Renewable and Sustainable Energy Reviews 48 (2015) 214–227

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

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

A review on optimization algorithms and application to wind energy

integration to grid
Sasmita Behera a,n, Subhrajit Sahoo b, B.B. Pati b
a
Dept. of Electrical & Electronics Engg., VS.S. University of Technology, Burla 768018, Odisha, India
b
Dept. of Electrical Engg., VS.S. University of Technology, Burla 768018, Odisha, India

art ic l e i nf o a b s t r a c t

Article history: In order to meet power needs, with concern for economics and environment, wind energy conversion is
Received 4 May 2014 gradually gaining interest as a suitable source of renewable energy. To maximize the power extraction
Received in revised form from the wind, optimization techniques are used at the various module of a wind farm starting from
30 November 2014
wind farm design for siting, sizing, optimal placement and sizing of distributed generation (DG) sources,
Accepted 8 March 2015
generation scheduling, tuning of PID controller, control of wind energy conversion system (WECS) etc.
This paper mainly focuses on the optimization algorithms (mostly the swarm based) in relation to
Keywords: integration of the wind farm with the grid. The paper here gives a precise idea about different
Optimization algorithms optimization techniques, their advantage and disadvantage with respect to a wind farm. This review will
Wind farm design for sitting
enable the researchers to open the mind to explore possible applications in this field as well as beyond
Optimal placement and sizing of DG
this area.
Generation scheduling
PID controller tuning & 2015 Elsevier Ltd. All rights reserved.
Control of wind energy conversion system
(WECS)

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
2. Salient features of the algorithms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
2.1. Biology based. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
2.1.1. Evolution based algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
2.1.2. Swarm based algorithm. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
2.2. Physics based algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2.2.1. Simulated Annealing (SA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2.2.2. Gravitational Search Algorithm (GSA). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2.2.3. Chaotic Optimization Algorithm (COA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
2.2.4. Intelligent Water Drops algorithm (IWD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2.2.5. Magnetic Optimization Algorithm (MOA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2.3. Geography based algorithms. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2.3.1. Tabu Search Algorithm (TS). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
2.3.2. Imperialistic Competition Algorithm (ICA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
3. Wind farm design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
3.1. Siting of wind farm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
3.2. Wind farm layout design. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
4. Optimal placement of distributed generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
5. Generation scheduling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
5.1. Economic load dispatch problem formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
6. PID controller parameter tuning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

n
Corresponding author. Tel.: +91 9437367106.
E-mail addresses: sasmitabehera2000m@gmail.com (S. Behera), 89subhrajit@gmail.com (S. Sahoo), pati_bibhuti@rediffmail.com (B.B. Pati).

http://dx.doi.org/10.1016/j.rser.2015.03.066
1364-0321/& 2015 Elsevier Ltd. All rights reserved.
 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227 215

7. Control of wind energy conversion system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223

8. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224

1. Introduction
These algorithms are sometimes termed as memetic algorithms [6]
without any further division in to subcategories. The biology based
The present world craves for optimization with respect to every
optimization algorithms are broadly divided into two types, namely:
possible aspect of the nature and its happenings due to the rapid
depletion of easy energy sources and profit maximization. Hence a
– Evolution based optimization algorithms and
lot of focus has been put on optimization of different fields of
– Swarm based optimization algorithms.
engineering problems and management problems mainly which are
multi-dimensional and mathematical in nature. Faster depletion of
the fossil fuels and concern over environmental pollution demand
for clean energy generation. Wind being the clean green energy is
2.1.1. Evolution based algorithm
preferred by many countries over other forms of energy. With rising
Evolutionary algorithms (EAs) are stochastic search methods
power demand, wind energy conversion is gradually gaining more
that mimic the process of biological evolution and/or the social
interest as a cost-effective suitable source of renewable energy.
behavior of species. The species surpasses others by learning,
Here an attempt has been taken to present a brief idea about the
adaptation, and evolution. There are a number of evolution based
optimization algorithms, mostly the swarm based and their use in
optimization algorithms, detailed as follows.
power system domain with advantages and disadvantages reviewed
as per the works done by various authors. Researchers have reviewed
earlier taking different aspects of optimization in relation to renew- 2.1.1.1. Genetic Algorithm (GA). GA is a stochastic global adaptive
able energy. Herbert et al. [1] have mainly reviewed different search optimization technique based on the mechanisms of
assessment models, technological aspects in planning, design of natural selection. It is initialized to a population containing a
wind farms and mechanical components of turbines, testing and number of chromosomes where each one represents a solution of
performance evaluation of design. Banosa et al. have reviewed the problem whose performance is evaluated by a fitness function.
optimization methods in context of different categories of renewable Basically, GA consists of three main stages: Selection, Crossover
sources [2]. But application in wind energy has not been detailed in and Mutation. The application of these three basic operations
depth and new optimization techniques have come up after their allows the creation of new individuals which may be better than
review. Salcedo-Sanz et al. [3] have detailed and summarized the their parents. This algorithm is repeated for many generations and
algorithms for lay out optimization of on shore and off shore wind finally stops at the optimum solution to the problem [7].
farms. In a latest review Iqbal et al. [4] have schematically pointed Despite excellent performance by GA for globally optimum
out main objectives considered for optimization with respect to solution in search space, some researchers have pointed out some
renewable but mostly review for hybrids of renewable energy deficiencies in GA performance. Those deficiencies are (i) poor
sources. But optimization in different areas of research on wind premature convergence, (ii) loss of best solution found, (iii) no
energy source has not been detailed. In earlier works on optimiza- absolute assurance that a genetic algorithm will find a global
tion, none of the papers have brought specific basic information of so optimum and (iv) indefinite or long time for convergence, not
many upcoming new algorithms under one roof which motivates the suitable for real time applications [8].
present one. Furthermore, this paper has focused on the major issues
of the power system i.e. maximized generation and integration of
wind power to grid. In Section 2 the salient features of the algorithms 2.1.1.2. Evolutionary Programming (EP). It is a technique in the field
are narrated in brief followed by their applications in the area of of computation developed by D. B. Fogel in 1990 [9,10]. It starts
wind energy in Sections 3–7. At the end the paper gives an with a population of randomly generated candidate solutions and
exhaustive conclusion and indicates for future research. evolves towards a better solution over a number of generations or
iterations. The main steps of EP are Initialization, Mutation,
Competition and Selection.
2. Salient features of the algorithms

The optimization algorithms are broadly divided in to three

categories according to the under laying principle shown in Fig. 1
namely:

1. Biology based algorithm;

2. Physics based algorithm;
3. Geography based algorithm.

2.1. Biology based

Biology based (Biology-derived) optimization algorithms are an

important part of computational sciences, which are derived from or
based on the analogy to natural evolution and biological activities [5]. Fig. 1. Classification of optimization algorithms.
216 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
2.1.1.3. Evolutionary Strategy (ES). Evolution strategies derive
inspiration from principles of biological evolution. A population, P,
of so-called individuals is assumed. Each individual consists of a
solution or object parameter vector x A Rn (the visible traits) and
further endogenous parameters, s (the hidden traits), and an
associated fitness value, f ðxÞ. In some cases the population contains
only one individual. Individuals are also denoted as parents or
offspring, depending on the context. In a generational procedure
– one or several parents are picked from the population (mating
selection) and new offspring are generated by duplication and
recombination of these parents.
The new offspring undergo mutation and become new mem-
bers of the population.
– Environmental selection reduces the population to its
original size.
Within this procedure, evolution strategies employ the main
principles that are specified and applied in the operation and
algorithms as detailed in [11].
2.1.1.4. Differential Evolution algorithm (DE). It has been proposed
by Storn et al. [12]. DE has been found to be a promising heuristic
optimization algorithm in numeric optimization problems. DE
designed to fulfill the requirement for practical minimization
technique such as consistent convergence to the global minimum
in consecutive independent trials can tackle non-differentiable, non-
linear and multimodal cost functions promptly [13]. Therefore, the
algorithm has gained a great attention since it was proposed.
Unlike simple GA that uses binary coding for representing
problem parameters, DE uses real coding of floating point numbers
[13]. The crucial idea behind DE is a scheme for generating trial
parameter vectors. Basically, DE adds the weighted difference
between two population vectors to a third vector.
The key parameters of control are NP–the population size, CR–
the crossover constant, F–the weight applied to random differen-
tial (scaling factor). It is worth noting that DE's control variables,
NP, F and CR, are not difficult to choose in order to obtain
promising results. Storn et al. [14] have come out with several
rules in selecting the control parameters. Adaptive parameters for
DE have also shown good results for bench mark functions [15].
2.1.1.5. Harmony Search algorithm (HS). Harmony search (HS) [16]
is an evolutionary algorithm that mimics musician's behaviors
such as random play, memory-based play and pitch adjusted play
to get a perfect state of harmony. The classical HS contains three
main operations: random search, memory consideration and pitch
adjustment.
A differential harmony search (DHS) algorithm is proposed by
combining the mechanisms of both harmony search and differential
evolution [147]. First, the pitch adjustment operation of the original
HS is cooperated with the differential mutation operation to enhance
exploitation ability. Second, the memory consideration and the
enhanced pitch adjustment operation are both employed to stren-
gthen the exploration ability. In the DHS, the use of differential
mutation and crossover can enhance the exploitation relative to the
pure HS. In comparison to the pure DE, the DHS may inherit
elements from as many individuals as its number of dimensions
when generating a new individual to enhance the exploration ability.
2.1.2. Swarm based algorithm
A swarm is a large number of homogenous, simple agents
interacting locally among themselves, and their environment, with
no central control to allow a global interesting behavior to emerge.
Swarm-based algorithms have recently emerged as a family of
nature-inspired, population-based algorithms. They are capable of
producing low cost, fast, and robust solutions to several complex
problems. Swarm is used to model the collective behavior of social
nature, such as ant colonies, honey bees, and bird flocks etc.
Although these agents (insects or swarm individuals) are relatively
simple with limited capabilities on their own, with their coopera-
tive interaction they attain tasks necessary for their survival, yet
keeping a distance from other agents. The Swarm based algo-
rithms are detailed as follows.
2.1.2.1. Artificial Immune System(AIS). Artificial immune system
(AIS) mimics biological principles of clone generation, proliferation
and maturation [17]. The main steps of AIS based on clonal selection
principle are activation of antibodies, proliferation and differentiation
on the encounter of cells with antigens, maturation by carrying out
affinity maturation process, eliminating old antibodies to maintain
the diversity of antibodies and to avoid premature convergence,
selection of those antibodies whose affinities with the antigen are
greater [18].
In order to emulate AIS in optimization, the antibodies and
affinity are taken as the feasible solutions and the objective
function respectively. Real number is used to represent the
attributes of the antibodies.
Initially, a population of random solutions is generated which
represent a pool of antibodies. These antibodies undergo prolifera-
tion and maturation. The proliferation of antibodies is realized by
cloning each member of the initial pool depending on their affinity.
In minimization problem, a pool member with lower objective value
is considered to have higher affinity. The proliferation rate is directly
proportional to the affinity of the antibodies. The maturation process
is carried through hyper-mutation which is inversely proportional to
the antigenic affinity of the antibodies. The next step is the applica-
tion of the aging operator. In this operator an antibody is allowed to
remain in the population for at most τB generations. This aging
operator eliminates old antibodies in order to maintain the diversity
of the population and to avoid the premature convergence to local
optima. During the cloning expansion, a clone inherits the age of its
parent and is assigned an age equal to zero when it is successfully
hyper-mutated i.e. when hyper-mutation improves its affinity.
2.1.2.2. Particle Swarm Optimization (PSO). It is proposed by
Kennedy et al. [19]. It is inspired by the paradigm of birds flocking.
It searches for the optimal solution from a population of moving
particles. Each particle represents a potential solution and has a
position in the problem space represented by a position vector xi . A
swarm of particles moves through the problem space, with the
moving velocity of each particle represented by a velocity vector vi .
At each time step, a fitness function ‘F’ representing a quality measure
is calculated by using xi as input. Each particle keeps track of its
individual best position xi;pbest , which is associated with the best
fitness it has achieved so far. Furthermore, the best position among all
the particles obtained so far in the swarm is kept track of as xgbest . This
information is shared by all particles for convergence to global best
solution [20]. Variants of PSO have been detailed in [21] for data
clustering.
2.1.2.3. Bacteria Foraging Optimization algorithm (BFO). It has been
proposed by Passino [22]. This algorithm imitates the foraging behavior
of Escherichia coli bacteria, present in our intestines. Foraging is the
methods for locating, handling, and ingesting food. The members of
the population are called bacteria. The bacteria undergo different
stages such as chemotaxis, swarming, reproduction, and elimination
and dispersal. In the chemotaxis stage, each bacterium can either
tumble or swim. During swarming, each bacterium signals other
bacteria via attractants to swarm (group) together. In reproduction,
 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227 217

the least healthy bacteria die and the healthiest splits into two bacteria solutions are abandoned. An employed bee is a forager associated with
that are then placed in the same location. Further, any bacterium is a certain food source which she is currently exploiting. She memorizes
eliminated from the total set of bacteria just by dispersing it to a the quality of the food source and then after returning to the hive
random location (elimination and dispersal). Chemotaxis and gives a positive feedback by sharing it with other bees waiting there
swarming give exploitation capability where as reproduction and via a peculiar communication called waggle dance. An onlooker bee is
elimination and dispersal aid to exploration capability of the an unemployed bee at the hive which tries to find a new food source
algorithm [23]. using the information provided by employed bees. A scout, ignoring
the other's information, searches around the hive randomly. In nature,
the recruitment of unemployed bees happens in a nearly similar way.
2.1.2.4. Cuckoo Search algorithm (CS). The CS algorithm was
In addition, when the quality of a food source is below a certain level,
proposed by Yang et al. [24] inspired by the typical cunning
it will be abandoned to make the bees explore for new food sources.
breeding behavior of some cuckoo species for survival. The female
Karaboga et al. have compared its performance to its rivals [27].
parasitic cuckoos can imitate the colors and patterns of the eggs of a
The Artificial Bee Colony (ABC) has been successful to solve many
few chosen host species. This reduces the probability of the eggs
optimization problems with constraints [28]. A modified ABC
being abandoned and, therefore, increases their reproductivity. It is
algorithm has been proposed recently [29].
worth mentioning that several host birds engage direct conflict
with intruding cuckoos. In this case, if host birds discover the eggs
are not their own, they will either throw them away or simply
2.1.2.6. Ant Colony Optimization algorithm (ACO). It is proposed by
abandon their nests and build new ones, elsewhere. Parasitic
Dorigo [30]. It is an evolutionary meta-heuristical algorithm based
cuckoos often choose a nest to lay their egg where the host bird
on the collective behavior emerging from the interaction of the
just laid its own eggs but secretly. In general, the cuckoo eggs hatch
different search threads that has proved effective in solving
slightly earlier than their host eggs. Once the first cuckoo chick is
combinatorial optimization problems. The peculiarity of ACO
hatched, his first instinct action is to evict the host eggs by blindly
algorithms is to relate the a priori information about the
propelling the eggs out of the nest. This action results in increasing
structure of a promising solution with an information about the
the cuckoo chick's share of food provided by its host bird.
structure of previously obtained good solutions.
Furthermore a cuckoo chick can imitate the call of host chicks to
ACO is found on the foraging behavior of ants and their indirect
gain access to more feeding opportunity.
communication based on pheromones, and has been applied to
Each egg in a nest represents a solution, and a cuckoo egg
several combinatorial problems. The ACO was inspired from natural
represents a new solution. The aim is to employ the new and
behavior of the ant colonies on how they find the food source and
potentially better solutions (cuckoos) to replace not-so-good
bring them back to their nest by building the unique trail formation.
solutions in the nests. In the simplest form, each nest has one
A hybrid of ACO has been proposed and applied to economic load
egg. The algorithm can be extended to more complicated cases in
despatch considering constraints by Mousa et al. [31].
which each nest has multiple eggs representing a set of solutions
[24]. The CS is based on three idealized rules.
Each cuckoo lays one egg at a time, and dumps it in a randomly
2.1.2.7. Coral Reef Optimization algorithm (CRO). The CRO is a novel
chosen nest. The best nests with high quality of eggs (solutions) will
bio-inspired meta-heuristic for optimization problems, based on an
carry over to the next generations. The number of available host
artificial simulation of the coral reefs' formation and reproduction
nests is fixed, and a host can discover an alien egg with probability
processes. It was first proposed by Salcedo-Sanz et al. [32]. The CRO
pa A [0,1]. In this case, the host bird can either throw the egg away or
algorithm emulates different phases of coral reproduction and fight for
abandon the nest to build a completely new nest in a new location.
space in the reef, and finally produces an efficient algorithm for
For simplicity, the last assumption can be approximated by a
solving difficult optimization problems. This algorithm mainly differs
fraction pa of the n nests being replaced by new nests, having new
in reproduction stage as there are three types of reproduction namely;
random solutions is fixed. For a maximization problem, the quality
broadcast spawning, brooding and budding where as others have only
or fitness of a solution can simply be proportional to the objective
one [33]. The proposed CRO approach can be seen as a cellular-
function. Other forms of fitness can be defined in a similar way to
type evolutionary scheme, with superior exploration, exploitation
the fitness function in genetic algorithms. The performance of the
properties.
CS can be improved by using Levy flights instead of simple random
walk. A Levy flight is a random walk in which the step-lengths are
distributed according to a heavy-tailed probability distribution.
2.1.2.8. Teaching–Learning Based Optimization algorithm (TLBO). Recently,
After a large number of steps, the distance from the origin of the
a new evolutionary method called Teaching–Learning Based
random walk tends to a stable distribution. The improved CS has
Optimization (TLBO) algorithm has been presented [34]. It is
been proposed by Valian et al. [25] where pa and step size can be
based on the effect of the influence of a teacher on the output of
adjusted during run after initialization. There in the basic steps of
learners in a class. TLBO method has the major advantage of not
the CS is summarized as the pseudo-code.
requiring any parameter of the algorithm for its operation with the
exception of the population size and maximum number of
2.1.2.5. Artificial Bee Colony algorithm (ABC). Artificial Bee Colony iterations. Furthermore the algorithm is easily implemented and
(ABC) model is proposed by Karaboga [26]. In this collective requires less computational memory [35].
intelligence search model, the honey bees are categorized as A Modified Teaching–Learning Based Optimization (MTLBO)
employed, onlooker and scout. The employed and unemployed bees algorithm was presented [36]. This new MTLBO method is
search for the rich food sources, which are close to the bee's hive. The obtained by modifying the TLBO algorithm to improve the perfor-
employed bees store the food source information and share the mance for global search. Basically, TLBO algorithm can be divided
information with onlooker bees. The number of employed bees is into two phases: ‘Teacher Phase’ and ‘Learner Phase’. In MTLBO
equal to the number of food sources and also equal to the amount of algorithm the proposed modification uses mutation operations in
onlooker bees. Employed bees whose solutions are poor and cannot be a similar way to Differential Evolution (DE) algorithm, such that
improved through a predetermined number of trials called ‘limit’, now the process of optimization is simulated with another addi-
specified by the user of the ABC algorithm, become scouts and their tional phase: ‘Mutation Phase’.
218 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
In [37] a new modification to TLBO procedure was proposed, in
which the algorithm uses a crossover technique, for improving the
algorithm performance for global search too.
Furthermore, Niknam et al. [38] proposed modification with
two mutation operations as well as two crossover operations to
enhance the ability of the algorithm for both local and global
search exploration adequately, although it may have the disad-
vantage of increasing the difficulty for the selection of the
parameters of the algorithm.
The modified TLBO has been tested in a comparative platform
by Satapathy et al. [39] and has excelled in many cases to other
rivals.
2.1.2.9. Firefly Algorithm (FA). Firefly Algorithm (FA) is invented by
Yang [40] for solving multimodal optimization problem. The
development of FA is based on flashing behavior of fireflies.
There are about 2000 firefly species where the flashes often
unique for a particular species. The flashing light is produced by
a process of bioluminescence where the exact functions of such
signaling systems are still on debate. Nevertheless, two
fundamental functions of such flashes are to attract mating
partners (communication) and to attract potential prey.
For simplicity, the following three ideal rules are introduced in
FA development as given in pseudo-code in [139]: 1) all fireflies
are unisex so that one firefly will be attracted to other fireflies
regardless of their sex, 2) attractiveness is proportional to their
brightness, thus for any two flashing fireflies, the less brighter one
will move towards the brighter one, and 3) the brightness of a
firefly is affected by the landscape of the objective function. For
maximization problem, the brightness can simply be proportional
to the value of the objective or fitness function.
2.1.2.10. Shuffled Frog Leaping Algorithm (SFLA). This is again
another meta-heuristic optimization method [41] which imitates
the behavior of a group of frogs searching for the location of
maximum food available. In this SFLA, the set of virtual frogs called
the population is partitioned into different groups. These groups
are termed to as memeplexes. The memeplexes have different
cultures of frogs, each performing a local search. Within each
memeplex, the individual frog's idea can be biased by the ideas of
other frogs. After a defined number of memetic evolution steps,
ideas are passed between memeplexes in a shuffling process. The
convergence criterion is achieved by repeated steps of local search
and the shuffling process [141].
In the first step of this SFLA, an initial population of F frogs is
randomly generated within the feasible search space. Defining S as
the number of decision variables in optimization problem, position of
the ith frog i.e. each possible solution is represented as,
X i ¼ ðX i1 ; X i2 ; …; X iS Þ. Next, the fitness for ith frog is evaluated. After
this the frogs are sorted in descending order to form m number of
memeplexes, each containing n frogs. Then, within each memeplex, a
procedure to improve the frog with the worst fitness only is applied.
Therefore, the frog with the worst fitness leaps towards the position
of the best frog, as follows:
Di ¼ randð:Þ:ðX b X w Þ
X new
i ¼ X current
i þ Di ; Di; min r Di r Di; max
where randð:Þ is a random number whose value lies between 1and 0.
Di;max and Di;min are the maximum and minimum allowed change in
frog's position respectively. X b and X w are best and worst frog position
respectively. If this process produces a better solution, it will replace the
worst frog. Else, the calculations in (1) and (2) are repeated after
replacing X b by X g (where X g is the global best fitness of frog). If no
improvement occurs then a feasible solution is randomly generated to
replace X w . The computation is carried out for a specific number of
iterations within each memeplex. Then, all frogs are shuffled to promote
global information exchange among the frogs. Local exploration and
global shuffling alternate until the convergence criteria are satisfied. A
binary SFLA has been successfully used in power system problem [42].
2.1.2.11. Pigeon Inspired Optimization (PIO). Pigeon inspired
optimization is a recently proposed population based algorithm
converging to the optimal solution through information sharing
and competing among them to reach the global best as quick as
possible. The evolutionary concept behind it is somewhat similar to
PSO. In this algorithm, homing pigeons search for their homes using
three parameters: magnetic field, sun and landmarks. It has been
further investigated that pigeons have excellent magnetic homing
detection particles placed at their beaks. The parameters of the
optimization, dimension of the function D, population N, map and
compass factor R, number of iterations Nc1max to carry out map and
compass operation and Nc2max to carry out landmark operation are
initialized at first, and then the velocities as well as position are
updated according to the map and compass operator as well as
landmark operator. The details of this algorithm and its use have
been given by Duan et al. [43]. A modified PIO has also been
proposed [44].
2.2. Physics based algorithm
Physics based optimization algorithms are the heuristic algo-
rithms which mimic the physical properties or physical behavior of
the matter or are in the same philosophy as the laws of physics. The
physics based optimization algorithms are classified as follows.
2.2.1. Simulated Annealing (SA)
Simulated Annealing [45] is a powerful general-purpose opti-
mization technique analogous to annealing process used for
crystallization a physical process in glass and metals for hardening
of the material. It has been applied to many optimization problems
specifically to power system as detailed in [46]. This method
converges asymptotically to the global optimum solution with
probability one, provided that certain conditions on the cost
function, the moves, and the annealing schedule are satisfied [47].
2.2.2. Gravitational Search Algorithm (GSA)
The gravitational search algorithm (GSA), one of the recent
additions to heuristic algorithms, was developed by Rashedi et al.
[48] and applied to optimization problems of different functions
[49,146]. GSA is followed by the physical law of gravity and the laws
of motion. In the proposed algorithm, agents are considered as
objects and their performance is measured by their masses. All
these objects attract each other by their gravitational force, and this
force causes a global movement of all objects towards the objects
with heavier masses. Hence, each mass will be informed through
gravitational force and updated by exchanging the information.
2.2.3. Chaotic Optimization Algorithm (COA)
Chaotic optimization algorithms as an emergent method of
ð1Þ
global optimization [50] have attracted attention for the features
of easy implementation, short execution time and robust mechan-
ð2Þ
isms of escaping from local optimum.
Chaos theory is recognized as very useful in many engineering
applications [163]. An essential feature of chaotic systems is that
small changes in the parameters or the starting values for the data
lead to vastly different future behaviors, such as stable fixed points,
periodic oscillations, bifurcations, and ergodicity. These behaviors can
be analyzed based on Lyapunov exponents and the attractor theory.
This sensitive dependence on initial conditions is generally
exhibited by systems containing multiple elements with nonlinear
 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227 219

interactions, particularly when the system is forced and dissipative.
Sensitive dependence on initial conditions is not only observed in
complex systems, but even in the simplest logistic equation.
The application of chaotic sequences can be an interesting
alternative to provide the search diversity in an optimization
procedure. Due to the non-repetition of chaos, it can carry out
overall searches at higher speeds than stochastic ergodic searches
that depend on probabilities. Details about analysis of chaotic
behavior can be found in [51–54].
2.2.4. Intelligent Water Drops algorithm (IWD)
Intelligent Water Drops algorithm (IWD) proposed by Shah-
Hosseini in [55] is a swarm-based nature inspired optimization
algorithm, which has been inspired from natural rivers and how
they find almost optimal path to their destination. A natural river
often finds good paths among lots of possible paths in its ways from
the source to destination. These near optimal or optimal paths follow
from actions and reactions occurring among the water drops and the
water drops with their river beds. In the IWD algorithm, several
artificial water drops cooperate to change their environment in such
a way that the optimal path is revealed as the one with the lowest
soil on its links. The solutions are incrementally constructed by the
IWD algorithm. Consequently, the IWD algorithm is generally a
constructive population based optimization algorithm. This Intelli-
gent Water Drop, IWD for short, has two important properties [56]:
– The soil it carries, denoted by soil (IWD);
– The velocity that it posses, denoted by velocity (IWD).
magnetic particles exhibit both attraction and repulsion to a given
particle based on the irrelative polarity. This dual function can be
useful in optimization to balance exploitation versus exploration.
The particles are placed in a lattice-like structure. In this algorithm,
the particles are attracted/ repelled to/by their neigh-boring parti-
cles in a lattice-like structured population. A basic version of this
concept of interaction that accounted only for long range force of
attraction, without the repulsion, has been tested for 14 numeric
benchmark problems. The long range force of attraction moves the
particles towards the ones with higher magnetism (fitness) similar
to PSO. The attraction force has exploitation parameters, α and ρ,
affecting the convergence speed. The MOA has been applied to
training a multi-layer perception [60] and a traveling salesman
problem [61] also. Unlike PSO where only best particles influence
the search in MOA an operation may discard best particles, thereby
increase chance to explore global optima. The operators in MOA
that introduce diversity are short range repulsion (SRR), explosion
(EXP), hybrid explosion–repulsion (HER) and crossover interaction
MOA(X-MOA). These operators are introduced with several para-
meters tuned by Tayarani et al. [62]. The short range force is
activated only when the distance between two particles is less
than a certain threshold. When the diversity becomes less than a
certain threshold, the population is re-scattered in the search space
by explosion. This acts as global short range force [62]. The HER
MOA uses both repulsion and explosion operators. Crossover and
mutation operators aim to complement the existing basic MOA
operators. Furthermore, the lattice structure also influences the
suitability for a problem, the best one being the cellular structure.

For each IWD, the values of both properties, soil (IWD) and 2.3. Geography based algorithms
velocity (IWD) may change as the IWD flows in its environment.
From the engineering point of view, an environment represents a Geography based optimization algorithms are the meta-
problem that is desired to be solved. A river of IWDs seeks an heuristic algorithm which generate the random solution in the
optimal path for the given problem. geographical search space. The geography based optimization
This environment depends on the problem at hand. In an algorithms are classified as follows.
environment, there are usually lots of paths from a given source
to a desired destination, which the position of the destination may
2.3.1. Tabu Search Algorithm (TS)
be known or unknown. If we know the position of the destination,
Tabu search is a meta-heuristic search method that employs
the goal is to find the best (often the shortest) path from the
local search. It was first suggested in Glover [63] in 1986.The
source to the destination [57]. In some cases, in which the
scheme aims to explore the search space of all feasible solutions by
destination is unknown, the goal is to find the optimum destina-
a sequence of moves. In each the move is carried out in the best
tion in terms of cost or any suitable measure for the problem.
direction among those available. To help the algorithm to escape
Duan et al. [58] in an application have considered an IWD moving
from local (but not global) minima a set of moves are forbidden at
in discrete finite-length steps. From its current location to its next
each iteration. Formally, the main elements of the algorithm are
location, the IWD velocity is increased by the amount nonlinearly
solution representation, cost function, neighborhood, tabu list,
proportional to the inverse of the soil between the two locations. Also,
aspiration criterion, and termination criterion [64].
the IWDs soil is increased by removing some soil of the path joining
the two locations. The amount of soil added to the IWD is inversely
– Solution representation: Every feasible solution to the optimi-
(and nonlinearly) proportional to the time needed for the IWD to pass
zation problem is represented distinctly in the search space.
from its current location to the next location. This duration of time is
– Cost function: A function cost maps each feasible solution to its
calculated by the simple laws of physics for linear motion. Thus, the
optimization cost. The algorithm aims to locate a solution that
time taken is proportional to the velocity of the IWD and inversely
minimizes this value. The solution with the minimum cost among
proportional to the distance between the two locations. Another
those in the neighborhood becomes the new current solution.
mechanism that exists in the behavior of an IWD is that it prefers the
– Neighborhood: It is a function mapping each feasible solution S
paths with low soils on its beds to the paths with higher soils on its
into a set of other solutions. That means new solutions are
beds. To implement this behavior of path choosing, a uniform random
chosen from the neighborhood of the current solution.
distribution among the soils of the available paths is used such that
– Tabu list: A list containing the last T moves carried out, so they
the probability of the next path to choose is inversely proportional to
are forbidden. Always a new solution S with a move contained
the soils of the available paths. The lower the soil of the path, the
in the tabu list cannot be a member of the neighborhood of S.
more chance it has for being selected by the IWD.
This list is also referred to as the short-term memory [64].
– Aspiration criterion: This is a criterion to judge the best move.
2.2.5. Magnetic Optimization Algorithm (MOA) The usual criterion is that the move produces the best solution
The MOA, recently proposed by Tayarani et al. [59], is based on obtained so far. If a tabu move (a move that is contained in the
the principles of magnetism, attraction/repulsion among magnetic tabu list) satisfies this criterion, then the solution obtained by
particles. MOA differs from GSA by the physical property that applying it to the current solution S can be considered to be in
220 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
Fig. 2. Application of optimization algorithms to different module of wind energy.
the neighborhood of S. Thus the best one can influence further
search moves.
– Termination criterion: TS algorithm stops optimization when
the termination criterion is satisfied. The solution with the
minimum cost among those in the neighborhood becomes the
new current solution.
2.3.2. Imperialistic Competition Algorithm (ICA)
Imperialistic Competition Algorithm (ICA) is introduced in
2007, by Gargari et al. [65]. Similar to other population based
ICA starts with initialization. The population is termed countries
here. They are classified into: colonies and imperialists. An empire
is formed by imperialist together with some colonies. ICA is based
on an imperialistic competition to form one empire in which all
the countries have same position. The last country is the best
result and output of optimization problem.
In the proposed (Modified Imperialistic Competition Algo-
rithm) [66], two steps are modified, for better result with different
operating and technical constraints. The first modification is
applied in generating new countries step, in order to initial new
countries with feasible variables. The second modification is
applied to prevent trapping in local optima on assimilation part.
This modification is setting the variables based on their upper and
lower limits while satisfying demand balance constraints.
Most of these algorithms are heuristic population (swarm)
based which take the advantage of exploration and exploitation
through self-adaptation and cooperation which gives better result
to find global optimum. A good informative comparison of some of
the available algorithms and the software incorporating them is
given by Rios et al. [67].
These optimization algorithms have been applied to a number of
areas in power system. This paper has reviewed the algorithms and
focused the application in different areas with respect to wind energy
integration to grid which can broadly be classified as shown in Fig. 2
1. Wind farm design for siting,
2. Optimal placement and sizing of distributed generators,
3. Sizing,
4. Generation scheduling,
5. PID controller parameter tuning,
6. Control of wind energy conversion system.
In Sections 3–7, the application of algorithms to the above
areas with comment on their performance is discussed and finally
Section 8 concludes the review.
3. Wind farm design
3.1. Siting of wind farm
Wind energy is one of the most profitable renewable energy
sources, constituting a proven technology to meet current and
future electricity demands. Most of the operating wind farm
turbines are on land; however an important part of the future
expansion of wind energy is off shore [68].
Offshore wind conditions are favorable with respect to sites on
land, presenting stronger and steadier wind speeds. However, the
advantages with respect to the potential wind resource contrast
with the increments of installation and maintenance costs, which
must be somehow compensated. This reason has motivated
scientist and engineers to focus on optimizing offshore wind farm
project designs, focusing on different aspects such as location [69],
installation, layout [70–73], availability, operation and mainte-
nance [74,75] etc.
3.2. Wind farm layout design
Once suitable wind farm site is established, the next step
should be wind farm design. Wind farm layout is distribution of
wind turbines on the wind farm site. It is affected by several
factors which must be considered: wind direction and wind
energy distribution, wake interactions between wind turbines,
land availability and possibility of construction. After wind
resource is probabilistically determined at a particular location, it
is possible to minimize expected wake effect losses, thus max-
imizing the expected efficient energy production and position the
turbines. When the wind goes through any turbine, a wake effect
is induced downstream decreasing wind speed and increasing
wind turbulence. This produces a reduction of energy production
in all turbines located within the area of influence of the wake. A
number of methods have been applied in order to optimize wind
farm layouts considering the above mentioned factors [76–78].
Ozturk and Norman [71] approach to maximize the profit in
selling electrical energy in a given market. Therefore, in objective
function they use the term which presents the expected profit per
kWh generated by the wind farm. The optimization is carried out in
which a turbine is inserted or removed or re-positioned in order to
maximize the profit defined in objective function. A perturbation is
added to try to avoid local minima. The wake model is very simple
and does not take into account wake interference in the middle of
the wind farm. The heuristic method performs quite fast and allows
the wake model to be independent from the optimization algorithm.
GA has been successfully used to approach the wind farm
optimization problem [70,72,79]. In [70], the position of wind
turbines that provide the maximum energy extraction with the
minimum installation costs is obtained. In [80,82], a square wind
farm is subdivided into a 10  10 grid. GA is utilized to determine
the cells to install turbines so as to minimize the cost per unit
energy subject to constraints of minimum turbine separation and
the wind farm boundaries. In the case of a wind farm, the
individuals are the possible layouts of the turbines; the population
is the assembly of the different individuals. In above studies the
effectiveness of genetic algorithms for solving the wind farm
layout optimization problem is demonstrated. However, due to
the binary coding method of the genetic algorithms used in [70],
turbines could only be installed at the center of selected cells.
Genetic algorithm method to the wind farm layout design problem
for certain number of turbines with several improvements in
calculation of fitness function is done in [80]. Here the positions
of turbines in a wind farm are adjusted freely, instead of being in
the center of each cell in order to reduce the wake effects further
and capture more wind energy.
 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
Donovan [81] created a model based on the generalized vertex
packing problem (GVP) and maximized the power generated in
accordance with the constraints based on the number of turbines,
their proximity and interference.
Elkinton et al. [73] are developing the OWFLO (Offshore Wind
Farm Lay out Optimization) project. Here optimization is carried out
considering energy production (taking into account wake effects,
electrical losses and turbine availability) with offshore wind farm
component cost. This work develops an analysis tool to identify the
major economic hurdles present for micro-siting of offshore wind
farm. In [83], the Unrestricted Wind Farm Layout Optimization
(UWFLO) methodology is proposed. It determines the optimum
wind farm layout and in addition the appropriate selection of
turbines rotor diameter that maximizes the net power generation.
An evolutive algorithm to optimize the wind farm layout onshore
is proposed in [84,85]. The optimization objective is based on a global
wind farm cost model using the initial investment and the present
value of the yearly net cash flow during the entire life span of the
wind farm. In [86] an evolutionary algorithm for optimal positioning
of wind turbines in wind farms for a realistic model is presented. It
includes orography, shape of the wind farm, simulation of the wind
speed and direction, and costs of installation, connection and road
construction among wind turbines. ACO algorithm is introduced for
maximizing the expected energy output in [87].
All optimization algorithms presented here for layout optimiza-
tion are based on heuristic procedures, until the algorithm is unable
to find a better solution. Basically, these methods focus on finding
an acceptable solution in an attempt to capture the global optimum.
But most of the times the solutions obtained do not even hold the
Karush–Kuhn–Tucker optimality conditions [88,89]. The limitation
of layout optimization problem in existing approaches is that
locations of turbines over a predefined grid are investigated which
limits the feasible region of possible locations considerably.
The heuristic programming approach instead of mathematical
programming for layout optimization is adopted mainly due to
non-convex nature of problem and shorter computational time of
the former. Furthermore, mathematical programming depends on
initial condition and global optimum may be skipped.
A combined method, heuristic and nonlinear programming has
been adopted to obtain the best offshore wind farm layout over a pre-
specified area in [68]. The global optimum lies in a convex subregion
that is identified by the mathematical programming solvers. The
initial solution for nonlinear programming is determined heuristically.
The proposed methodology has the following advantages:
– Current state-of-the-art nonlinear mathematical programming
solvers are more reliable, numerically robust, and computa-
tionally efficient.
– Alternative constraints can be including easily, which do not
alter the flow of the methodology.
– The heuristic method used to generate initial solutions is
capable of searching convex subregions. This allows tackling
non-convexities.
– It is easy to include parallelization features in order to increase
computational efficiency and reduce computational costs.
– The final solution holds the Karush–Kuhn–Tucker optimality
conditions.
– It does not require reducing the feasible solution region by
gridding the possible location area [68].
PSO algorithm is employed to optimize the farm layout for
maximizing the total power generation. A robust constraint
handling technique, based on constraint dominance principles
[83,77,90], is employed to deal with the problem constraints.
Performance of PSO, a stochastic search algorithm, to deal with
multimodal problems is significantly better than gradient based
221
algorithms. It is simpler than most other optimization algorithms
to implement and involves fewer user-defined parameters that
need to be adjusted [91,92].
A simple evolution strategy (ES) is applied to optimize the
placement of the turbines in [93]. In general, an ES is effective
because it is easily parallelized and it self-adapts the extent to which
it perturbs decision variables when generating a new potential
solution. However, the application of algorithm given in [93] is
limited to a small number of turbines. A more powerful ES called
CMA-ES (Covariance Matrix Adaption Evolution Strategy) has been
used for the same problem [94]. The limitation of earlier approach is
overcome and it allows optimization of huge layouts for up to 1000
turbines. But it is still computationally expensive and needs weeks to
solve for large wind farm layouts on computing servers.
4. Optimal placement of distributed generation
Most of the wind farms are located in high wind speed regions
that are far off the HV grids but in the vicinity of load in the low
voltage distribution network. Distributed generation (DG) resources
have attracted research due to their positive technical impact on
distribution system reflected in terms of improved distribution
system reliability, reduced customer interruption costs, reduction
in losses and improvement in voltage profile as well as power
quality at the consumer terminal.
The installation of DG at optimum location boosts the perfor-
mance of distribution system as well as presents a cost effective
solution thus giving a new dimension to distribution system
planning. Optimal planning of distribution system with distributed
generation resources is concerned with the sizing of distributed
generators and placement of generators.
Distributed Generators offer a lot of advantages on distribution
system performance [95–104]. The benefits of distributed genera-
tion can be classified into following categories:
A. Technical benefits
– Line losses reduction.
– Improvement in voltage profile.
– Power quality improvement.
– Enhancement in system reliability and security.
B. Economic benefits
– Reduced investments for upgradation of facilities in system.
– Reduced O&M costs of some DG technologies.
– Enhanced productivity.
– Reduced health care costs due to improved environment.
– Reduced fuel costs due to increased overall efficiency.
– Reduced reserve requirements and the associated costs.
– Lower operating costs due to peak shaving.
C. Environmental benefits
– Reduced emissions of pollutants.
– Encouragement to renewable energy based generation.
Griffin et al. [105] have proposed a simple heuristic iterative
approach for DG placement with objective of loss minimization
using B loss coefficient. Minimization of investments and opera-
tion costs is obtained by El-Khattam et al. using the iterative
technique[106].
Nara et al. and Golshan and Arefifar have used TS (Tabu Search)
for optimization of DG size and location. The approach extensively
explores its memory structures to effectively and economically
direct the search to attractive regions in the solution space
[107,108].
For finding out suitable DG size and location Beromi et al. [109]
used PSO with manifold objective of improving voltage profile,
222 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
minimizing losses, reduction of THD at the same time dealing with
the costs.
Falaghi et al. [110] have presented a model to determine
optimal location and size of DGs in a distribution system which
is solved using ACO as the optimization tool.
Wang et al. [111] have used ant colony system algorithm for DG
placement scheme for radial distribution networks and to derive
the optimal recloser as well.
Sutthibun and Bhasaputra [112] have shown the superiority of
SA (Simulated Annealing) for optimal location and size with the
less computing time than genetic algorithm and Tabu search for
multi-objective problem. The DGs placed in the optimal location
work efficiently comparing with single objective.
GA based approach has been reported for sizing and placement
of DG keeping in view of system power loss minimization in
different loading conditions [113].
Borges and Falcao [114] have presented a method for optimal
DG unit allocation and sizing in order to maximize a benefit/cost
relation. Here the benefit is in terms of the reduction of electrical
losses and the cost is dependent on investment and installation.
Constraints to guarantee acceptable reliability level and voltage
profile along the feeders are incorporated.
MTLBO algorithm [35] is adopted to determine the optimal
placement and size of DG units in distribution systems, which improve
the performance for global search as compared to TLBO algorithm.
Differential evolution (DE) is employed for optimal DG capacity
determination for reduction of total system real power losses
[116].
Cuckoo Search algorithm (CS) is presented to evaluate the DG
site and size in the distribution network [117]. Using the CS, the
optimization can be solved efficiently, as there are fewer para-
meters to be fine-tuned in CS than in other optimization algo-
rithms. It can tackle complex optimization problems, and gives
better quality solution and higher precision factor compare to GA
and PSO methods [115].
The Shuffled frog leaping algorithm has been used for the
placement of Distributed Generators (DG) in the radial distribution
systems to reduce the real power losses and cost of the DG [118].
BFOA is employed to minimize the objective function for the
optimal size of DG units at candidate locations [119] and for
solving optimum power flow [120].
As the number of renewable integrated to power supply system
is increasing, optimizing the size and location of active and
reactive power resources like DG and capacitors is also being
optimized [121]. Optimal sizing of renewable power generation
along with battery is emerging as a new area of research where
the optimization algorithms can be applied [122].
5. Generation scheduling
Generating Power at minimum cost to match the demand
satisfying all the constraints of the system is the aim of generation
scheduling or Economic Load Dispatch (ELD). Significant cost
savings can be obtained by small improvements in the optimal
output scheduling can contribute to significant cost savings. There
exist a lot of research works [123,124] to solve this fundamental
problem.
Various types of physical and operational constraints make ELD
for large scale systems, a highly nonlinear constrained optimiza-
tion problem. In many cases, generally the cost function of the
generator is approximated as a single quadratic function. However,
higher order nonlinear and discontinuities are seen in real input–
output characteristics due to valve-point effects in fossil fuel
burning plants. Besides, the units may have prohibited operating
zones due to faults in machines or associated auxiliaries, such as
boilers, feed pumps, etc. leading to instabilities in certain ranges of
the unit loading [125]. Furthermore, the operating range for online
units is actually restricted by their ramp-rate limits. These con-
straints transform an ELD problem into a hard non-convex
optimization problem [126].
With the integration of wind energy into existing power
systems, it introduces many complexities to operation and control
due to its nonpredictable and nondispatchable wind power nature
[127,128].
Lambda iteration method (LIM), gradient search, linear pro-
gramming, quadratic programming, dynamic programming, New-
ton based methods etc. are some of the optimization methods to
solve the classical ELD problem [129]. Several heuristic non-
classical techniques have also been reported for the ELD problem.
These include evolutionary programming [130], genetic algorithm
(GA) [131,132], hybrid EP combined with sequential quadratic
programming (SQP) [133], hybrid particle swarm optimization
(PSO) combined with the SQP method (PSO-SQP) [134], Adaptive
PSO [135], hybrid GA-simulated annealing (SA) [136], hybrid GA
[137], and pattern search method [138].
5.1. Economic load dispatch problem formulation
Economic Load Dispatch (ELD) Problem formulation is not an easy
task since there are a lot of factors need to be considered. Minimizing
the selected objective functions while maintaining an acceptable
system performance in terms of generator capability limits and the
output of the compensating devices solves ELD problem.
The objective function is to minimize the overall generating
cost, Ct. This cost function of plant output:
X
ng
Ct ¼ Ci
i¼1
X
ng
¼ αi þβi P i þ γ i P 2i ð3Þ
i¼1
Now it is subjected to the following equality and inequality
constraints [129]:
X
ng
P iðminÞ ¼ P D þ P L ð4Þ
i¼1
P iðminÞ rP i rP iðmaxÞ i ¼ 1; 2; …; ng ð5Þ
th
where P i is the production of i plant, P D is the total load demand,
P L is the total loss and ng is the total number of generating plants
and αi ; βi ; …; γ i are the cost coefficients [129].
Many optimization algorithms applied to solve the ELD pro-
blem are given below.
Firefly Algorithm (FA) applies to solve the Economic Load
Dispatch problem [139].
In this regard, this paper proposes, craziness based shuffled
frog leaping optimization algorithm (CRSFL) to solve the ELD
problem incorporating additional physical and operational con-
straints. The ELD problem is formulated by incorporating wind
power in power balance equation [140].
A new approach is proposed to solve non-smooth ELD problem
with valve-point effect using intelligent water drop (IWD) algo-
rithm [142]. The later technique imitates some of the processes that
happen in nature between the water drops of a river and the soil of
the river bed. The performance, effectiveness, and robustness of the
proposed method are accessed via intensive testing and comparison
of results with other methods reported in recent literatures [141].
GSA described earlier has been successfully applied in solving
various nonlinear functions [143,144]. The results obtained therein
confirm the high performance and efficiency of the proposed
 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
method. GSA has a flexible and well-balanced mechanism to
enhance exploration and exploitation abilities [144]. GSA method
has been proposed for economic dispatch problem with valve-
point effect for 3 nos. and 13 nos. of units test systems in [145].
A hybrid DE has been proposed for this problem [146] with
repair procedure and three simple selection rules to handle the
constraints of the ELD problem. The simulation results and
comparisons to some existing algorithms based on some testing
problems with various constraints including valve point effects,
multi-fuels, ramp rate limit and prohibited operation zones
demonstrate the effectiveness of the proposed DHS [147].
The CSA is very effective in solving non-convex and large-scale
ED problems reported in [148].
Different test cases with different wind energy penetration
levels using the MICA are applied. The results are compared with
other methods, which have been applied to the GS generation
scheduling problem. The results evaluate the proposed MICA
capability in solving GS problem.
In comparison with other algorithms, such as PSO, MICA gives
better results. Less run time results in the fast convergence
characteristic. This is due to modification in initialing and assim-
ilating stages of algorithm. The new assimilation mechanism,
which changes the assimilation direction, checks the algorithm
from dropping in local minimums [66]. But still PSO has been
applied in similar problems in the latest work [149].
6. PID controller parameter tuning
PID controller represents the simplest feedback controller from
a modern perspective [150,151]. Many tuning methods for PID
controllers are found in the literatures. Starting from Ziegler
Nichols (ZN) method to modern ones like ANN, Fuzzy, Bio-
inspired evolutionary computation algorithms, sliding mode,
robust design via some norm, Linear Matrix Inequality (LMI)
technique etc. [152–154] to obtain the optimal P, I, D gains for a
desired application performance index. Here, the objective of the
optimization algorithm is to optimize the gains of the three term
controller for the given process.
Particle swarm optimization (PSO) algorithm used to find the
optimal parameters of the various PI controllers yield excellent
results [20].
For optimal control of RO (Reverse Osmosis) plant with mini-
mum overshoot and fast settling time compared with conven-
tional tuning method, Kim et al. [155] proposed an improved GA
method for PID controller. Yin et al. [156] have successfully used
GA to tune PID controller for low damping, and slow response
plant. Application of GA for optimization of PID parameters to
control a single-link flexible manipulator in vertical motion has
been explored by Zain et al. [157]. Results of simulation revealed
that PID parameters chosen enable the system to target well in
reducing vibration at the end-point of the manipulator.
GAs with the abilities for global optimization and good robust-
ness, and without knowing anything about the underlying mathe-
matics, are expected to overcome the shortcomings of the
traditional PID tuning process. GA is found to be more acceptable
for industrial control [158].
Ruijun [159] predicted the performance of DE and PSO in
optimizing the PID controller parameter in the 1st order process.
They found that DE is more robust (with respect to reproducing
constant results in different runs) than PSO. Youxin and Xiaoyi [160]
have applied DE algorithm in tuning the PID controller for electric-
hydraulic servo system of parallel platform. Simulation results show
the controlled system has satisfactory response and the proposed
parameter optimum method is an effective tuning strategy. Bingul
[161] have demonstrated the usefulness of DE in tuning the PID
223
controller for unstable and integrating processes time delay, where
it produces smaller settling time with minimum overshoot.
Chaotic optimization algorithms, which have the features of easy
implementation, short execution time and robust mechanisms of
escaping from local optimum, are a promising tool for engineering
applications. In this paper [162], a tuning method for determining
the parameters of PID control for an automatic regulator voltage
(AVR) system using a chaotic optimization approach based on Lozi
map is proposed. Damping controllers with renewable penetrating
the power system has also effectively tuned by PSO [163] and can
be done with other algorithms as well.
DC motor speed control using the ACO and ABC algorithm has
been explored [164]. Here, in this controller, ACO algorithm is used
to optimize the gains and the values are applied into the controller
of the plant. The objective of this algorithm is to optimize the gains
of the PID controller for the given plant. The design requirements of
minimum overshoot, rise time, steady state error and settling time
in the step response of the closed loop system are better as
compared to the conventional method.
The coordinated PI design in control system for 7 nos. PI
controllers in wind energy conversion system (WECS) is carried
out successfully using a bacterial foraging optimization technique
scheme in [165]. In another application hybrid of BFO and PSO has
been proposed for tuning fractional order PID controller [166].
7. Control of wind energy conversion system
Wind turbine technology has undergone rapid development in
response to the demands for increased use of renewable energy
[167]. Powerful control systems are required to achieve control of
the (WECS) to reduce cost, improve reliability on the other hand
lessen the loads on the different parts [168–170].
In recent years, many researchers have worked on control of WECS
with objective to improve quality and quantity of the produced
renewable energy from the wind turbine. Controllers such as PI
regulator [171,172], optimal control in LQ [173], and LQG form [174]
have been developed. These control strategies which use the pitch angle
as a control input give acceptable results for rotor speed regulation, but
showed poor performances in power regulation. In [175], it was shown
that the generator torque alone is able to regulate the electrical power
in an acceptable way. However, large variation of the rotor speed is
observed that are not desirable for the wind turbine structure.
PSO has been applied to solve the control problem of energy
generation for MPPT (Maximum Power Point Tracking) for both
fixed-speed and variable-speed wind turbines, given that histor-
ical data of wind speed are limited or insufficient. In MPPT the
control objective is to determine the optimum rotor speed of a
fixed speed wind turbine and the optimum tip-speed ratio of a
variable speed wind turbine, in such a way that annual energy
generation is maximized [176].
Wind turbines can operate with either fixed speed (actually within
a speed range about 1%) or variable speed. Pitch-adjusting variable-
Fig. 3. Control strategy of wind turbine [177].
224 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
speed wind turbines have become the dominating type of yearly
installed wind turbines in recent years. Several reasons, such as the
reduction of both the mechanical structure stresses and the acoustic
noise and the possibility to control active and reactive power, have
driven the choice for variable speed operation of wind turbines.
Wind turbine operation can be divided in two different opera-
tion regions as shown in Fig. 3. The first region describes wind
turbine operation is below the rated wind speed. In this region, the
wind power is lower than the rated power output of wind turbine
generator. The second operation region describes wind turbine
operation during above the rated wind speed. In this region wind
power is greater than the rated power output of wind turbine
generator. In the first zone, the main task of control system is to
maximize the power output of wind turbine by maximizing wind
power capture. Variable speed wind turbine has been developed to
maximizing wind power capture. Many control strategies are
applied to variable speed wind turbine [177].
In wind speed lower than rated value in zone 1, the speed
controller can continuously adjust the speed of the rotor to
maintain the tip speed ratio constant at the level which gives
the maximum power coefficient there by appreciably increasing
the efficiency of the turbine. The generator is controlled by power
electronic equipment, which makes it possible to control the rotor
speed. In this way the power fluctuations caused by wind varia-
tions can be more or less absorbed by changing the rotor speed
and thus power variations originating from the wind conversion
and the drive train can be reduced [178].
Pitch angle regulation is necessary in conditions above the
rated wind speed in zone 2 when the rotational speed is to be kept
constant. Small changes in pitch angle can have a dramatic effect
on the power output [179]. Therefore, control systems for variable
speed wind turbines should continue to evolve towards more and
more effective and innovative solutions, based on soft-computing
methodologies, such as artificial intelligence and GA.
Several control strategies have been applied, among others,
robust digital control [180] and intelligent control based on RBFNN
(Radial Basic Function Neural Network) [181]. A hybrid technique
as improved Elma Neural Network based Modified Particle Swarm
Optimization has been proposed in [182].
In [183], Belghazi et al. used the GA as controller; differently
such as an optimization technique for the extraction of the
maximum power from the turbine.
In [177], a maximum output power tracking (MOPT) strategy of
WECS is developed to tackle both operation regions. The maximum
output power is obtained by adjusting the generator speed and
pitch angle of wind turbines. Generator speed is controlled using PI-
FLC (PI-Fuzzy Logic Controller) based on direct field oriented control
(DFOC) method. Pitch angle of wind turbine is controlled using
RENN (Recurrent Elman Neural Network). Control strategy is shown
in Fig. 3. In zone 1, the wind speed is less than the rated
ðv1 r ν r v2 ), while in zone 2 the wind speed is higher than the
rated for wind turbine ðv2 rν rv3 ). For zone 1, the generator speed
ωm is controlled at maximum power coefficient C p_max with pitch
angle β constant, such that the maximum output power can be
obtained at each wind speed variations. For zone 2 the generator
speed ωm is maintained constant and pitch angle β is controlled to
lower the power coefficient, to limit the output power of wind
turbine with the rating of maximum power rating of the generator.
8. Conclusion
A number of optimization techniques have been presented here
with the basic parameters and relative merits and demerits. It reports
various such techniques in wind turbine for achieving the maximum
power output. This comprehensive study however claims to be
although incomplete but indicates sufficient conceptual view point
on many optimization methods and their potential applications.
Many operational and planning problems in wind energy
integration to existing power system can be solved with the
techniques described here. In addition, technical details with
problem formulations are provided in some of the cases to attract
the reader to perform similar study.
PSO shares many structure similarities with GA but have faster
convergence and simpler implementation. However, any optimization
technique cannot be said the best or worst based on one application.
For some application one may be better than other. The variants of
MOA like SR-MOA and HER-MOA have shown better promising results
for larger size of problems for multimodal functions as compared to
GA, PSO and DE [63]. But this has not been applied to area of siting of
large wind farm where GA fails and becomes slow and computation-
ally expensive. This is a possible area of application of this new
algorithm. New methods are again proposed every day to this effect.
Hybrid algorithms can be developed combining more than one of the
existing algorithms to suit the complexity of the problem. Behavior of
similar species is excluded in this review for the sake of brevity.
Most of the software tools and the optimization algorithms
available are for analysis and planning. But in real time application
to power systems where the action should be taken within
milliseconds, the considerable computational time taken for opti-
mization has been a demerit. Application of parallel processing in
the area of wind energy is rare. If applied, this can improve
computational speed with existing hardware configuration [184].
Reliability of power supply in power system is emerging as a
key issue as renewable sources like wind and sun are dependent
on weather. Therefore, optimal sizing of renewable generators
with stand by supply/storage has significant research potential.
Lastly, this review explores the outcomes of many related
research works in the area of wind energy integration to grid
and has the main focus on the objectives such as: maximization of
annual energy, minimization of installation and operational cost;
internal control to maximize power and reduction of mechanical
loads on the components. However, inept handling of the algo-
rithms may lead to no significant result.
Acknowledgment
We thank the editor and the anonymous reviewer whose com-
ments on an earlier version of this survey have helped to significantly
improve the presentation and the content of this survey.
References
[1] Herbert GMJ, Iniyan S, Sreevalsan E, Rajapandian S. A review of wind energy
technologies. Renew Sustain Energy Rev 2007;11(6):1117–45.
[2] Banosa R, Manzano-Agugliarob F, Montoyab FG, Gila C, Alcaydeb A, Gomezc
J. Optimization methods applied to renewable and sustainable energy: a
review. Renew Sustain Energy Rev 2011;15:1753–66.
[3] Salcedo-Sanz S, Saavedra-Moreno B, Paniagua-Tineo A, Prieto L, Portilla-
Figueras A. A review of recent evolutionary computation-based techniques in
wind turbines layout optimization problems. Cent Eur J Comput Sci 2011;1
(1):101–7.
[4] Iqbal M, Azam M, Naeem M, Khwaja AS, Anpalagan A. Optimization
classification, algorithms and tools for renewable energy: a review. Renew
Sustain Energy Rev 2014;39:640–54.
[5] Yang XS. Ch. 32: Biology-derived algorithms in engineering optimization. In:
Handbook of bio-inspired algorithms and applications. Chapman & Hall/CRC
Computer & Information Science Series; 2005. p. 585–96. (eds Olarius &
Zomaya)
[6] Neri F, Carlos C. Memetic algorithms and memetic computing optimization:
a literature review. Swarm Evol Comput 2012;2:1–14.
[7] Saad MS, Jamaluddin H, Darus IZ. PID controller tuning using evolutionary
algorithms. WSEAS Trans Syst Control 2012;7(4):139–49.
[8] Vanitha M, Thanushkodi K. Solution to economic dispatch problem by
differential evolution algorithm considering linear equality and inequality
constraints. Int J Res Rev Electr Comput Eng 2011;1(1):21–6.
 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
[9] Sinha N, Chakrabarti R, Chattopadhyay PK. Fast evolutionary programming
techniques for short-term hydrothermal scheduling. Electr Power Syst Res
2003;66(2):97–103.
[10] Bäck T, Schwefel H-P. An overview of evolutionary algorithms for parameter
optimization. Evol Comput 1993;1(1):1–23.
[11] Hansen N, Arnold DV and Auger A. Evolution strategies; 5 Apr 2013.
[12] Storn R, Price K. Differential evolution—a simple and efficient heuristic for
global optimization over continuous spaces. J Glob Optim 1997;11
(4):341–59.
[13] Jayakumar, Thanushkodi. Steady state security and economic power dispatch
using multiobjective evolutionary programming. Int J Appl Eng Res 2008;3
(8):1073–85.
[14] Storn, R. On the usage of differential evolution for function optimization. In:
Proceedings of the fuzzy information processing society. Berkeley (CA, USA);
1996. p. 519–23.
[15] Brest J, Greiner S, Boskovic B, Mernik M, Zumer V. Self-adapting control
parameters in differential evolution: a comparative study on numerical
benchmark problems. IEEE Trans Evol Comput 2006;10(6):646–57.
[16] Geem ZM. Music-inspired harmony search algorithm: theory and applica-
tions. Berlin: Springer; 2009.
[17] Basu M. Artificial immune system for fixed head hydrothermal power
system. Energy 2011;36(1):606–12.
[18] Basu M. Artificial immune system for combined heat and power economic
dispatch. Int J Electr Power Energy Syst 2012;43(1):1–5.
[19] Kennedy J, and Eberhart R. Particle swarm optimization. In: Proceedings of
IEEE international conference on neural networks, vol. 4, no. 2; 1995.
p. 1942–8.
[20] Qiao W, Venayagamoorthy GK, and Harley RG. Design of optimal PI
controllers for doubly fed induction generators driven by wind turbines
using particle swarm optimization. In: Proceedings of IEEE international
joint conference on neural networks; 2006. p. 1982–7.
[21] Rana S, Jasola S, Kumar R. A review on particle swarm optimization
algorithms and their applications to data clustering. Artif Intell Rev
2011;35(3):211–22.
[22] Passino Kevin M. Bacterial foraging optimization. Int J Swarm Intell Res
2010;1(1):1–16.
[23] Bensenouci Ahmed. PID controllers design for a power plant using bacteria
foraging algorithm. In: Proceedings of IEEE Saudi international electronics,
communications and photonics conference (SIECPC); 2011. p. 1–6.
[24] Yang XS, Deb S. Engineering optimisation by cuckoo search. Int J Math Model
Numer Optim 2010;1(4):330–43.
[25] Valian E, Mohanna S, Tavakoli S, Haghi A. Improved cuckoo search for
reliability optimization problems. Comput Ind Eng 2013;64(1):459–68.
[26] Karaboga D. Artificial bee colony algorithm. Scholarpedia 2010;5(3):6915.
[27] Karaboga D, Basturk B. On the performance of artificial bee colony (ABC)
algorithm. Appl Soft Comput 2008;8(1):687–97.
[28] Abu-Mouti SF, El-Hawary ME. Optimal distributed generation allocation and
sizing in distribution systems via artificial bee colony algorithm. IEEE Trans
Power Deliv 2011;26(4):2090–101.
[29] Akay B, Karaboga D. A modified artificial bee colony algorithm for real-
parameter optimization. Inf Sci 2012;192:120–42.
[30] Dorigo, M. Ant colony optimization and swarm intelligence. In: Proceedings
of the 5th international workshop, ANTS 2006, vol. 415. Brussels (Belgium);
4–7 Sept 2006.
[31] Mousa AAA. Hybrid ant optimization system for multi-objective economic
emission load dispatch problem under fuzziness. Swarm Evol Comput
2014;18:11–21.
[32] Salcedo-Sanz S, Del Ser J, Landa-Torres I, Gil-López S, Portilla-Figueras A. The
coral reefs optimization algorithm: a novel meta-heuristic for solving hard
optimization problems. Sci World J 2014:1–15.
[33] Salcedo-Sanz S, Gallo-Marazuela D, Pastor-Sánchez A, Carro-Calvo L, Portilla-
Figueras A, Prieto L. Offshore wind farm design with the coral reefs
optimization algorithm. Renew Energy 2014;63:109–15.
[34] Rao RV, Savsani VJ, Vakharia DP. Teaching-learning-based optimization: a
novel method for constrained mechanical design optimization problems.
Comput Aided Des 2011;43:303–15.
[35] Martín García JA, Gil Mena AJ. Optimal distributed generation location and
size using a modified teaching–learning based optimization algorithm. Int J
Electr Power Energy Syst 2013;50:65–75.
[36] Hosseinpour H, Niknam T, Taheri SI. A modified TLBO algorithm for place-
ment of AVRs considering DGs. In: Proceedings of the 26th international
power system conference; 2011. p. 1–8.
[37] Azizipanah-Abarghooee R, Niknam T, Roosta A, Malekpour AR, Zare M.
Probabilistic multi-objective wind-thermal economic emission dispatch
based on point estimated method. Energy 2012;37:322–35.
[38] Niknam T, Fard AK, Baziar A. Multi-objective stochastic distribution feeder
reconfiguration problem considering hydrogen and thermal energy produc-
tion by fuel cell power plants. Energy 2012;42:563–73.
[39] Satapathy Suresh Chandra, Naik Anima. Modified teaching–learning-based
optimization algorithm for global numerical optimization—a comparative
study. Swarm Evol Comput 2014;16:28–37.
[40] Yang XS. Firefly algorithms for multimodal optimization. In: Proceedings of
the stochastic algorithms: foundation and applications, SAGA, 5792; 2009.
p. 169–78.
225
[41] Huynh TH. A modified shuffled frog leaping algorithm for optimal tuning of
multivariable PID controllers. In: Proceedings of IEEE international confer-
ence on industrial technology, IEEE ICIT; 2008. p. 1–6.
[42] Gomez-Gonzalez M, Ruiz-Rodriguez FJ, Jurado F. A binary SFLA for probabil-
istic three-phase load flow in unbalanced distribution systems with techni-
cal constraints. Int J Electr Power Energy Syst 2013;48:48–57.
[43] Duan H, Qiao P. Pigeon-inspired optimization: a new swarm intelligence
optimizer for air robot path planning. Int J Intell Comput Cybern 2014;7
(1):24–37.
[44] Sun, H and Duan, H. PID controller design based on prey-predator pigeon-
inspired optimization algorithm. In: Proceedings of 2014 IEEE international
conference on mechatronics and automation. Tianjin (China); August 3–6
2014. p. 1416–20.
[45] Kirkpatrick S, Gelatt CD, Vecchi MP. Optimization by simulated annealing. Sci
J 1983;220(4598):671–80.
[46] Deeb N. Simulated annealing in power systems. In: Proceedings of IEEE
international conference on man systems and cybernetics; 1992. p. 1086–9.
[47] Tiwari A,Ajjarapu V. Optimal allocation of dynamic VAR for enhancing
stability and power quality. In: Proceedings of 2008 IEEE power and energy
society general meeting—conversion and delivery of electrical energy in the
21st century; 2008. p. 1–7.
[48] Rashedi E, Nezamabadi-pour H, Saryazdi S. GSA: a gravitational search
algorithm,. Inf Sci 2009;179:2232–48.
[49] Rashedi E, Nezamabadi-Pour H, Saryazdi S. Filter modeling using gravita-
tional search algorithm. Eng Appl Artif Intell 2011;24(no. 1):117–22.
[50] Yan XF, Chen DZ, Hu SX. Chaos-genetic algorithms for optimizing the
operating conditions based on RBF-PLS model. Comput Chem Eng 2003;27
(10):1393–404.
[51] Lozi R. Un attracteur estrange du type attracteur de Henon. J Phys Colloq
1978;39(C5):9–10.
[52] Strogatz SH. Nonlinear dynamics and chaos. Massachussets: Perseus Publish-
ing; 2000.
[53] Parker TS, Chua LO. Practical numerical algorithms for chaotic system. Berlin,
Germany: Springer; 1989.
[54] Alligood KT, Sauer TD, Yorke JA. Chaos: an introduction to dynamical
systems. London, UK: Springer; 1996.
[55] Shah-Hosseini H. Optimization with the nature-inspired intelligent water
drops algorithm. Int J Intell Comput Cybern 2008;1(2):193–212.
[56] Shah-Hosseini H. Ch 16: Optimization with the nature-inspired intelligent
water drops algorithm. In: dos Santos WP, editor. Evolutionary Computation;
2009. p. 297–320.
[57] Khan K, Ali Z, Philip N, Deane G, Sahai A. An intelligent water droplet-based
evaluation of health oriented distance learning. J Adv Inf Technol 2013;4
(2):91–100.
[58] Duan H, Liu S, Wu J. Novel intelligent water drops optimization approach to
single UCAV smooth trajectory planning. Aerosp Sci Technol 2009;13
(8):442–9.
[59] Tayarani, NMH, and Akbarzadeh-T, MR. Magnetic optimization algorithms a
new synthesis. In: Proceedings of IEEE congress on evolutionary computa-
tion, CEC 2008 (IEEE World Congress on Computational Intelligence); 2008.
p. 2659–64.
[60] Mirjalili S, and Sadiq AS. Magnetic optimization algorithm for training multi
layer perceptron. In: Proceedings of IEEE 3rd international conference on
communication software and networks (ICCSN); 2011. p. 42–6.
[61] Ismail MM, Zakaria MI, Abidin AFZ, Juliani JM, Lit A, Mirjalili S, Nordin NA,
and Saaid MFM. Magnetic optimization algorithm approach for travelling
salesman problem. In: Proceedings of the world academy of science,
engineering and technology; 2012. p. 1393–7.
[62] Tayarani NMH, Akbarzadeh-T M-R. Magnetic-inspired optimization algo-
rithms: operators and structures. Swarm Evol Comput 2014;19:82–101.
[63] Glover F. Heuristics for integer programming using surrogate con-straints.
Decis Sci 1977;8:156–66.
[64] Glover F, Taillard E, de Werra D. A user's guide to tabu search. Ann Oper Res
1993;41(1):3–28.
[65] Gargari EA, Lucas C. Imperialist competitive algorithm: an algorithm for
optimization inspired by imperialistic competition. In: Proceedings of IEEE
congress on evolutionary computation; 2007. p. 4661–7.
[66] Mahari A, Zare K. A solution to the generation scheduling problem in power
systems with large-scale wind farms using MICA. Int J Electrl Power Energy
Syst 2014;54:1–9.
[67] Rios LM, Sahinidis NV. Derivative-free optimization: a review of algorithms
and comparison of software implementations. J Glob Optim 2013;56
(3):1247–93.
[68] Perez B, Roberto M, Raúl G. Offshore wind farm layout optimization using
mathematical programming techniques. Renew Energy 2013;53:389–99.
[69] Espejo A, Mínguez R, Tomás A, Menendez M, Mendez JM, Losada IJ.
Directional calibrated wind and wave reanalysis databases using instru-
mental data for optimal design of off-shore wind farms. In: Proceedings of
OCEANS 2011, IEEE-Spain; 2011. p. 1–9.
[70] Mosetti G, Poloni C, Diviacco B. Optimization of wind turbine positioning in
large wind farms by means of a genetic algorithm. J Wind Eng Ind Aerodyn
1994;51(1):105–16.
[71] Ozturk UA, Norman BA. Heuristic methods for wind energy conversion
system positioning. Electr Power Syst Res 2004;70(3):179–85.
[72] Grady SA, Hussaini MY, Abdullah MM. Placement of wind turbines using
genetic algorithms. Renew Energy 2005;30(2):259–70.
226 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
[73] Elkinton CN, Manwell JF, McGowan JG. Offshore wind farm layout optimiza-
tion (OWFLO), project: preliminary results, In: Proc. of the 44th AIAA
Aerospace Sciences Meeting and Exhibit, American Society of Mechanical
Engineering, Reno, NV; January 2006.
[74] LWMM Rademakers, Braam, H, Zaaijer, MB, van Bussel, GJW. Assessment and
optimization of operation and maintenance of offshore wind turbines. In:
Proceedings of the European wind energy conference (EWEC). Madrid
(Spain); 2003.
[75] Rašuo BP, Bengin AČ. Optimization of wind farm layout. FME Trans 2010;38
(3):107–14.
[76] González JS, Payán MB, Santos JMR, Longatt FG. A review and recent
developments in the optimal wind-turbine micro-siting problem. Renew
Sustain Energy Rev 2014;30:133–44.
[77] Chowdhury S, Jie Z, Achille M, Luciano C. Optimizing the arrangement and
the selection of turbines for wind farms subject to varying wind conditions.
Renew Energy 2013;52:273–82.
[78] Mínguez R, Martínez M, Castellanos OF, Guanche R. Component failure simula-
tion tool for optimal electrical configuration and repair strategy of off-shore wind
farms. In: Proceedings of OCEANS, 2011 IEEE-Spain; 2011. p. 1–10.
[79] Moreno SV. Wind farm optimisation. Leicestershire: Centre for Renewable
Energy Systems Technology, Loughborough University; 2001.
[80] Rašuo, B, Bengin, A, and Veg, A. On aerodynamical optimization of wind-
farm layout by genetic algorithm method. In: Proceedings of the 6th
international congress on industrial and applied mathematics – ICIAM
2007 (GAMM 2007). Zurich (Switzerland); 16–20 Jul 2007. p. 98.
[81] Donovan S, Sherali HD, Smith JC. Wind farm optimization, In: Proceedings of
the 40th annual ORSNZ conference, 2005, p.196-205.
[82] Kusiak A, Song Z. Design of wind farm layout for maximum wind energy
capture. Renew Energy 2010;35(3):685–94.
[83] Chowdhury S, Zhang J, Messac A, Castillo L. Unrestricted wind farm layout
optimization (UWFLO): investigating key factors influencing the maximum
power generation. Renew Energy 2012;38:16–30.
[84] Elkinton CN, Manwell JF, McGowan JG. Algorithms for offshore wind farm
layout optimization. Wind Eng 2008;32(1):67–84.
[85] Serrano J, González AG, Castro J, Riquelme J, Burgos M. Optimization of wind
farm turbines layout using an evolutive algorithm. Renew Energy
2010;35:1671–81.
[86] Saavedra-Moreno B, Salcedo-Sanz S, Paniagua-Tineo A, Portilla-Figueras A.
Seeding evolutionary algorithms with heuristics for optimal wind turbines
positioning in wind farms. Renew Energy 2011;36(11):2838–44.
[87] Eroglu Y, Ulusam Y. Design of wind farm layout using ant colony algorithm.
Renew Energy 2012;44:53–62.
[88] Vanderplaats GN. Numerical optimization techniques for engineering design
with applications. . New York: McGraw-Hill; 1984.
[89] Bazaraa MS, Sherali HD, Shetty CM. Nonlinear programming: theory and
algorithms. 2nd ed.. New York: John Wiley & Sons; 1993.
[90] Deb KK, Pratap A, Agarwal S, Meyarivan T. A fast and elitist multi-objective
genetic algorithm: NSGA-II. IEEE Trans Evol Comput 2002;6(2):182–97.
[91] Chowdhury S, Jie Z, Achille M, Luciano C. Unrestricted wind farm layout
optimization (UWFLO): investigating key factors influencing the maximum
power generation. Renew Energy 2012;38(1):16–30.
[92] Wagner M, Day J, Neumann F. A fast and effective local search algorithm for
optimizing the placement of wind turbines. Renew Energy 2013;51:64–70.
[93] Kusiak A, Song Z. Design of wind farm layout for maximum wind energy
capture. Renew Energy 2010;35(3):685–94.
[94] Wagner M, Veeramachaneni K, Neumann F, O’Reilly UM. Optimizing the
layout of 1000 wind turbines. European wind energy association annual
event; 2011. p. 205–9.
[95] Chiradeja, P. Benefits of distributed generation: a line loss reduction analysis.
In: Proceedings of IEEE transmission and distribution conference and
exhibition: Asia and Pacific, 2005 IEEE/PES; 2005. p. 1–5.
[96] Kejun Q, Zhou C, Yuan Y, Xiaodan S and Allan M. Analysis of the environ-
mental benefits of distributed generation. In: Proceedings of IEEE power and
energy society general meeting conversion and delivery of electrical energy
in the 21st century; 2008. p. 1–5.
[97] Gil HA, Joos G. Models for quantifying the economic benefits of distributed
generation. IEEE Trans Power Syst 2008;23:327–35.
[98] Chiradeja P, Ramakumar R. An approach to quantify technical benefits of
distributed generation. IEEE Trans Energy 2004;19(4):764–73.
[99] Waseem, I, Pipattanasomporn, M and Rahman, S. Reliability benefits of
distributed generation as back up source. In: Proceedings of IEEE power and
energy society general meeting, PES-2009; Publication year 2009. p. 1–8.
[100] Momoh, JA, Yan, X, and Boswell, GD. An approach to determine distributed
generation benefits in power networks. In: Proceedings of the 40th NA
power symposium; Publication year 2008. p. 1–7.
[101] Chiradeja, P and Ramakumar, R. A review of distributed generation and storage.
In: Proceedings of the frontiers of power conference; 1998. p. VIII 1–VIII.
[102] Dugan RC and Price SK. Issues for distributed generations in the US. In:
Proceedings of IEEE power engineering society winter meeting, vol. 1; Jan
2002. p. 121–6.
[103] Dugan RC, Mcdermont TE. Power quality and distributed generation. IEEE Ind
Appl Mag 2002;9(5):19–25.
[104] Ramakumar, R. Technology and economic market integration of DG. In:
Proceedings of the frontiers of power conference; 2001. p. X 1–4.
[105] Griffin, T, Tomsovic, K, Secrest, D, and Law, A. Placement of dispersed
generations system for reduced losses. In: Proceedings of the 33rd Hawaii
international conference on system science. Maui (HI); 4–7 Jan 2000. p. 1–9.
[106] El-Khattam W, Bhattacharya K, Hegazy Y, Salama MMA. Optimal investment
planning for distributed generation in a competitive electricity market. IEEE
Trans Power Syst 2004;19(3):1674–84.
[107] Nara, K, Hayashi, Y, Ikeda, K, and Ashizawa, T. Application of tabu search to
optimal placement of distributed generators. In: Proceedings of IEEE power
engineering society, winter meeting; 2012. p. 918–23.
[108] Golshan MEH, Arefifar SA. Optimal allocation of distributed generation and
reactive sources considering tap positions of voltage regulators as control
variables. Eur Trans Electr Power 2007;17(3):219–39.
[109] Alinejad-Beromi Y, Sedighizadeh M, Sadighi M. A particle swarm optimiza-
tion for sitting and sizing of distributed generation in distribution network to
improve voltage profile and reduce THD and losses. In: Proceedings of the
43rd international UPEC; 2008. p. 1–5.
[110] Falaghi H, Haghifam M. ACO based algorithm for distributed generation
sources allocation and sizing in distribution systems, In: Proceedings of IEEE
PowerTech conference; 2007, 555–560.
[111] Wang L, Singh C. Reliability-constrained optimum placement of reclosers and
distributed generators in distribution networks using an ant colony system
algorithm. IEEE Trans Syst Man Cybern C: Appl Rev 2008;38(6):757–64.
[112] Sutthibun T, and Bhasaputra P. Multi-objective optimal distributed genera-
tion placement using simulated annealing. In: Proceedings of IEEE interna-
tional conference, ECTI-CON; 2010. p. 810–3.
[113] Singh D, Singh D, Verma KS. GA based optimal sizing & placement of
distributed generation for loss minimization. World Academy of Science,
Engineering and Technology, International Science Index 2007;1(11):82–8.
[114] Borges CLT, Falcao DM. Optimal distributed generation allocation for relia-
bility, losses, and voltage improvement. Electr Power Energy Syst 2006;28
(6):413–20.
[115] Paliwal P, Patidar NP and Nema RK. A comprehensive survey of optimization
techniques used for distributed generator siting and sizing. In: Proceedings of
IEEE Southeastcon; 2012. p. 1–7.
[116] Arya LD, Koshti A, Choube SC. Distributed generation planning using
differential evolution accounting voltage stability consideration. Int J Electr
Power Energy Syst 2012;42(1):196–207.
[117] Tan, WS, Hassan, MY, Majid, MS, and Rahman, HA. Allocation and sizing of DG
using cuckoo search algorithm. In: Proceedings of IEEE international con-
ference on power and energy (PECon); 2012. p. 133–8.
[118] Chandrasekhar Y, Maheswarapu S, Matam S. Multiobjective optimization for
optimal placement and size of DG using shuffled frog leaping algorithm.
Energy Proc 2012;14:990–5.
[119] Mohamed Imran A, Kowsalya M. Optimal size and siting of multiple
distributed generators in distribution system using bacterial foraging opti-
mization. Swarm Evol Comput 2014;15:58–65.
[120] Ravi K, and Devabalaji KR. Solving optimum power flow in dynamic
environments to minimize tracking errors using dynamic bacterial foraging
algorithm. In: Proceedings of the international conference on advanced
computing and communication systems (ICACCS); 2013. p. 1–8.
[121] Jain N, Singh SN, Srivastava SC. PSO based placement of multiple wind DGs
and capacitors utilizing probabilistic load flow model. Swarm Evol Comput
2014;19:15–24.
[122] Bashir M, and Sadeh J. Optimal sizing of hybrid wind/photovoltaic/battery
considering the uncertainty of wind and photovoltaic power using Monte
Carlo. In: Proceedings of 11th international conference on environment and
electrical engineering (EEEIC); 2012. p. 1081–6.
[123] Chowdhury BH, Rahman S. A review of recent advances in economic
dispatch. IEEE Trans Power Syst 1990;5(4):1248–59.
[124] Momoh JA, Adapa R, El-hawary ME. A review of selected optimal power flow
literature to 1993. I. Nonlinear and quadratic programming approaches. IEEE
Trans Power Syst 1999;14(1):96–104.
[125] Wood AJ, Wollenberg BF. Power generation operation and control. 2nd ed..
New York, USA: Wiley; 1996.
[126] Gaing ZL. Particle swarm optimization to solving the economic dispatch
considering the generator constraints. IEEE Trans Power Syst 2003;18
(3):1187–95.
[127] Demeo EA, Grant W, Milligan MR, Schuerger MJ. Wind plant integration. IEEE
Power Energy Mag 2005;3(6):38–46.
[128] Ahlstrom M, Jones L, Zavadil R, Grant W. The future of wind forecasting and
utility operations. IEEE Power Energy Mag 2005;3(6):57–64.
[129] Aravindhababu P, Nayar KR. Economic dispatch based on optimal lambda
using radial basis function network. Int J Electr Power Energy Syst 2002;24
(7):551–6.
[130] Yang H, Yang P, Huang C. Evolutionary programming based economic
dispatch for units with non-smooth fuel cost functions. IEEE Trans Power
Syst 1996;11(1):112–8.
[131] Walters DC, Sheble GB. Genetic algorithm solution of economic dispatch with
valve point loading. IEEE Trans Power Syst 1993;8(3):1325–32.
[132] Chiang CL. Genetic-based algorithm for power economic load dispatch. IET
Gener Transm Distrib 2007;1(2):261–9.
[133] Attaviriyanupap P, Kita H, Tanaka E, Hasegawa J. A hybrid EP and SQP for
dynamic economic dispatch with non smooth fuel cost function. IEEE Trans
Power Syst 2002;17(2):411–6.
[134] Victoire TAA, Jeyakumar AE. Hybrid PSO-SQP for economic dispatch with
valve-point effect. Electr Power Syst Res 2004;71(1):51–9.
 S. Behera et al. / Renewable and Sustainable Energy Reviews 48 (2015) 214–227
[135] Panigrahi BK, Pandi VR, Das S. Adaptive particle swarm optimization
approach for static and dynamic economic load dispatch. Energy Convers
Manag 2008;49(6):1407–15.
[136] Wong KP, Wong YW. Thermal generator scheduling using hybrid genetic/
simulated-annealing approach,. IEE Proc Gener Transm Distrib 1995;142
(4):372–80.
[137] He DK, Wang FL, Mao ZZ. Hybrid genetic algorithm for economic dispatch
with valve-point effect. Electr Power Syst Res 2008;78(4):626–33.
[138] Al-sumait JS, Al-othman AK, Sykulski JK. Application of pattern search
method to power system valve-point economic load dispatch. Electr Power
Energy Syst 2007;29(10):720–30.
[139] Sulaiman, MH, Mustafa, MW, Zakaria, ZN, Aliman, O, and Rahim, SRA. Firefly
algorithm technique for solving economic dispatch problem. In: Proceedings
of IEEE international power engineering and optimization conference
(PEDCO). Melaka (Malaysia); 2012. p. 90–5.
[140] Jadhav HT, and Roy R. Economic load dispatch of a power system with wind
energy using craziness based shuffled frog leaping algorithm. In: Proceedings
of IEEE innovative smart grid technologies-India (ISGT India), 2011 IEEE PES;
2011. p. 23–7.
[141] Hamed SH. The intelligent water drops algorithm: a nature-inspired swarm-
based optimization algorithm. Int J Bio-Inspired Comput 2009;1(1/2):71–9.
[142] Rao RS. An intelligent water drop algorithm for solving economic load
dispatch problem. Int J Electr Electron Eng 2011;5(2):43–9.
[143] Swain RK, Sahu NC, Hota PK. Gravitational search algorithm for optimal
economic dispatch. Proc Technol 2012;6:411–9.
[144] Bhattacharya A, Roy PK. Solution of multi-objective optimal power flow using
gravitational search algorithm. IET Gener Transm Distrib 2012;6(8):751–63.
[145] Duman S, Güvenç U, Yörükeren N. Gravitational search algorithm for
economic dispatch with valve-point effects. Int Rev Electr Eng 2010;5
(6):2890–5.
[146] Wang S-K, Chiou J-P, Liu CW. Non-smooth/non-convex economic dispatch by
a novel hybrid differential evolution algorithm. IET Gener Transm Distrib
2007;1(5):793–803.
[147] Wang L, Li L. An effective differential harmony search algorithm for the
solving non-convex economic load dispatch problems. Int J Electr Power
Energy Syst 2013;44(1):832–43.
[148] Vo DN, Schegner P, Ongsakul W. Cuckoo search algorithm for non-convex
economic dispatch. IET Gener Transm Distrib 2013;7(6):645–54.
[149] Bilil H, Aniba G, Maaroufi M. Probabilistic economic/environmental power
dispatch of power system integrating renewable energy sources. Sustain
Energy Technol Assess 2014;8:181–90.
[150] Ogata K. Modern control engineering. 3rd ed. Englewood Cliffs, NJ: Prentice
Hall; 1996.
[151] Astrom KJ, Hagglund T, Zhang CC, Ho WK. Automatic tuning and adaptation
for PID controllers—a survey. IFAC J Control Eng Pract 1993;1(4):699–714.
[152] Zheng F, Wang QG, Lee TH. On the design of multivariable PID controllers via
LMI approach. Automatica 2002;38:517–26.
[153] Bensenouci, A, and Abdelghany, AM. Simulated annealing optimized and
neural networks self-tuned PID voltage regulator for a single-machine power
system. In: Proceedings of 32nd IEEE/IECON 2006. Paris (France); 7–10 Nov
2006. p. 241–6.
[154] Abdelghany, AM, and Bensenouci, A. PID controllers for a one-machine
infinite-bus power system using generalized Ziegler–Nichols method. In:
Proceedings of 4th Saudi technical conference & exhibition, STCEX'2006.
Riyadh (Saudi Arabia); 2–6 Dec 2006.
[155] Na SG, Kim JS, Park SM, Lee YH, Heo H. Auto tuning PID controller based on
improved genetic algorithm for reverse osmosis plant. In: Proceedings of the
world academy of science and engineering; 2008.
[156] Yin F, Wang J, and Guo C. Design of PID controllers using genetic algorithms
approach for low damping slow response plants. In: Advances in neural
networks. L.N.i.C. Science. Heidelberg: Springer-Verlag Berlin; 2004. p. 219–20.
[157] Zain BAM, Tokhi MO, and Toha SF. PID-based control of a single-link flexible
manipulator in vertical motion with genetic optimization. In: Proceedings of
IEEE third Uksim European symposium on computer modeling and simula-
tion, EMS'09; 2009. p. 355–60.
[158] Pereira, DS, and Pinto, JOP. Genetic algorithm based system identification and
PID tuning for optimum adaptive control. In: Proceedings of IEEE/ASME
international conference on advanced intelligent mechatronics; 2005. p.
801–6.
[159] Ruijun D. Differential evolution versus particle swarm optimization for PID
controller design. In: Proceedings of IEEE fifth international conference on
natural computation. ICNC'09, vol. 3; 2009. p. 236–40.
227
[160] Youxin L and Xiaoyi C. Tuning PID control parameters on hydraulic servo
control system based on differential evolution algorithm. In: Proceedings of
IEEE 2nd international conference on advanced computer control (ICACC),
vol. 3; 2010. p. 348–51.
[161] Bingul Z. A new PID tuning technique using differential evolution for
unstable and integrating processes with time delay. In: Pal NR, Kasabov N,
Mudi RK, Pal S, Parui SK, editors. Neural information processing. Berlin
Heidelberg: Springer; 2004. p. 254–60.
[162] dos Santos Coelho L. Tuning of PID controller for an automatic regulator
voltage system using chaotic optimization approach. Chaos Solitons Fractals
2009;39(4):1504–14.
[163] Huang H, Chung CY. Coordinated damping control design for DFIG-based
wind generation considering power output variation. IEEE Trans Power Syst
2012;27(4):1916–25.
[164] Kaushal Jitender. Comparative performance study of ACO & ABC optimization
based PID controller tuning for speed control of DC motor drives. Master
thesis. Patiala: Department of Electrical and Instrumentation Engineering,
Thapar University; 2012.
[165] Mishra Y, Mishra S, Li F. Coordinated tuning of DFIG-based wind turbines and
batteries using bacteria foraging technique for maintaining constant grid
power output. IEEE Syst J 2012;6(1):16–26.
[166] Luo, Y, and Jianying, L. The controlling parameters tuning and its application
of fractional order PID bacterial foraging-based oriented by particle swarm
optimization. In: Proceedings of IEEE international conference on intelligent
computing and intelligent systems, ICIS 2009, vol. 1; 2009. p. 4–7.
[167] Kaldellis JK. The wind potential impact on the maximum wind energy pene-
tration in autonomous electrical grids. Renew Energy 2008;33(7)1665–77.
[168] Chadli, M, El Hajjaji, A. Wind energy conversion systems control using T-S
fuzzy modeling. In: Proceedings of 18th Mediterranean conference on
control and automation, IEEE-MED'10. Marrakech (Morocco); 2010.
p. 1365–70.
[169] Kusiak A, Li W, Song Z. Dynamic control of wind turbines. Renew Energy
2010;35(2):456–63.
[170] Leithead WE, De la Salle S, Reardon D. Role and objectives of control of wind
turbines. IEE Proc C 1991;138:135–48.
[171] Hand MM, Balas MJ. Non-linear and linear model based controller design for
variable speed wind turbines. Report no. CP-500-26244. National Renewable
Energy Laboratory; 1999. p. 1–8.
[172] Knudsen T, Andersen P, Toffner-Clausen S. Comparing PI and robust pitch
controllers on a 400 kW wind turbine by full-scale tests. In: Proceedings of
the European wind energy conference; 1997. p. 546–50.
[173] Ekelund T. Modeling and linear quadratic optimal control of wind turbines.
PhD thesis. Sweden: Chalmers University of Technology; 1997.
[174] Munteanu I, Cultululis NA, Bratcu AI, Ceanga E. Optimization of variable
speed wind power systems based on LQG approach. Control Eng Pract
2005;13:903–12.
[175] Boukhezzar B and Siguerdidjane H. Nonlinear control of variable speed
windturbines for power regulation. In: Proceedings of 2005 IEEE conference
on control applications, CCA; 2005. p. 114–9.
[176] Kongnam C, Nuchprayoon S. A particle swarm optimization for wind energy
control problem. Renew Energy 2010;35(11):2431–8.
[177] Yuhendri M, Mochamad A, Purnomo MH. Maximum output power tracking
of wind turbine using intelligent control. Telkomnika 2011;9(2):217–26.
[178] Karthikeyan Arul M, Muthukumar S, and Krishnan N. Estimator based neuro-
fuzzy control for maximum power extraction from wind electrical power
generation system. In: Proceedings of IEEE international conference on
computational intelligence and computing research (ICCIC); 2010. p. 1–6.
[179] Mansour, M, Mansouri, MN, and Mimouni, MF. Study of performance of a
variable-speed wind turbine with pitch control based on a permanent
magnet synchronous generator. In: Proceedings of IEEE 8th international
multi-conference on systems, signals and devices (SSD); 2011. p. 1–6.
[180] Camblong H. Digital robust control of a variable speed pitch regulated wind
turbine for above rated wind speeds. Control Eng Pract 2008;16:946–58.
[181] Yilmaz AS, Ozer Z. Pitch angle control in wind turbine above the rated wind
speed by multi-layer perceptron and RBFNN. Expert Syst Appl
2009;36:9767–75.
[182] Lin WM, Hong CM. A new Elman neural network-based control algorithm for
adjustable-pitch variable speed wind-energy conversion systems. IEEE Trans
Power Electron 2011;66(2):1847–53.
[183] Belghazi O, Cherkaoui M. Pitch angle control for variable speed wind turbines
using genetic algorithm controller. J Theor Appl Inf Technol 2012;39(1):6–10.
[184] Alba E. Parallel metaheuristics: a new class of algorithms, Vol. 47. Hoboken,
New Jersey: John Wiley & Sons; 2005.