i þ ci Þ  U ih þ SUC ih  ð1  U iðh1Þ Þ  U ih ; ð2Þ
h¼1 i¼1
where U ih is the position or status of ith unit at hth hour. Start up
cost is warmth-dependent. Start up cost is that cost which occurs
while bringing the thermal generating unit online. It is expressed
in terms of the time (in hours) for which the units have been shut

Table 4
Generation scheduling of 10-unit test system (for 10% spinning reserve).

The bold results show the superiority of proposed Hybrid Harmony search algo-
rithm over others well known heuristics, meta-heuristics and evolutionary The bold results show the superiority of proposed Hybrid Harmony search algorithm
algorithms. over others well known heuristics, meta-heuristics and evolutionary algorithms.
 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249 235

down. On the other hand, shut down cost is a fixed amount for each The Various Constraints linked with Unit Commitment Problem
unit which is shut down. Mathematically, start up cost can be are mentioned below:
expressed as:
Load balance or power balance constraints
(
HSC i ; for MDT i 6 MDT ON
i 6 ðMDT i þ CSHi Þ
SUC ih ¼ ði 2 G; h ¼ 1; 2; 3; . . . ; HÞ; ð3Þ
CSC i ; for MDT ON
i > ðMDT i þ CSHi Þ The load balance or system power balance constraint requires
that the sum of generation of all the committed units at hth hour
where CSC i and HSC i are cold startup and hot start-up cost of ith must be greater than or equal to the demand at a particular hour ‘h’.
unit respectively and MDT i is the minimum down time of ith unit,
X
NG
MDT ON
i is the number of hours that ith unit has been on-line since Pih U ih ¼ Dh : ð4Þ
it was turned ON earlier and CSHi is the cold start hour of unit i. i¼1

Table 5
Comparison of results for 10-generating unit system (for 10% spinning reserve).

Sr. no. Method Overall generation cost ($) Average time (s)
Best Average Worst
1 Genetic Based Method [65] NA 623,441 – –
2 Hybrid Continuous Relaxation and Genetic Algorithm (CRGA) [52] NA 563,977 – 46
3 Continuous Relaxation and Genetic Algorithm (CRGA) [81] – 563,977 – –
4 Integer Coded Genetic Algorithm (ICGA) [94] – 566,404 – –
5 Lagrangian Search Genetic Algorithm (LSGA) [54] 609023.69 – – –
6 Improved Binary Particle Swarm optimization (IBPSO) [16] 599,782 – – 14.48
7 New Genetic Algorithm [90] 591,715 – – 677
8 PSO [55] 581,450 – – –
9 Binary Particle Swarm Optimization with bit Change Mutation (MPSO) [59] 574,905 – – 15.73
10 HPSO [91] 574,153 – – –
11 LCA-PSO [58] 570,006 – – 18.34
12 Two-Stage Genetic Based Technique (TSGA) [83] 568,315 – – –
13 Hybrid PSO-SQP [66] 568032.3 – – –
14 BCGA [94] 567,367 – – –
15 SM [93] 566,686 566,787 567,022 –
16 LR [93] 566,107 566,493 566,817 –
17 GA [93] 565,866 567,329 571,336 –
18 Genetic Algorithm (GA) [88] 565,852 – 570,032 221
19 Enhanced Simulated Annealing (ESA) [61] 565,828 565,988 566,260 3.35
20 Lagrangian Relaxation (LR) [88] 565,825 – – –
21 Dynamic Programming (DP) [88] 565,825 – – –
22 Improved Lagrangian Relaxation (ILR) [92] 565823.23 – – –
23 LRPSO [92] 565275.2 – – –
24 Lagrangian Relaxation and Genetic Algorithm (LRGA) [95] 564,800 564,800 – 518
25 Evolutionary Programming (EP) [89] 564,551 565,352 – 5.61
26 EP [93] 564,551 565,352 566,231 100
27 Particle Swarm Optimization (PSO) [67] 564,212 565,103 565,783 –
28 Ant Colony Search Algorithm (ACSA) [71] 564,049 – – –
29 Hybrid Ant System/Priority List (HASP) [80] 564,029 564,324 564,490 –
30 B. SMP [74] 564017.73 564121.46 564401.08 –
31 Annealing Genetic Algorithm (AGA) [96] 564,005 – – –
32 Binary Differential Evolution [73] 563,997 563,997 563,997 –
33 Social Evolutionary Programming (SEP) [56] 563,987 – – –
34 Methodological Priority List (MPL) [82] 563977.1 – – –
35 Genetic Algorithm (GA) [88] 563,977 564,275 5,665,606 221
36 IBPSO [16] 563,977 564,155 565,312 –
37 Genetic Algorithm Based on Unit Characteristics (UCC-GA) [51] 563,977 – 565,606 85
38 Enhanced Adaptive Lagrangian Relaxation (EALR) [53] 563,977 – – 4
39 Local Search Method (LCM) [57] 563,977 – – 0.6
40 Quantum-Inspired Binary PSO (QBPSO) [63] 563,977 – – –
41 Binary PSO [68] 563,977 563,977 563,977 –
42 Quantum-Inspired Binary PSO (QIBPSO) [68] 563,977 563,977 563,977 –
43 Extended Priority List (EPL) [70] 563,977 – – –
44 Muller Method [85] 563,977 – – 0.516
45 Improved Particle Swarm Optimization (IPSO) [67] 563,954 564,162 564,579 –
46 Advanced Fuzzy Controlled Binary PSO (AFCBPSO) [86] 563,947 564,285 565,002 5.54
47 Hybrid PSO (HPSO) [97] 563942.3 564772.3 565782.3 –
48 Fuzzy Quantum Computation Based Thermal Unit Commitment (FQEA) [64] 563,942 – – –
49 Advanced Quantum-Inspired Evolutionary Algorithm (AQEA) [62] 563,938 – – –
50 Particle Swarm-Based-Simulated Annealing (PSO-B-SA) [69] 563,938 564,115 564,985 –
51 QEA-UC [62] 563,938 564,012 564,711
52 IQEA-UC [62] 563,938 563,938 563,938
53 Gravitational Search Algorithm [84] 563,938 564,008 564,241 2.89
54 A.SMP [74] 563937.26 564040.3 564320.61 –
55 Harmony Search (HS) [72] 564367.69
56 Harmony Search Algorithm (HAS) [87] 563,977 564168.6 – 3.00
57 Hybrid HS-Random Search [Proposed Method] 563937.6875 563965.31 563995.33 16.831236

The bold results show the superiority of proposed Hybrid Harmony search algorithm over others well known heuristics, meta-heuristics and evolutionary algorithms.
236 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249

Spinning reserve constraints load demand. This excess capacity of generation is known as Spin-
ning Reserve and mathematically given as:
Considering the important aspect of reliability, there is a provi-
X
N
sion of excess capacity of generation which is required to act ðP max ÞU ih P Dh þ Rh : ð5Þ
i
instantly when there is a failure of already running unit or sudden i¼1

Table 6
Committed status of 10-unit test system (with 5% spinning reserve).

Table 7
Generation scheduling of 10-unit test system (for 5% spinning reserve).
 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249 237

Thermal constraints Minimum down time

If all the units are already running, they cannot be shut-down
A thermal generation unit needs to undergo gradual tempera- simultaneously. This constraint is given as:
ture changes and thus it takes some period of time to bring a ther-
mal unit online. Also, the operation of a thermal unit is manually
X off
i ðtÞ P MDT i ; ð7Þ
controlled. So a crew is required to perform the operation and where X off
i ðtÞ is duration for which unit i is continuously OFF (in hrs)
maintenance of any thermal unit. This leads to many restrictions
and MDT i minimum down time (in hrs).
in the operation of thermal unit and thus it gives rise to many
constraints.
Crew constraints
If a plant consists of two or more units, they cannot be turned
Minimum up time on at the same time since there are not enough crew members
If the units have already been shut down, there will be a mini- to attend both units while starting up.
mum time before they can be restarted. This constraint in given as:

X on Maximum and minimum power limits

i ðtÞ P MUT i ; ð6Þ
Every unit has its own maximum/minimum power level of gen-
where X on
i ðtÞ is duration for which unit i is continuously ON (in hrs)
eration, beyond and below which it cannot generate.
and MUT i is unit i minimum up time (in hrs).

Table 8
Comparison of results for 10-generating unit system (for 5% spinning reserve).

Method Overall generation cost ($)

Best cost ($) Average cost ($) Worst cost ($)
BPSO [75] 565,804 566,992 567,251
GA [75] 570,781 574,280 576,791
APSO [76] 561,586 – –
BP [76] 565,450 – –
TSGB [77] 560263.92 – –
IPSO [78] 558114.80 – –
Hybrid PSO-SQP [79] 568032.30 – –
B.SMP [74] 558844.76 558937.24 559154.98
Hybrid HS-Random Search algorithm 557905.6427 558267.2 558682.0107

The bold results show the superiority of proposed Hybrid Harmony search algorithm over others well known heuristics, meta-heuristics and evolutionary algorithms.

Table 9
Committed status of 20-unit test system.
238 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249

Pmin
i 6 Pih 6 Pmax
i : ð8Þ solution, then the possibility to make a good solution is increased
next time. Random Search Algorithm is derivative free method for
continuous domain, which is based on direct search and most suit-
Initial operating status of generating units able for Stochastic and Global optimization problem. In the pro-
The initial operating status of every unit should take the last posed algorithm HS is combined with Random Search algorithm
day’s previous schedule into account, so that every unit satisfies for random population search. The major steps of proposed hybrid
it’s minimum up/down time. algorithm are mentioned below:

Hybrid harmony search-random search algorithm  Initialization of Harmony Memory (HM)

 Harmony Memory Considering (HMC) Rule
Harmony Search (HS) is a population based meta-heuristics  Pitch Adjusting Rate (PAR)
search algorithm inspired from the musical process of searching  Random Initialization Rule
for a perfect state of harmony. HS has been proposed by Geem  Harmony Memory Updating
et al. in 2001. The pitch of each musical instrument determines  Ensemble Consideration
the aesthetic quality, just as the fitness function value determines  Violated Harmony Consideration
the quality of decision variables. In the musical improvisation pro-  Randomization
cess, all players sound pitches within possible range together to
make one harmony. If all the pitches make a good harmony, each Step-I: initialization of Harmony Memory (HM)
player stores in his memory that experience and the possibility
of making a good harmony is increased next time. The same thing The initial population HM contains of HMS vectors is generated
in optimization, the initial solution is generated randomly from randomly. The Harmony Memory (HM) matrix is filled with HMS
decision variables within the possible range. If the objective func- vectors as follows:
tion values of these decision variables is good to make a promising

Table 10
Generation scheduling of 20-unit test system (for 10% spinning reserve).
 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249 239

2 3
p11 p12 p13 ... p1G Step-IV: random initialization rule
6 p ... p2G 7
6 21 p22 p23 7
6 7 If the condition r1 < HMCR fails, the new first decision variable
HM ¼ 6 .. .. .. .. .. 7 ; ð9Þ
6 . . . . . 7
4 5 in the new vector pNewij is generated randomly as follows:
..
pHMS1 pHMS2 pHMS3 . pHMSG HMSG
pNew
ij ¼ Pmin
ij þ ðPmax
ij  Pmin
ij Þ  randð0; 1Þ; ð12Þ
where pi ¼ pij ; i 2 f1; 2; 3; . . . ; HMSg and j 2 f1; 2; 3; . . . ; Gg (see
where Pmin and Pmax are lower and upper bounds for generating
Fig. 1). ij ij
units and randð0; 1Þ is random vector within the range [0, 1].
Step-II: Harmony Memory Considering (HMC) rule
Step-V: harmony memory updating
For this rule, a new random number r1 is generated within the
range [0, 1]. After the Harmony Vector pNew
ij is generated, it will replace the
If r1 < HMCR, then the first decision variable in the new vector worst harmony vector pWorst
ij in the Harmony memory if its objec-
pNew
ij is chosen randomly from the values in the current HM as tive function value is better than the objective function value of
follows: the worst harmony vector. PSEUDO code for updation of Worst
  Harmony Vector (WHV) with new random harmony vector is men-
pNew
ij ¼ pij ; pij 2 p1j ; p2j ; p3j ; . . . ; pHMSj ; ð10Þ tioned below (see Fig. 4):
where HMCR is the Harmony Memory Consideration Rate (see
Fig. 2).  
If pNew
ij < pWorst
ij then
Step-III: Pitch Adjusting Rate (PAR) Update the HM as pWorst ¼ pNew
ij ij
end if
The obtained decision variables from the harmony memory
consideration rule is further examined to determine if it needs to
pitch adjustment or not, which can be mathematically calculated
as:
( Step-VI: ensemble consideration
pij  randð0; 1Þ  BW; with probability PAR
pNew ¼ ;
ij
pij ; with probability ð1-PARÞ After the new harmony pNewij ¼ pij ; pij 2 fp1j ; p2j ; p3j ; . . . ; pHMSj g is
obtained, one more operation can be considered from the relation-
ð11Þ
ship among decision variables. Just as a player has even stronger
where PAR is Pitch Adjustment Rate and BW is a bandwidth factor, relationship with specific player in a music group, the new opera-
which is used to control the local search around the selected deci- tion, ensemble consideration [40], enables the algorithm to com-
sion variable in the new vector (see Fig. 3). bine closely related variables together.

Table 11
Comparison of results for 20-generating unit system (for 10% spinning reserve).

Sr. no. Method Overall generation cost ($) Average time (s)
Best cost Average cost Worst cost
1 Genetic Based Method [65] – 1,215,066 –
2 New Genetic Algorithm [90] – 1,133,786 – 1095
3 Hybrid Continuous Relaxation and Genetic Algorithm (CRGA) [52] – 1,236,981 – 98
4 Lagrangian Relaxation and Genetic Algorithm (LRGA) [95] – 1,122,622 – 1147
5 Integer Coded Genetic Algorithm (ICGA) [94] – 1,127,244 – –
6 Annealing Genetic Algorithm (AGA) [96] – 1,124,651 – –
7 Improved Binary Particle Swarm optimization (IBPSO) [16] 1,196,029 – – 60.65
8 Binary Particle Swarm Optimization with bit Change Mutation (MPSO) [59] 1,152,966 – – 65.87
9 LCA-PSO [58] 1,139,005 – – 57.32
10 Lagrangian Relaxation (LR) [88] 1,130,660 – – –
11 BCGA [94] 1,130,291 – – –
12 GA [93] 1,128,876 1,130,160 1,131,565 –
13 LR [93] 1,128,362 1,128,395 1,128,444 –
14 SM [93] 1,128,192 1,128,213 1,128,403 –
15 DP and Lagrangian Relaxation (DPLR) [53] 1,128,098 – – –
16 Enhanced Simulated Annealing (ESA) [61] 1,126,254 – – –
17 Enhanced Simulated Annealing (ESA) [93] 1,126,251 1,127,955 1,129,112 16.8
18 Genetic Algorithm (GA) [88] 1,126,243 1,132,059 733
19 Particle Swarm Optimization (PSO) [67] 1,125,983 – 1,131,054 –
20 Intelligent Mutation based Genetic Algorithm (UCC-GA) [51] 1,125,516 – 1,128,790 225
21 Genetic Algorithm (GA) [84] 1,125,516 – 1,128,790 733
22 Evolutionary Programming (EP) [89] 1,125,494 1,127,257 – 340
23 EP [93] 1,125,494 1,127,257 1,129,793 –
24 Evolutionary Programming (EP) [89] 1,125,494 – 1,129,793 340
25 Improved Particle Swarm Optimization (IPSO) [67] 1,125,279 – 1,127,643 –
26 Social Evolutionary Programming (SEP) [56] 1,125,170 – – –
27 B.SMP [74] 1,124,838 1,125,102 1,125,283
28 Harmony Search [72] – 1,127,377 – 92
29 Hybrid HS-Random Search Algorithm [proposed method] 1124889.39 1124912.84 1124951.55 35.01579

The bold results show the superiority of proposed Hybrid Harmony search algorithm over others well known heuristics, meta-heuristics and evolutionary algorithms.
240 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249

Step-VII: violated harmony consideration
Once the new harmony is obtained using the above-mentioned
rules, it is then checked whether it violates problem constraints.
Although the new harmony violates the constraints, it has still
chance to be included in HM, just as rule-violated harmony was
still used by musicians such as famous composer Ludwig van
Beethoven [40]. Violated harmony can be considered by adding a
penalty. The suitable penalty can be mathematically described as:
b
Penalty ¼ a0  ðViolationAmountÞ 0  c0 ;
where a0, b0 and c0 are penalty coefficients; a makes the penalty
amount proportional to the violated amount; b0 makes the penalty
amount exponential to the violated amount; and c0 contributes con-
stant amount to the penalty amount in order to prevent any slight
violation (see Figs. 5 and 6).
Randomization in harmony search algorithm
Randomization in Harmony Search algorithm is to increase the
diversity of the solutions. Although the pitch adjustment has a sim-
ilar role, it is limited to certain area and thus corresponds to a local
search. The use of randomization can drive the system further to
explore various diverse solutions so as to attain the global optimal-
ð13Þ ity. The Pseudo code of Proposed Algorithm (Fig. 7). The probability
of randomization is
PRandom ¼ 1  r accept ; ð14Þ

Table 12
Generation scheduling of 40-unit test system (for 10% spinning reserve).

Hour Committed status of generating units for 40-units test system (units 1–20)
U1 U2 U3 U4 U5 U6 U7 U8 U9 U10 U11 U12 U13 U14 U15 U16 U17 U18 U19 U20
1 455 225 0 0 0 0 0 0 0 0 455 150 0 0 0 0 0 0 0 0
2 455 150 0 0 0 0 0 0 0 0 455 455 0 0 0 0 0 0 0 0
3 455 390 0 0 0 0 0 0 0 0 455 150 0 0 0 0 0 0 0 0
4 455 455 0 0 25 0 0 0 0 0 455 410 0 0 25 0 0 0 0 0
5 455 455 0 0 155 0 0 0 0 0 455 455 0 0 25 0 0 0 0 0
6 455 455 130 130 25 0 0 0 0 0 455 455 130 130 35 0 0 0 0 0
7 455 455 130 130 25 0 0 0 0 0 455 455 130 130 105 0 0 0 0 0
8 455 455 130 130 25 0 0 0 0 0 455 455 130 130 45 0 0 0 0 0
9 455 455 130 130 162 20 25 0 0 0 455 455 130 130 162 0 25 0 0 0
10 455 455 130 130 162 20 25 10 10 10 455 455 130 130 162 72 25 10 0 0
11 455 455 130 130 162 52 25 10 10 10 455 455 130 130 162 80 25 10 10 10
12 455 455 130 130 162 80 25 43 10 10 455 455 130 130 162 80 25 43 10 10
13 455 455 130 130 162 20 25 10 10 10 455 455 130 130 162 72 25 10 0 0
14 455 455 130 130 162 20 25 0 0 0 455 455 130 130 162 0 25 0 0 0
15 455 455 130 130 25 0 0 0 0 0 455 455 130 130 45 0 0 0 0 0
16 455 455 130 130 25 0 0 0 0 0 455 150 130 130 25 0 0 0 0 0
17 455 455 130 130 25 0 0 0 0 0 455 150 130 130 25 0 0 0 0 0
18 455 150 130 130 25 0 0 0 0 0 455 455 130 130 25 0 0 0 0 0
19 455 455 130 130 25 0 0 0 0 0 455 455 130 130 45 0 0 0 0 0
20 455 455 130 130 162 20 25 10 10 10 455 455 130 130 162 72 25 10 0 0
21 455 455 130 130 162 20 25 0 0 0 455 455 130 130 114 20 25 0 0 0
22 455 455 0 0 162 20 25 0 0 0 455 455 130 0 126 20 25 0 0 0
23 455 365 0 0 25 0 0 0 0 0 455 455 0 0 0 0 0 0 0 0
24 455 455 0 0 0 0 0 0 0 0 455 320 0 0 0 0 0 0 0 0

Committed status of generating units for 40-units test system (units 21–40)
U21 U22 U23 U24 U25 U26 U27 U28 U29 U30 U31 U32 U33 U34 U35 U36 U37 U38 U39 U40
1 455 455 0 0 0 0 0 0 0 0 455 150 0 0 0 0 0 0 0 0
2 455 425 0 0 0 0 0 0 0 0 455 150 0 0 0 0 0 0 0 0
3 455 455 0 130 0 0 0 0 0 0 455 455 0 0 0 0 0 0 0 0
4 455 455 0 130 25 0 0 0 0 0 455 455 0 0 0 0 0 0 0 0
5 455 455 0 130 25 0 0 0 0 0 455 455 0 0 25 0 0 0 0 0
6 455 455 0 130 25 0 0 0 0 0 455 455 0 0 25 0 0 0 0 0
7 455 455 130 130 25 0 0 0 0 0 455 455 0 0 25 0 0 0 0 0
8 455 455 130 130 25 0 0 0 0 0 455 455 130 130 25 0 0 0 0 0
9 455 455 130 130 51 0 25 0 0 0 455 455 130 130 25 0 25 0 0 0
10 455 455 130 130 162 20 25 0 0 0 455 455 130 130 162 20 25 0 0 0
11 455 455 130 130 162 80 25 10 10 0 455 455 130 130 162 80 25 0 0 0
12 455 455 130 130 162 80 25 43 10 10 455 455 130 130 162 80 25 43 10 10
13 455 455 130 130 162 20 25 0 0 0 455 455 130 130 162 20 25 0 0 0
14 455 455 130 130 51 0 25 0 0 0 455 455 130 130 25 0 25 0 0 0
15 455 455 130 130 25 0 0 0 0 0 455 455 130 130 25 0 0 0 0 0
16 455 455 130 130 25 0 0 0 0 0 455 180 130 130 25 0 0 0 0 0
17 455 285 130 130 25 0 0 0 0 0 455 150 130 130 25 0 0 0 0 0
18 455 380 130 130 25 0 0 0 0 0 455 455 130 130 25 0 0 0 0 0
19 455 455 130 130 25 0 0 0 0 0 455 455 130 130 25 0 0 0 0 0
20 455 455 130 130 162 20 25 0 0 0 455 455 130 130 162 20 25 0 0 0
21 455 455 130 130 162 20 25 0 0 0 455 455 0 0 162 20 25 0 0 0
22 455 455 0 0 162 20 25 0 0 0 455 455 0 0 0 20 25 0 0 0
23 455 455 0 0 25 0 0 0 0 0 455 455 0 0 0 0 0 0 0 0
24 455 150 0 0 0 0 0 0 0 0 455 455 0 0 0 0 0 0 0 0
 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249 241

and the actual probability of the pitch adjustment is
PPitch ¼ r accept  rpa ;
where r accept is the Harmony memory accepting rate and r pa repre-
sents the Pitch Adjustment rate.
Flow chart of proposed algorithm
In proposed hybrid HS-random Search Algorithm, firstly the
position of population matrix is updated using the below Harmony
search algorithm and random search algorithm is used for updat-
ing mutation operator. The flow chart of Harmony Search algo-
rithm is shown in Fig. 8a and PSEUDO code for proposed hybrid
harmony search algorithm is shown in Fig. 8b. The flow chart for
handling various UCP constraints is shown in Fig. 8c and PSEUDO
code is mentioned in Appendicies A and B (see Figs. 9 and 10).
Test systems
The simulation includes runs for 4, 10, 20 and 40 units test sys-
tems. Scheduling periods are 8 h for 4-units test system and 24-h
for 10, 20 and 40 units test system. The generating units character-
istics and load demand data for 4-units test system are taken from
[74] and are shown in Tables 17 and 18 respectively. The charac-
teristics of 10-units test system are taken from [74] and are shown
in Table 19 and load demand pattern is shown in Table 20. For the
20 Units test system, the characteristics of 10-Units system were
duplicated and load demand data was multiplied by 2. The prob-
ð15Þ
lem data of 10-Units test system were scaled appropriately for
the problem with 20 and 40-units test system. The load demand
pattern for 10, 20 and 40-units test system is shown in Fig. 11.
The corresponding results has been obtained using hybrid har-
mony search algorithm using population size of 40 for 4 and 10-
units test system, population size of 80 for 20-units test system
and population size of 160 for 40-units test system. The proposed
algorithm is tested for number of searches from 150 to 1000. The
Flow chart and PSEUDO code of proposed Hybrid HS-Random
Search algorithm is shown in Figs. 8a and 8b. The performance of
the proposed algorithm is tested in MATLAB 2013a (8.1.0.604) soft-
ware on IntelÒ coreTM i-5-3470S CPU@ 2.90 GHz, 4.00 GB RAM
system.
Results and discussion
In order to stochastic nature of Hybrid HS-random Search algo-
rithm, 50 test trials were made for each problem set, with each run
starting with different initial populations. The Population size of 40
(for 4 and 10-units test system) was taken in all runs. The simula-
tion results are shown in Tables 1–15. As shown in comparison
Table 2 for 4-units test system, Table 5 for 10-Units test system
with 10 % spinning reserve, Table 8 for 10-Units test system with
5% Spinning reserve, Table 11 for 20-units test system and Table 13

Table 13
Comparison of results for 40-generating unit system (for 10% spinning reserve).

Sr. no. Method Overall generation cost ($) Average time (s)
Best cost Average cost Worst cost
1 Hybrid Continuous Relaxation and Genetic Algorithm (CRGA) [52] – 2,243,796 – 265
2 Lagrangian Relaxation and Genetic Algorithm (LRGA) [95] – 2,242,178 – 2165
3 Integer Coded Genetic Algorithm (ICGA) [94] – 2,254,123
4 Improved Binary Particle Swarm optimization (IBPSO) [16] 2,401,728 – – 316.86
5 Binary Particle Swarm Optimization with bit Change Mutation (MPSO) [59] 2,323,435 – – 317.29
6 LCA-PSO [58] 2,277,396 – – 274.67
7 Advanced Fuzzy Controlled Binary PSO (AFCBPSO) [86] 2,266,040 – – –
8 LR [38] 2,258,503 – –
9 Enhanced Simulated Annealing (ESA) [93] Considering Ramp Rate 2,255,864 2,256,971 2,258,897 199.55
10 GA [93] 2,252,909 2,262,585 2,269,282 –
11 Genetic Algorithm (GA) [88] 2,251,911 – 2,259,706 2697
12 LR [93] 2,250,223 2,250,223 2,250,223 –
13 Enhanced Simulated Annealing (ESA) [61] 2,250,063 2,252,125 2,254,539 88.28
14 Particle Swarm Optimization (PSO) [67] 2,250,012 – 2,257,146
15 Intelligent Mutation based Genetic Algorithm (UCC-GA) [51] 2,249,715 – 2,256,824 614
16 Genetic Algorithm (GA) [88] 2,249,715 – 2,256,824 2697
17 SM [93] 2,249,589 2,249,589 2,249,589 –
18 Evolutionary Programming (EP) [89] 2,249,093 2,252,612 – 1176
19 EP [93] 2,249,093 2,252,612 2,256,085 –
20 Harmony Search (HS) [72] – 2,250,968 – 467
21 Harmony Search Algorithm (HAS) [87] – 2,248,740 – 78
22 Hybrid HS-Random Search Algorithm [proposed method] 2,248,508 2248652.78 2,248,757 179.66679

The bold results show the superiority of proposed Hybrid Harmony search algorithm over others well known heuristics, meta-heuristics and evolutionary algorithms.

Table 14
Conclusion of results for 4, 10, 20 and 40 units test system.

No. of units Generation cost ($) Computational time (s)

Best cost Average cost Worst cost Best time Average time Worst time
4 74,476 74,476 74,476 20.68704 22.863569 22.9709
10 (SR = 10%) 563937.6875 563965.3094 563995.3262 16.831236 16.9158306 16.99832
10 (SR = 5%) 557905.6427 558267.22 558682.0107 14.36105 15.88731 16.33696
20 1124889.395 1124912.84 1124951.548 35.01579 38.21 39.38754
40 2248507.702 2248652.78 2248757.422 176.9953 179.66679 181.713
242 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249

for 40-units test system, proposed hybrid Harmony Search-
Random Search algorithm gives better solution in comparison with
other well known metaheuristics algorithms. Table 1 gives com-
mitment and generation schedule for 4-units test system (Total
cost: $ 74476s). Tables 3 and 4 gives commitment and generation
schedule for 10-units test system with 10% spinning reserve (Total
cost: $ 563937.6875). Tables 6 and 7 gives commitment and gen-
eration schedule for 10-units test system with 5% spinning reserve
(Total cost: $ 557905.6427). Tables 9 and 10 gives commitment
and generation schedule for 20-units test system with 10% spin-
ning reserve (Total cost: $ 1124889.395). Table 12 gives commit-
ment and generation schedule for 40-units test system with 10%
spinning reserve (Total cost: $ 2248507.702). The commitment
schedule from Tables 3, 6, 9 and 12 indicates that all physical

Table 15
Comparison of proposed algorithm with other harmony search algorithms.

Method 10-Unit system 20-Units system 40-Units system

Best cost Mean cost Worst cost Execution time Best cost Execution time Best cost Execution time
Harmony Search [72] 564367.69 – – – 1,127,377 92 2,250,968 467
Harmony Search Algorithm (HAS) [87] 563,977 564168.6 – 3.00 1,124,715 24 2,248,740 78
Proposed Method 563937.69 563965.31 563995.3262 16.831236 1124912.84 35.01579 2248653 179.66679

The bold results show the superiority of proposed Hybrid Harmony search algorithm over others well known heuristics, meta-heuristics and evolutionary algorithms.

Table 16
Parameter sensitivity analysis for proposed algorithm.

HMCR B.W.=0.2 B.W.=0.6 Average execution time (s)

Pitch Adjustment Rate (PAR)
0 0.5 0.95 0 0.5 0.95
0 558272.26 558395.3696 559010.74 558377.95 558516.46 560380.70 28.936581
0.2 558647.14 559265.02 559795.69 558957.9345 560982.29 560732.24 28.66208633
0.4 559118.3878 559923.88 561264.74 559481.04 561047.80 561169.74 28.43884067
0.8 559157.2897 559567.66 561231.15 559486.37 561100.07 561796.42 28.19135733
1 559554.8208 560643.75 561414.84 560177.59 561612.86 562413.29 27.37707567

Table 17
4-Unit test system [74].

Unit no. P max

ih P min
ih
a ($/MW2 h) b ($/MW h) c ($/h) MUTi MDTi HSCi CSCi CSHi ISi

U1 300 75 684.74 16.83 0.0021 5 4 500 1100 5 8

U2 250 60 585.62 16.95 0.0042 5 3 170 400 5 8
U3 80 25 213 20.74 0.0018 4 2 150 350 4 5
U4 60 20 252 23.6 0.0034 1 1 0 0.02 0 6

Table 18
Load demand for 4-unit test system.

Hour 1 2 3 4 5 6 7 8
Load demand 450 530 600 540 400 280 290 500

Table 19
Test data for 10-unit system [74].

Unit no. P max

ih P min
ih
a ($/MW2) b ($/MW h) c ($/h) MUTi MDTi HSCi CSCi CSHi ISi

U1 455 150 1000 16.19 0.00048 8 8 4500 9000 5 8

U2 455 150 970 17.26 0.00031 8 8 5000 10,000 5 8
U3 130 20 700 16.6 0.002 5 5 550 1100 4 5
U4 130 20 680 16.5 0.00211 5 5 560 1120 4 5
U5 162 25 450 19.7 0.00398 6 6 900 1800 4 6
U6 80 20 370 22.26 0.00712 3 3 170 340 2 3
U7 85 25 480 27.74 0.00079 3 3 260 520 2 3
U8 55 10 660 25.92 0.00413 1 1 30 60 0 1
U9 55 10 665 27.27 0.00222 1 1 30 60 0 1
U10 55 10 670 27.79 0.00173 1 1 30 60 0 1
 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249 243

Table 20
Load demand pattern for 24-h for 10-unit system.

Hour 1 2 3 4 5 6 7 8 9 10 11 12
Demand 700 750 850 950 1000 1100 1150 1200 1300 1400 1450 1500
Hour 13 14 15 16 17 16 19 20 21 22 23 24
Demand 1400 1300 1200 1050 1000 1100 1200 1400 1300 1100 900 800

and operational constraints are satisfied. In comparison with the
results produced by reported methods, the proposed method gives
satisfactory solution in reasonable computation time. The parame-
ter sensitivity analysis of proposed algorithm is presented in
Table 16. In analysis, test system consisting of 10-generating units
has been taken into consideration with spinning reserve of 5%. The
performance of the algorithm is tested for two different bandwidth
value of 0.2 and 0.6 and analysis features the PAR and HMCR set
from 0 to 1. Average execution time (in seconds) is presented for
analysis purpose. The algorithm with high value of HMCR does
not perform well for unit commitment problem. However, for
HMCR of 1, execution time is significantly low. Also, low PAR value
results in better cost. The analysis for global capability of proposed
hybrid algorithm is presented in Table 21 for unimodal and multi-
modal benchmark functions.
Conclusion and future scope
model has been successfully evaluated using proposed algorithm.
Experimental results on various unimodal and multimodal bench-
mark functions reveals that performance of proposed Hybrid HS-
Random Search algorithm is much better than others well known
and recently developed evolutionary, heuristics and meta-
heuristics search algorithm. From simulation results, proposed
algorithm can be determine satisfactory commitment schedule in
reasonable computation time and such a powerful algorithm can
be used for the solution of multi-objective and multi-area unit
commitment problem of electric power system.
Acknowledgments
The corresponding author wish to thank DAV University,
Jalandhar and I.K. Gujral Punjab Technical University, Jalandhar
(Punjab), for providing advanced research facilities during research
work.

In this paper, researchers have presented the solution of Unit

Commitment Problem of electric power system using Hybrid Har- Appendices: PSEUDO code for various repairing of UCP
mony Search-Random Search Algorithm. The results for standard constraints
IEEE test system consisting of 4, 10, 20 and 40-units Generating
See Table 17–21.
244 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249

Table 21
The analysis for global capability of proposed hybrid for unimodal and multimodal benchmark functions.

Unimodal benchmark functions

P
Function : F 1 ðxÞ ¼ ni¼1 x2i Parameter space Convergence Curve for F1
Dimension: 30
Range: [100, 100] 4
Hybrid HS algorithm
No. of iterations = 500 x 10
0
f min ¼ 4:0117e28 10
2

Best fitness obtained---->

-5
10
1.5

F1( x1 , x )
2
-10
1 10

0.5 -15
10

0 -20
100 10

50 100
50
0 -25
x2 0 10
-50 -50 x1
-100 -100 100 200 300 400 500
Iteration----->
Pn
Function : F 2 ðxÞ ¼ i¼1 jxi j þ Pni¼1 jxi j Parameter space Convergence Curve for F2
Dimension: 30
10
Range: [10, 10] 10 Hybrid HS algorithm

No. of iterations = 500

f min ¼ 5:9877e17 12000

Best fitness obtained---->

105
10000

8000
F2(x1, x2 )

100
6000

4000
10-5
2000

0
100 10-10
50 100
0 50
x2 0 10-15
-50 -50 x1
-100 -100 100 200 300 400 5
Iteration----->
P Pi 2
Function : F 3 ðxÞ ¼ ni¼1 j1 xj Parameter space Convergence Curve for F3
Dimension: 30
Hybrid HS algorithm
Range: [100, 100] x 10
4
4
No. of iterations = 500 10
f min ¼ 2:7791e05 5
Best fitness obtained---->

4 2
10
F3( x , x )
2

3
1

2 0
10

-2
0 10
100
50 100
0 50
-4
x2 0 10
-50 -50 x1
-100 -100 100 200 300 400 500
Iteration----->
 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249 245

Table 21 (continued)

Unimodal benchmark functions

Function : F 4 ðxÞ ¼ maxi fjxi j; 1 6 i 6 ng Parameter space Convergence Curve for F4

Dimension: 30
Hybrid HS algorithm
Range: [100, 100]
No. of iterations = 500
100

Best fitness obtained---->

f min ¼ 5:9439e07 0
10
80

F4(x , x )
2
60
-2

1
10
40

20
-4
0 10
100
50 100
0 50
0
x2 -50 -50 10
-6
x1
-100 -100 100 200 300 400 50
Iteration----->
Parameter space Convergence Curve for F5
P 2 2
Function : F 5 ðxÞ ¼ n1 2
i¼1 ½100ðxiþ1  xi Þ þ ðxi  1Þ  8
10 Hybrid HS algorithm
Dimension: 30 x 10
11

Range: [30, 30]

7
No. of iterations = 500 2 10

Best fitness obtained---->

f min ¼ 27:9558
1.5 10
6
F5(x1 , x2 )

1 10
5

0.5 4
10

0 3
200 10
100 200
0 100 2
10
x2 0
-100 -100 x1
-200 -200 100 200 300 400
Iteration----->
P
Function : F 6 ðxÞ ¼ ni¼1 ½xi þ 0:52 Parameter space Convergence Curve for F6
Dimension: 30
Range: [100, 100] 4
Hybrid HS algorithm
No. of iterations = 500 x 10
f min ¼ 0:74774 10
4
2.5
Best fitness obtained---->

2
3
10
F6( x , x )
2

1.5
1

1
2
10
0.5

0 1
100 10

50 100
0 50
x2 0 0
-50 10
-50 x1
-100 -100 100 200 300 400
Iteration----->

(continued on next page)

246 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249

Table 21 (continued)

Unimodal benchmark functions

P 4
Function : F 7 ðxÞ ¼ ni¼1 ixi þ random½0; 1Þ Parameter space Convergence Curve for F7
Dimension: 30
Range: [1.28, 1.28] Hybrid HS algorithm
No. of iterations = 500
f min ¼ 0:0016663 4 10
1

Best fitness obtained---->

3
0

F7( x1 , x 2 )
10
2

1 -1
10

0
1
-2
0.5 1 10
0 0.5
x2 0
-0.5 x1
-0.5
-1 -1 50 100 150 200 250 300 350 400 450 500
Iteration----->
Multimodal benchmarkfunctions pffiffiffiffiffiffiffi
Pn
Function : F 8 ðxÞ ¼ i¼1 xi sin jxi j Parameter space Convergence Curve for F8
Dimension: 30
Hybrid HS algorithm
Range: [500, 500]
3.2
No. of iterations = 500 -10
f min ¼ 6154:0161 1000

Best fitness obtained---->

3.3
-10
500
F8(x , x )
2

3.4
-10
0
1

3.5
-10
-500

3.6
-1000 -10
500
500 3.7
-10
0
x2 0
x1
-500 -500 100 200 300 400 500
Iteration----->
P
Function : F 9 ðxÞ ¼ ni¼1 ½x2i  10 cosð2pxi Þ þ 10 Parameter space Convergence Curve for F9
Dimension: 30
Hybrid HS algorithm
Range: [5.12, 5.12]
0
No. of iterations = 500
Best fitness obtained---->

100 10
f min ¼ 5:6843e14
80
F9( x , x )
2

60
-5
10
1

40

20

0
5 10
-10

5
0
x2 0
-5 -5
x1
50 100 150 200 250 300 350 400 450 500
Iteration----->
 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249 247

Table 21 (continued)

Unimodal benchmark functions

Pn  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn 
Function : F 10 ðxÞ ¼ i¼1  20 exp 0:2 1n i¼1 x2i Parameter space Convergence Curve for F10
1 Pn 2 
 exp n i¼1 xi cosð2pxi Þ þ 20 þ e
Hybrid HS algorithm
Dimension: 30 10
0

Range: [32, 32]

25

Best fitness obtained---->

No. of iterations = 500 10
-2

f min ¼ 9:3259e14 20

F10(x , x )
-4

2
10
15

1
-6
10 10

5 -8
10
0
20 -10
10
10 20
0 10 -12
0 10
x2 -10 -10 x1
-20 -20 100 200 300 400
Iteration----->
Pn  
 Pni¼1 cos piffi þ 1
x
Function : F 11 ðxÞ ¼ 4000
1 2
i¼1 xi i
Parameter space Convergence Curve for F11
Dimension: 30
Hybrid HS algorithm
Range: [600, 600]
0
No. of iterations = 500 10
150
f min ¼ 0

Best fitness obtained---->

100
F11(x , x )
2

-5
10
1

50

-10
10
0
500
500
0
x2 0 10
-15

x1
-500 -500 100 200 300 400 5
Iteration----->
Function : F 12 ðxÞ ¼ pn f10 sinðpy1 Þ Parameter space Convergence Curve for F12
P
ðy  1Þ2 ½1 þ 10 sin ðpyiþ1 Þ þ ðyn  1Þ2 g
2
þ n1
Pi¼1 i 8
þ ni¼1 uðxi ; 10; 100; 4Þ 10 Hybrid HS algorithm
x 8 9
yi ¼ 1 þ iþ1 4
m
< lðxi  aÞ ; xi > a =
m
uðxi ; a; l; mÞ ¼ lðxi  aÞ ; xi < a
: ; 150
Dimension: 30 0; a < xi < a 6
10
Range: [50, 50]
Best fitness obtained---->

No. of iterations = 500

f min ¼ 0:037298 100
F12(x , x )
2

4
10
1

50

2
10

0
10
5 10 0
10
5
0
x2 0
-5 -5 x1
-10 -10 100 200 300 400 500
Iteration----->

References [4] Kumar Vikram, Bath Sarbjeet Kaur. Single area unit commitment problem by
modern soft computing techniques. Int J Enhanc Res Sci Technol Eng 2013;2
(3). ISSN No: 2319-7463.
[1] Bhardwaj A, Tung NS, Shukla VK, Kamboj VK. The important impacts of unit
[5] Sriyanyong P, Song YH. Unit commitment using particle swarm optimization
commitment constraints in power system planning. Int J Emerg Trend Eng
combined with lagrange relaxation. In: Proc IEEE power engineering society
Develop 2012;5(2):301–6.
general meeting, San Francisco, CA, vol. 3; 2005. p. 2752–9.
[2] Zhu J. Unit commitment. In: Optimization of power system
[6] Xiong W, Li MJ, Cheng YL. An improved particle swarm optimization algorithm
operation. Hoboken, NJ: Wiley IEEE Press; 2009. p. 251–93 [chapter 7].
for unit commitment. In: Proc international conference on intelligent
[3] Rajan CCA, Mohan MR, Manivannan K. Neural based tabu search method for
computation technology and automation (ICICTA-2008), Changsha, Hunan,
solving unit commitment problem. In: Proc international conference on power
China, vol. 2; 2008. p. 21–5.
system management and control (conf publ no 488), London, UK; 2002. p.
[7] Jeong YW, Park JB, Jang SH, Lee KY. A new quantum-inspired binary PSO for
180–5.
thermal unit commitment problems. In: Proc 15th international conference
248 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249
on intelligent system applications to power systems, Curitiba, Brazil; 2009.
p. 1–6.
[8] Ge W. Ramp rate constrained unit commitment by improved priority list and
enhanced particle swarm optimization. In: Proc 2010 international conference
on computational intelligence and software engineering (CiSE 2010), Wuhan,
China; 2010. p. 1–8.
[9] Borghetti A, Frangioni A, Lacalandra F, Lodi A, Martello S, Nucci CA, et al.
Lagrangian relaxation and tabu search approaches for the unit commitment
problem. In: Proc IEEE power tech conference, Porto, Portugal, vol. 3; 2001. p.
1–7.
[10] Gaing ZL. Discrete particle swarm optimization algorithm for unit
commitment. In: Proc IEEE power engineering society general meeting,
Toronto, Canada, vol. 1; 2003. p. 418–24.
[11] Rajan CCA, Mohan MR, Manivannan K. Neural based tabu search method for
solving unit commitment problem. IEE Proc Gener Transm Distrib 2003;150
(4):469–74.
[12] Gaing ZL. Discrete particle swarm optimization algorithm for unit
commitment. In: IEEE power engineering society general meeting, 13–17
July 2003, vol. 1; 2003. p. 418–24.
[13] Zhao B, Guo CX, Bai BR, Cao YJ. An improved particle swarm optimization
algorithm for unit commitment. Int J Elect Power Energy Syst
2006;28:482–90.
[14] Lee TY, Chen CL. Unit commitment with probabilistic reserve: an IPSO
approach. Energy Convers Manage 2007;48(2):486–93.
[15] Samudi C, Das GP, Ojha PC, Sreeni TS, Cherian S. Hydro-thermal scheduling
using particle swarm optimization. In: IEEE/PES transmission and distribution
conference and exhibition; April 2008. p. 1–5.
[16] Yuan X, Nie H, Su A, Wang L, Yuan Y. An improved binary particle swarm
optimization for unit commitment problem. Expert Syst Appl: Int J 2009;36
(4):8049–55. http://dx.doi.org/10.1016/j.eswa.2008.10.047.
[17] Tahanan M, van Ackooij W, Frangioni A, Lacalandra F. Large-scale unit
commitment under uncertainty. 4OR 2015;13(2):115–71.
[18] Mirjalili Seyedali, Lewis Andrew. Adaptive gbest-guided gravitational search
algorithm. Neural Comput Appl 2014;25(7–8):1569–84.
[19] Dhillon JS, Kothari DP. Power system optimization. 2nd ed., New Delhi: PHI;
2010.
[20] Mirjalili Seyedali, Mirjalili Seyed Mohammad, Lewis Andrew. Grey wolf
optimizer. Adv Eng Softw 2014;69:46–61.
[21] Hara K, Kimura M, Honda N. A method for planning economic unit
commitment and maintenance of thermal power systems. IEEE Trans Power
Apparatus Syst 1966;PAS-85(5):421–36.
[22] Guy JD. Security constrained unit commitment. IEEE Trans Power Apparatus
Syst 1971;90:1385–9.
[23] Lowery PG. Generating unit commitment by dynamic programming. IEEE
Trans Power Apparatus Syst 1966;PAS-85(5):422–6.
[24] Hobbs WJ, Hermon G, Warner S, Sheble GB. An enhanced dynamic
programming approach for unit commitment. IEEE Trans Power Syst 1988
(3):1201–5.
[25] Tao Li, Shahidehpour SM. Price-based unit commitment: a case of Lagrangian
relaxation versus mixed integer programming. IEEE Trans Power Syst 2005;20
(4):2015–25.
[26] Venkatesh B, Jamtsho T, Gooi HB. Unit commitment a fuzzy mixed integer
linear programming solution. IET Gener Transm Distrib 2007;1(5):836–46.
[27] Mohan Saini L, Soni Mahender Kumar. Artificial neural network-based peak
load forecasting using conjugate gradient methods. IEEE Trans Power Syst
2002;17.3:907–12.
[28] Guan X, Zhai Q, Papalexopoulos A. Optimization based methods for unit
commitment: Lagrangian relaxation versus general mixed integer
programming. In: Proc IEEE power engineering society general meeting,
Toronto, Canada, vol. 2; 2003. p. 1095–100.
[29] Cohen AI, Yoshimura M. A branch-and-bound algorithm for unit commitment.
IEEE Trans Power Apparatus Syst 1983;102(2):444–51.
[30] Salam Md-Sayeed, Hamdan Abdul-Razak, Mohamed Nor Khalid. Integrating an
expert system into a thermal unit-commitment algorithm. IEE Proc C 1991;
l38:553–9.
[31] Kadam DP, Sonwane PM, Dhote VP, Kushare BE. Fuzzy logic algorithm for unit
commitment problem. In: Proc international conference on control,
automation, communication and energy conversation (INCACEC-2009),
Perundurai, Erode, India; 2009. p. 1–4.
[32] Yalcinoz T, Short MJ, Cory BJ. Application of neural networks to unit
commitment. IEEE Trans Power Syst 1999:649–54.
[33] Simopoulos D, Contaxis G. Unit commitment with ramp rate constraints using
the simulated annealing algorithm. In: Proc IEEE mediterranean
electrotechnical conference (MELECON 2004), Dubrovnik, Croatia; May 12–
15, 2004. p. 845–9.
[34] Mantawy AH, Abdel-Magid YL, Selim SZ. Unit commitment by tabu search. IEE
Proc Gener Transm Distrib 1998;145(1):56–64.
[35] Mirjalili Seyedali, Lewis Andrew. Adaptive gbest-guided gravitational search
algorithm. Neural Comput Appl 2014;25(7–8):1569–84.
[36] Geem ZW, Kim JH, Loganathan GV. A new heuristic optimization algorithm:
harmony search. Simulation 2001;76(2):60–8.
[37] Novel derivative of harmony search algorithm for discrete design variables.
Appl Math Comput 199: 223–30. http://dx.doi.org/10.1016/j.amc.2007.09.049.
[38] Improved harmony search from ensemble of music players. Lect Notes Comput
Sci 86–93. http://dx.doi.org/10.1007/11892960_11.
[39] Geem ZW, Tseng CL, Park Y. Harmony search for generalized orienteering
problem: best touring in China. In: Wang L, Chen K, Ong Y, editors. Advances in
natural computation. Berlin: Springer; 2005. p. 741–50.
[40] Geem Z. Improved harmony search from ensemble of music players. In: Gabrys
B, Howlett RJ, Jain L, editors. Knowledge-based intelligent information and
engineering systems. Heidelberg: Springer; 2006. p. 86–93.
[41] Mahdavi M, Fesanghary M, Damangir E. An improved harmony search
algorithm for solving optimization problems. Appl Math Comput 2007;188
(2):1567–79.
[42] Omran MGH, Mahdavi M. Global-best harmony search. Appl Math Comput
2008;198(2):643–56.
[43] Mukhopadhyay A, Roy A, Das S, Abraham A. Population-variance and
explorative power of harmony search: an analysis. In: Second national
conference on mathematical techniques emerging paradigms for electronics
and IT industries (MATEIT 2008), New Delhi, India; 2008.
[44] Degertekin S. Optimum design of steel frames using harmony search
algorithm. Struct Multidiscip Optim 2008;36(4):393–401.
[45] Chakraborty P, Roy GG, Das S, Jain D, Abraham A. An improved harmony search
algorithm with differential mutation operator. Fundam Inform 2009;95:1–26.
[46] Hasancebi O, Erdal F, Saka MP. An adaptive harmony search method for
structural optimization. J Struct Eng 2009;1:72.
[47] Saka M, Hasancebi O. Adaptive harmony search algorithm for design code
optimization of steel structures. In: Geem Z, editor. Harmony search
algorithms for structural design optimization. Berlin, Heidelberg: Springer-
Verlag; 2009. p. 79–120.
[48] Kattan A, Abdullah R, Salam RA. Harmony search based supervised training of
artificial neural networks. In: International conference on intelligent systems,
modelling and simulation (ISMS); 2010. p. 105–10.
[49] Wang CM, Huang YF. Selfadaptive harmony search algorithm for optimization.
Expert Syst Appl 2010;37(4):2826–37.
[50] Al-Betar M, Khader A, Liao I. A harmony search with multi-pitch adjusting rate
for the university course timetabling. In: Geem Z, editor. Recent advances in
Harmony search algorithm. Berlin, Heidelberg: Springer-Verlag; 2010. p.
147–61.
[51] Senjyu Tomonobu, Yamashiro Hirohito, Uezato Katsumi, Funabashi Toshihisa.
A unit commitment problem by using genetic algorithm based on unit
characteristics classifications. In: Proceeding 2002 IEEE power engg soc winter
meeting, vol. 1; 2002. p. 58–63.
[52] Tokoro Ken-ichi, Masuda Yasushi, Nishino Hisakazu. Solving unit commitment
problem by combining of continuous relaxation method and genetic
algorithm. In: SICE annual conference 2008. Japan: The University Electro-
Communications; August 20–22, 2008.
[53] Ongsakul W, Petcharaks N. Unit commitment by enhanced adaptive
Lagrangian relaxation. IEEE Trans Power Syst 2004;19(1):620–8.
[54] Sheble GB et al. Unit commitment by genetic algorithm with penalty method
and a comparison of Lagrangian search and genetic algorithm economic
dispatch example. Int J Elect Power Energy Syst 1997;19(1):45–55.
[55] Grefensttete JJ. Optimization of control parameters for genetic algorithm. IEEE
Trans Syst Man Cybern 1986;16:122–8.
[56] Zhe Wang, Yiyin Yu, Hongpeng Zhang. Social evolutionary programming based
unit commitment. Proc CSEE 2004;24(4):24–8.
[57] Fei Lan, Jinghua Li. A solution to the unit commitment problem based on local
search method. In: International conference on energy and environment
technology. In: Proceeding international conference on energy and
environment technology, 2009 (ICEET ’09), 16–18 October 2009, Guilin,
Guangxi, vol. 2; 2009. p. 51–6.
[58] Wang Bo, Li You, Watada Junzo. Re-scheduling the unit commitment problem
in fuzzy environment. In: IEEE international conference on fuzzy systems, 27–
30 June, 2011, Taipei, Taiwan; 2011.
[59] Lee S, park H, Jeon M. Binary particle swarm optimization with bit change
mutation. IEICE Trans Fundam Electron Commun Comput Sci 2007;E-90A
(10):2253–6.
[60] Valenzuela J, Smith AE. a seeded memetic algorithm for large unit
commitment problems. J Heuristics 2002;8:173–95.
[61] Simopoulos Dimitris N, Kavatza SD, Vournas CD. Unit commitment by an
enhanced simulated annealing algorithm. IEEE Trans Power Syst 2006;21
(1):68–76.
[62] Chung CY, Yu H, Wong KP. An advanced quantum-inspired evolutionary
algorithm for unit commitment. IEEE Trans Power Syst 2006;26(2):847–54.
[63] Jeong Y, Park J, Jang S, Lee KY. A new quantum-inspired binary PSO:
application to unit commitment problems for power systems. IEEE Trans
Power Syst 2010;25(3):1486–95.
[64] Chakraborty Shantanu, Senjyu Tomonobu, Yona Atsushi, Funabashi Toshihisa.
Fuzzy quantum computation based thermal unit commitment strategy with
solar battery system injection. In: IEEE international conference on fuzzy
systems, June 27–30, 2011, Taipei, Taiwan; 2011.
[65] Marifeld Tim T, Sheble Gerald B. Genetic based unit commitment algorithm.
IEEE Trans Power Syst 1996;11(3):1359–70.
[66] Victoire TAA, Jeyakumar AE. Hybrid PSO-SQP for economic dispatch with
valve-point effect. Electr Power Syst Res 2004;71(1):51–9.
[67] Zhao B, Guo CX, Bai BR, Cao YJ. An improved particle swarm optimization
algorithm for unit commitment. Electr Power Energy Syst 2006;28(7):482–90.
[68] Jeong YW, Park JB, Jang SH, Lee KY. A new quantum-inspired binary PSO for
thermal unit commitment problems. In: Proc IEEE 15th int conf intelligent
system applications to power systems; 2009. p. 1–6.
 V.K. Kamboj et al. / Electrical Power and Energy Systems 77 (2016) 228–249
[69] Sadati Nasser, Hajian MahdI, Zamani Majid. Unit commitment using particle
swarm based simulated annealing optimization approach. In: Proceeding of
the IEEE swarm intelligence symposium (SIS2007); 2007. p. 297–302.
[70] Senjyu T et al. A fast technique for unit commitment problem by extended
priority list. IEEE Trans Power Syst 2003;18(2).
[71] Sum-im T, Ongsakul W. Ant colony search algorithm for unit commitment. In:
IEEE con ICIT; 2003.
[72] Najafi S, pourjamal Y. A new heuristic algorithm for unit commitment
problem. Energy Procedia 2012;14:2005–11.
[73] Jeong Yun-Won, Lee Woo-Nam, Kim Hyun-Houng, Park Jong-Bae, Shin Joong-
Rin. Thermal unit commitment using binary differential evolution. J Electr Eng
Technol 2009;4(3):323–9.
[74] Khanmohammadi S, Amiri M, Tarafdar Haque M. A new three-stage method
for solving unit commitment problem. Energy 2010:3072–80. http://dx.doi.
org/10.1016/j.energy.2010.03.049.
[75] Gaing ZL. Discrete particle swarm optimization algorithm for unit commit-
ment. In: IEEE power engineering society general meeting, vol. 1; 2003. p. 13–
7.
[76] Pappala VS, Erlich I. A new approach for solving the unit commitment problem
by adaptive particle swarm optimization. In: Power and energy society general
meeting-conversion and delivery of electrical energy in the 21st century. USA:
IEEE; 2008. p. 1–6.
[77] Eldin AS, El-sayed MAH, Youssef HKM. A two-stage genetic based technique
for the unit commitment optimization problem. In: 12th international middle
east power system conference, MEPCO, Aswan; 2008. p. 425–30.
[78] Xiong W, Li MJ, Cheng YL. An improved particle swarm optimization algorithm
for unit commitment. In: Proceedings of the 2008 international conference on
intelligent computation technology and automation, vol. 01; 2008. p. 21–5.
[79] Victoire TAA, Jeyakumar AE. Hybrid PSO–SQP for economic dispatch with
valve-point effect. Electr Power Syst Res 2004;71(1):51–9.
[80] Chusanapiputt S, Nualhong D, Jantarang S, Phoomvuthisarn S. A solution to
unit commitment problem using hybrid ant system/prioritylist method. In:
IEEE 2nd international conference on power and energy, PECon 08, Malaysia;
2008. p. 1183–8.
[81] Tokoro KI, Masuda Y, Nishina H. Solving unit commitment problem by
combining of continuous relaxation method and genetic algorithm. In: SICE
annual conference. Japan: The University Electro-Communications; 2008. p.
3474–8.
[82] Tingfang Y, Ting TO. Methodological priority list for unit commitment
problem. In: International conference on computer science and software
engineering, CSSE, vol. 1; 2008. p. 176–9.
249
[83] Eldin AS, El-sayed MAH, Youssef HKM. A two-stage genetic based technique
for the unit commitment optimization problem. In: 12th international middle
east power system conference, MEPCO, Aswan; 2008. p. 425–30.
[84] Roy Provas Kumar. Solution of unit commitment problem using gravitational
search algorithm. Electr Power Energy Syst 2013;53:85–94.
[85] Chandram K, Subrahmanyam N, Sydulu M. Unit commitment by improved
pre-prepared power demand table and Muller method. Int J Electr Power
Energy Syst 2011;33:106–14.
[86] Chakraborty S, Ito T, Senjyu T, Saber AY. Unit commitment strategy of thermal
generators by using advanced fuzzy controlled binary particle swarm
optimization algorithm. Int J Electr Power Energy Syst 2012;43(1):1072–80.
[87] Afkousi-Paqaleh M, Rashidinejad M. An implementation of harmony search
algorithm to unit commitment problem. Electr Eng 2010;92:215–25. http://
dx.doi.org/10.1007/s00202-010-0177.
[88] Kazarlis SA, Bakirtzis AG, Petridis V. A Genetic algorithm solution to the unit
commitment problem. IEEE Trans Power Syst 1996;11(1):83–92.
[89] Juste KA, Kita H, Tanaka E, Hasegawa J. An evolutionary programming solution
to the unit commitment problem. IEEE Trans Power Syst 1999;14(4):1452–9.
http://dx.doi.org/10.1109/59.801925.
[90] Ganguly Debjani, Sarkar Vaskar, Pal Jagdish. A new genetic approach for solving
the unit commitment problem. In: International conference on power system
technology-POWERCON 2004, Singapore; 21–24 November, 2004. p. 542–7.
[91] Gaing Zwe-Lee. Particle swarm optimization to solving the economic dispatch
considering the generator constraints. IEEE Trans Power Syst 2003;18
(3):1187–95.
[92] Sriyanyong P, Song YH. Unit commitment using particle swarm optimization
combined with lagrange relaxation. IEEE Trans 2005:1–8.
[93] Simopoulos DN, Kavatza SD, Vournas CD. Unit commitment by an enhanced
simulated annealing algorithm. In: Power systems conference and exposition,
2006. PSCE ’06. 2006 EEE PES October 29 2006–November 1 2006; 2006. p.
193–201.
[94] Damousis IG, Bakirtzis AG, Dokopoulos PS. A solution to the unit commitment
problem using integer-coded genetic algorithm. IEEE Trans Power Syst
2004;19(2):1165–72.
[95] Cheng C-P, Liu C-W, Liu CC. Unit commitment by Lagrangian relaxation and
genetic algorithms. IEEE Trans Power Syst 2000;15(2):707–14.
[96] Cheng C-P, Liu C-W, -C Liu C. Unit commitment by annealing genetic
algorithms. Electr Power Energy Syst 2000;24:149–58.
[97] Ting TO, Rao MVC, Loo CK. A novel approach for unit commitment problem via
an effective hybrid particle swarm optimization. IEEE Trans Power Syst
2006;21(1):411–8.