Journal of AI and Data Mining

Vol 4, No 1, 2016, 35-48 10.5829/idosi.JAIDM.2016.04.01.05

Applications of two new algorithms of cuckoo optimization (CO) and

forest optimization (FO) for solving single row facility layout problem
(SRFLP)

M. Maadi1, M. Javidnia2* and M. Ghasemi2

1. Department of Industrial Engineering, Damghan University, Damghan, Iran.

2. Department of Computer Engineering, Damghan University, Damghan, Iran.

Received 30 August 2015; Accepted 10 November 2015

*Corresponding: [email protected] (M. Javidnia).

Abstract
Nowadays, due to the inherent complexity of the real optimization problems, it is a challenging issue to
develop a solution algorithm to these problems. Single row facility layout problem (SRFLP) is an NP-hard
problem of arranging a number of rectangular facilities with varying lengths on one side of a straight line
with the aim of minimizing the weighted sum of the distances between all the facility pairs. In this work, the
two new algorithms cuckoo optimization (CO) and forest optimization (FO) are applied and compared to
solve SRFLP for the first time. The operators of these two algorithms are adapted according to the
characteristics of SRFLP, and the results obtained are compared for two groups of benchmark instances of
the literature. These groups consist of instances with the number of facilities less and more than 30. The
results obtained from the two groups of instances show that the proposed cuckoo optimization algorithm
(COA) has a better performance than the proposed forest optimization algorithm (FOA) in both aspects of
finding the best solution and the computational time.

Keywords: Facility Layout Problem (FLP), Single Row Facility Layout Problem (SRFLP), Cuckoo
Optimization Algorithm (COA), Forest Optimization Algorithm (FOA).

1. Introduction
In a facility layout problem (FLP), we wish to arrange the facilities to minimize the weighted
arrange a number of facilities in a given space to sum of the distances between all the facility pairs.
satisfy an objective function. Single row facility Denoted byΠn , the set of all permutations π of N=
layout problem (SRFLP) is a specific case of FLP, {1, 2 … n}, SRFLP can be formulated as follows
which is the arrangement of n facilities on a line [2]:
so as to minimize the transportation costs among n−1 n
facilities. SRFLP has attracted significant {∑ ∑ cij dπij ⎮π ∈ Πn } (1)
attention in the recent years [1]. Generally, i=1 j=i+1
SRFLP can be described as follows.
where, dπij is the distance between the facilities i
Assume that there are n rectangular facilities.
They should be arranged on one side of a straight and j with respect to the permutationπ. In This
line in a given direction. The parameters involved model, all the permutations of N are searched to
in the problem are the length li (i=1, 2, …, n) of find a permutation with a minimum objective
function value. One should note that due to the
each facility i and ann × n matrix C = [cij ], where
symmetrical property of the cost matrix C, there is
cij is usually the flow between the facilities i and j no difference between cij andcji .
(i, j=1, 2, …, n with i<j). This matrix is a In 1974, Garey et al. showed that the minimum
symmetric matrix. The distance between each pair linear arrangement problem that is a special case
of facilities is calculated as the distance between of SRFLP is NP-hard, so by implication SRFLP is
their centers. The objective of the problem is to NP-hard too[3]. In this paper, with regard to this
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

model, two algorithms are proposed to solve different engineering problems. In the literature,
SRFLP, and they are compared with each other. the heuristic and meta-heuristic algorithms used to
In the literature, because of the theoretical solve SRFLP are divided into 2 groups of
attractiveness and numerous practical situations of construction and improvement methods.
SRFLP, many papers have been presented in this Construction methods construct the sequence of
area. SRFLP has numerous practical applications facilities until a complete permutation is obtained.
in the real world. For example, arrangement of Neghabat in 1974 presented a constructive
rooms in hospitals, departments in office algorithm to obtain a complete solution by adding
buildings or supermarkets [4], arrangement of one machine at a time to the end of the current
machines in flexible manufacturing systems [5], solution [26].A heuristic method based on the
design of warehouse layouts, and assignment of eigenvectors of a transformed flow matrix was
files to disk cylinders in computers storage [6] are introduced by Drezner in 1987 [27]. A linear
some of these applications mentioned in the mixed-integer formulation of the SRFLP and a
literature. penalty technique to solve it was presented by
In addition to these applications, there are some Heragu and Kusiak in 1991 [28]. Kumar et al. in
practical applications of a special case of SRFLP, 1995and Djellab and Gourgand in 2001
in which the facilities have unit lengths. These introduced a constructive greedy heuristic method
applications include the design of error correcting to solve SRFLP [29, 30]. Also some heuristic
codes [7], wire length minimization in VSLI procedures, used to extract a feasible solution to
design, graph drawing, and reordering of large SRFLP from an optimal solution to the semi-
sparse matrices [8, 9]. Regarding the numerous definite programming (SDP) relaxation, were
applications of SRFLP, several methods have presented by Anjos et al. in 2005, Anjos et al. in
been proposed in the literature to solve it. These 2008,Anjos and Yen in 2009, and Hungerlander
methods can be categorized into exact, heuristic, and Rendle in 2011[7,11,13,31].
and meta-heuristic algorithms. Exact methods The improvement methods start with one or more
have been applied to solve small instances of permutations of facilities as the initial solutions,
SRFLP to optimality, and their relaxations have and improve them until the stopping criteria is
been used to obtain good bounds for larger reached or the solution cannot be improved. The
SRFLP instances. The first exact method used to seven meta-heuristic algorithms tabu search (TS),
solve SRFLP was a branch and bound algorithm genetic algorithm (GA), particle swarm
presented by Simmons in 1969 [4].After that, optimization (PSO), simulated annealing (SA), ant
different exact methods have been reported in the colony optimization (ACO), scatter search (SS),
literature including combinatorial branch and and imperialist competitive algorithm (ICA) have
bound, mathematical programming [10,4], been reported in the literature to solve SRFLP.
dynamic programming [6], branch and cut Also some hybrid algorithms based on SA, GA,
[5,6,11], cutting planes approach [11,12], and ACO, PSO, etc. have been studied in different
semi-definite programming [7,13,14]. papers to solve SRFLP. Romero and Sanchez-
Since in this paper we study two meta-heuristic Flores in 1990, Heragu and Alfa in 1992, and
algorithms to solve SRFLP, we tend to review the Gomes de Alvarenga et al. in 2000 applied SA to
heuristic and meta-heuristic algorithms suggested solve SRFLP [32,33,34]. Solimanpur et al. in
in the literature for SRFLP. For more information 2005 presented an ant colony algorithm to solve
on the exact methods that have been presented in this problem [35].Different papers including Datta
the literature to solve SRFLP, the researchers are et al. in 2011 and Ficko et al. in 2004usedGA to
referred to review the paper presented by Kothari solve SRFLP [36,37]. Kumar et al. in 2008 and
and Ghosh [1]. Kothari and Ghosh in 2014 presented a scatter
Since the exact algorithms for SRFLP are search algorithm to solve SRFLP [38,39]. Gomes
computationally expensive, they have been de Alvarenga et al. in 2000 and Samarghandi and
applied to relatively small instances, with up to 42 Eshghi in 2010 applied TS to solve SRFLP
facilities. The heuristic and meta-heuristic [34,40]. In 2010, Samarghandi et al. presented
algorithms are faster than the exact methods but PSO to solve this problem [41]. Akbari and Maadi
they do not guarantee an optimal solution [1]. In and Lian et al. in 20ll proposed an ICA to solve
the recent years, different meta-heuristic SRFLP [42, 43]. Also there are some hybrid
algorithms have been applied to solve various algorithms to solve SRFLP such as a hybrid
engineering problems because of their algorithm based on SA and GA by Braglia in
complexity. For example, references [15-25] 1996 and a hybrid algorithm based on ACO and
present different meta- heuristic algorithms for PSO by Teo and Ponnambalam in 2008 [44,45].

36
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

Cuckoo optimization algorithm (COA) is a new cuckoos migrate to places more suitable for
meta-heuristic algorithm introduced by Rajabioun generation survival and egg-laying. The number
in 2011 [46]. COA has proven its excellent of grown eggs shows the nest suitability of the
capabilities such as faster convergence and better area. The goal of a cuckoo optimization problem
global minimum achievement rather than other is to find a situation in which a maximum number
meta-heuristic algorithms [47]. COA is applied to of eggs are saved. The environmental features and
solve different non-linear problems, and has migration of groups of cuckoos hopefully lead
shown good capability in diverse optimization them to converge and find the best environment
tasks. This algorithm has been tested so far on for breeding and reproduction. This best
different types of practical instances in some environment is global or the best solution to the
scopes such as teleportation systems, machine problem. In this process, after chicks become
error compensation, noise canceller design, mature, they make some groups. Each group has
chemical machine process, and machine process its own typical habitat. The habitat with the best
[48-55].This algorithm has been much better than situation is the destination of the cuckoos of other
the rest of the meta-heuristic algorithms, and has groups. After moving, each group resides in the
shown its efficiency [47]. This subject can be a area near the current location. Regarding the
motivation to apply COA in other scopes of number of each cuckoo eggs and the distance of
optimization such as SRFLP. cuckoo to the best residence, some egg laying
Forest optimization algorithm (FOA) is a new radii are assigned to it. After that, cuckoos start to
evolutionary algorithm, which is inspired by few lay eggs in some random nests inside this radius.
trees in the forest. This algorithm was introduced This process continues until the best position is
by Ghaemi and Feizi-Derakhshani in 2014 [56]. obtained. In the best position, most of the cuckoo
Since FOA is a newly introduced algorithm, there habitats are gathered around the same global
is no paper in the literature using it, and the solution [46]. Figure 1 shows a COA flowchart.
present paper can be a start to apply FOA to solve COA was introduced basically to solve
different optimization problems. Also FOA can continuous optimization problems. Until now, this
solve continuous problems, and this paper algorithm has been modified to solve discrete
presents a forest-based algorithm that solves optimization problems, and has had impressive
discrete SRFLP with changing FOA operators results. In the next sections, the stages of the
regarding the characteristics of SRFLP. proposed COA to solve SRFLP are described. The
As only seven meta-heuristic algorithms have implementation of COA in SRFLP is as follows.
been reported in the literature to solve SRFLP,
this paper introduces the two new COA and FOA 2.1. Generating initial cuckoo habitat
to solve SRFLP for the first time. The As the goal of an optimization problem is to find
performances of these two algorithms are also an optimal solution in terms of the variables, a
compared to each other using the SRFLP representation pattern that is usually a vector of
instances of the literature with different sizes. the decision variables should be defined. For
Our paper is organized as follows. In section 2, example, in genetic algorithm (GA), this vector is
the proposed COA to solve SRFLP is presented. defined as chromosome and in particle swarm
Section 3 describes the proposed FOA for solving optimization (PSO), this array is named particle
this problem. In section 4, computational results position. In COA, it is called habitat.
of the two proposed algorithms are compared, and In an Nvar dimensional optimization problem, a
it is followed by the conclusions in section 5. habitat is an array of 1 × Nvarrepresenting the
current living position of cuckoos. This array is
2. Proposed cuckoo optimization-based defined as follows:
algorithm
As stated earlier, Cuckoo optimization algorithm Habitat = [x1 , x2 , … , xNvar ] (2)
(COA) is an evolutionary algorithm inspired by Like the other meta-heuristic algorithms, the
the life of a bird named Cuckoo. This algorithm is profit of a habitat is obtained by evaluation of the
based on the specific egg-laying and breeding of profit function fp at a habitat, as follows:
cuckoos. COA starts with an initial population of
cuckoos that have some eggs to lay in some host Profit = fp (habitat) = fp (x1 , x2 , … , xNvar ) (3)
birds' nest. Those eggs that are more similar to the As it can be seen, COA is an algorithm that
host birds' eggs survive and can grow to become a maximizes the benefit function.
mature cuckoo. Other eggs are detected by the
host birds and are expelled out. In such cases,

37
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

where, α is an integer, supposed to handle the

Start
maximum value for ELR. To solve SRFLP with n
facilities using COA, according to the problem
Initialize cuckoos
definition, a habitat is represented as an n
with eggs dimensional vector of integer numbers between 1
and n that shows a permutation allocating
Determine egg laying rectangular facilities to a straight line. For
Lay eggs in different radius for each
nests cuckoo example, figure 2 shows a habitat of COA in
SRFLP with 10 facilities.
2 3 6 8 9 10 1 4 7 5
Some of eggs are Move all cuckoos Figure 2. Habitat representation.
detected and killed toward best
environment Using (1) as the cost function that should be
minimized for COA, the profit of each habitat can
No be calculated using (4). To generate the initial
Population is
less than max Determine cuckoo population, two procedures are used in this paper.
value? societies The first one is based on theorem 1 in the paper
Kill cuckoos in presented by Samarghandi and Eshghi [40]. In
worst area
Yes this procedure, it is assumed that in the cost
Find nests with best function coefficients, cij = c. Now, if we sort
survival rate
Check Survival of eggs facilities in a non-descending order such that the
in nests
(Get profit values)
shortest facility is denoted by 1 and the longest
one by n, then figure 3 shows the optimum
Let eggs grow solution when n is an odd number, and figure 4
shows the optimum solution when n is an even
Stop
condition?
number. With this procedure and another
No procedure that randomly generate the permutation
Yes allocating the rectangular facilities to a straight
End line, the initial population is created. After
producing the initial population, the process of
Figure 1. Flowchart of COA. laying eggs will be started.
To use COA for solving the minimization … i … 3 1 2 … i-1 …
problems, it is sufficient to multiply a negative
mark in a cost function as follows: Figure 3. Optimal layout when n is odd.

Profit = fp (habitat) = −fp (x1 , x2 , … , xNvar ) (4) … i … 2 1 3 … i-1 …

To start COA, a matrix of sizeNpop × Nvar should Figure 4. Optimal layout when n is even.
be created, and for each habitat of initial
population, some randomly-produced eggs are 2.2. Cuckoos' style for egg laying
supposed to be assigned. The number of eggs that In this stage, each cuckoo starts laying eggs
each cuckoo lays is between 5 and 20.These randomly in some other host birds' nests within
values are the upper and lower limits for egg- her ELR. A clear view of this concept is shown in
laying at different iterations. Another habit of real figure 5.
cuckoos is that they lay eggs within a maximum
distance from their habitat. This maximum range
is called the egg laying radius (ELR). In an
optimization problem with an upper limit of varhi
and a lower limit of varlow for the variables, for
each cuckoo, an ELR, which is proportional to the
total number of eggs, number of current cuckoo's
eggs, and also variable limits of varhi and varlow
is determined. The ELR is defined as follows:
Number of current cuckoo′ seggs
ELR = α ×
total number of eggs (5) Figure 5. Random egg-laying in ELR. Central red star is
× (varhi − varlow ) the initial habitat of cuckoo with 5 eggs; pink stars are
eggs’ new nests.

38
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

After all cuckoos' eggs are laid in the host birds' way and also have deviation. This movement is
nests, some of them that are less similar to the shown in figure 6. As it is clear in this figure, each
host birds' own eggs are detected by the host birds cuckoo flies only λ% of the total direction toward
and thrown out of the nest. Thus after the laying the destination habitat, and also has a deviation of
process, p% of the eggs with less profit values φ radians. The two parameters λ and φ help
will be destroyed. P is usually 10 in this step. The cuckoos to search the environment more. λ is a
rest of the eggs grow in the host nests and are fed random number between 0 and 1, and φ is a
by the host birds. The important point about the number between ‒π /6 and π /6. When all cuckoos
cuckoo chicks is that only one egg per nest is migrate to the target habitat, and also the new
allowed to grow because when a cuckoo egg habitats are specified, some eggs are given to the
hatches and the chick comes out, she throws out mature cuckoos. After that, considering the
the host bird eggs, and if the host bird chicks are number of eggs for each cuckoo, an ELR is
hatched sooner, the cuckoo chicks eat the most specified for it, and a new egg-laying process
food that the host birds bring (because of their3 restarts.
times bigger bodies, they push other chicks and
eat more), and after some days, the host bird
chicks starve to death and just the cuckoo chicks
survive.
For implementation of this step of COA in
SRFLP, in the proposed algorithm, because of the
definition of habitat and structure of the problem,
there is no ELR, and some changes in the
structure of habitats are used for the laying eggs
operator. For this aim, the swap operator is used.
The steps of the proposed swap operator are as
follow:
1- Consider a mature cuckoo habitat for
laying eggs.
2- To lay eggs, exchange the position of two
facilities of a habitat and make an egg.
Using this operator, the eggs are laid and the
algorithm continues to the next section. Figure 6. Immigration of a sample cuckoo toward a goal
habitat.

2.3. Immigration of cuckoos To simulate this operator of COA for SRFLP,

After growing young cuckoos, they live in their considering the characteristics of SRFLP, the
own area and society for some time. At the time steps proposed for the immigration operator are
of spawning, migration to new and better habitats, applied in the following procedure.
where there is more chance for eggs to survive At first, the target habitat toward which the other
and more food available for young cuckoos, starts. cuckoos should migrate is considered (a habitat
After forming cuckoo groups in different areas with the highest profit).Then a group of facilities
(the search space of the problem), the society with in the target habitat is selected. Figure 7 shows an
the best profit is selected as the target point of a example of a target habitat with 10 facilities and
group for other cuckoos to migrate. When mature selected group of facilities of the habitat. An array
cuckoos live all over the environment, it is tough of a migrant cuckoo who wants to immigrate
to recognize which cuckoo belongs to which toward the target habitat is considered (Figure 8).
group. To solve this problem, the grouping of The selected group of facilities that is in target
cuckoos is done by the K-means clustering habitat is copied exactly to the array of new
method (a k value between3 and 5 seems to be position of the migrant cuckoo. For the other
sufficient). After formation of groups, the average facilities in the new position of the migrant
profit value is calculated to achieve the relative cuckoo that are not in the selected group, from the
optimality of that group habitat. After that, the first cell of the migrant cuckoo habitat array, each
group with the highest amount of profit is selected facility that is not in the facilities of the selected
as the target group, toward which other groups group is placed in the cells of the array of the new
will migrate. During movement toward the target position of the migrant cuckoos sequentially.
point, the cuckoos do not fly all the way to the Figure 9 shows the array of the new position of
destination habitat. They only fly a part of the the migrant cuckoo.

39
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

8 5 2 1 6 4 3 9 10 7 3. Proposed forest optimization-based

Figure7. Target habitat and selected group of facilities. algorithm
Forest optimization algorithm (FOA) is a new
4 7 6 5 10 9 3 8 2 1
evolutionary algorithm that is inspired by few
Figure 8. A migrant cuckoo habitat. trees in the forest which can survive for several
7 5 10 1 6 4 3 9 8 2 decades, while other trees could live for a limited
period. As mentioned earlier, this algorithm was
Figure 9. New position of migrant cuckoo.
introduced by Ghaemi and Feizi-Derakhshani in
2.4. Eliminating cuckoos in worst habitats 2014 [56].
Since there is always equilibrium in the birds' FOA involves three main stages: local seeding of
population in nature (because of food limitation, the trees, population limiting, and global seeding
being killed by predators, and inability to find of the trees. Like other meta-heuristic algorithms,
proper nest for eggs), a numberNmax is determined FOA starts with the initial population named
as the maximum number of cuckoos that can live trees. Each tree represents a solution of the
in the environment. In this step, only those Nmax optimization problem. A tree besides the cells of
number of cuckoos that have better profit values variables of the problem has a part that represents
survive, and the others demise. After some the age of the tree.
iteration, the algorithm moves to one best habitat At first, the age of a tree is set '0'. After
with the characteristic of maximum similarity of initialization of the trees, the operator of local
eggs to the host birds and also with the maximum seeding will generate new young trees (or seeds in
food resources. This habitat has the best profit fact) from the trees with age 0 and add new trees
value, and is the best solution to the problem. In to the forest. Then all trees, except the new
COA, convergence of 95% of cuckoos to the same generated ones, get old, and their age increases by
habitat puts an end to the algorithm. The stopping '1'. The next step is the population limiting. At
condition for this algorithm, like other this step, there is a control on the population of
evolutionary algorithms, can be the number of trees in the forest, and some trees will be omitted
iterations. In the proposed algorithm, the number from the forest. The omitted trees form the
of iterations is considered as the stopping candidate population for the global seeding stage.
condition. The pseudo-code of the proposed COA In the third stage, which is the global seeding of
used to solve SRFLP is described in algorithm 1. the trees, a percentage of the candidate population
Algorithm 1: Proposed COA
is chosen to move far in the forest. The aim of
Input: number of facilities (n), flow matrix (F), facilities length global seeding is adding some new potential
matrix (L), maximum number of iterations (maxiter), minimum solutions to the forest in order to get rid of the
number of eggs (minegg), maximum number of eggs (maxegg).
local optimums. After that, the trees are ranked
Output: An approximation of an optimal solution to the PCSP with regard to their fitness values. The tree with
instance. the best value of fitness function is chosen as the
Initialize cuckoo habitats with N individuals using section 2.1. best tree, and its age is set '0' in order to avoid the
Compute fitness for each cuckoo aging of the best tree as the result of local seeding
Find best cuckoo as best cuckoo
for i=0 to maxiter stage. In this way, it will be possible for the best
for each cuckoo tree to locally optimize its location by the local
dedicate some eggs with random number between {minegg ,
maxegg}
seeding operator.
end These 3 stages will be continued until the
stopping conditions are met. Figure 10 shows the
for each cuckoo
for each egg of selected cuckoo flowchart of the forest optimization algorithm. As
Perform laying operator using section 2.2. mentioned earlier, since FOA is a new introduced
end algorithm, there is no paper in the literature that
Sort cuckoo habitats according to their fitness value and update uses this algorithm, and this paper can be regarded
best cuckoo as a start to apply FOA to solve different
Select the best N cuckoos and reduce the extended cuckoos.
optimization problems. Also as FOA is introduced
for each cuckoo to solve continuous non-linear programming, this
Let new cuckoo populations immigrate toward best cuckoo paper introduces a discrete version of FOA to
using section 2.3.
end solve SRFLP. In the next sections, using operators
Compute fitness for each cuckoo and update best cuckoo of FOA, a forest optimization-based algorithm is
end introduced to solve SRFLP.
Return the best cuckoo as the result

40
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

3.1. Initialize trees Tree = [x1 , x2 , … , xNvar , Age] (6)

As mentioned earlier, in FOA, the potential
solution to each problem is considered as a tree. A In the definition of a tree, the age of a tree has a
tree is usually a vector of variables, and the maximum range that is a predefined parameter
variables have lower and upper limits. In addition named lifetime. This parameter should be
to the variables, each tree has a part related to the determined at the beginning of the algorithm.
age of the tree. The age of a tree for each newly When the age of a tree reaches the lifetime
generated tree is set '0' as a result of local seeding parameter, that tree is omitted from the population
or global seeding. After the local seeding stage, of the trees, and is added to the candidate
the age of the trees, except for the new generated population. The big number of lifetime parameter
ones in the local seeding stage, increases by '1'. increases the age of the trees in each iteration, and
This increase in age is used later as a controlling the forest will be full of the old trees that do not
population in the limiting stage. Thus in FOA, a take part in the local seeding stage. Otherwise, a
tree is represented as an array of 1 × (Nvar + 1), very small value for this parameter causes the
where Nvar is the diminution of the optimization trees to get old very soon, and they will be
problem, and one cell is considered as the age of a omitted at the beginning of the competition. Thus
tree. Equation 6 represents a tree. this parameter should provide a good chance for a
local search. The best lifetime value was
determined to be 6 in FOA.
Start In SRFLP, a tree is represented as an n+1
dimensional vector, whose first n cells are integer
numbers of 1 ton that represent a permutation,
allocating rectangular facilities to a straight line,
Initialize forest with 0-aged trees and the last cell is the age. Figure 11 shows a tree
with 10 facilities.
4 7 6 5 10 9 3 8 2 1 0
Figure 11. A tree representation in SRFLP with 10
Local seeding on 0-aged trees facilities and age of zero.

The fitness function for FOA is the same as that

for COA, which is (1) and should be minimized.
For generating the initial population, the same
Population limiting and form procedures used in COA are applied for FOA. All
thecandidate population
the initial trees in this step have the age '0'. With
this population, the algorithm starts the next stage,
which is the local seeding of the trees.
Global seeding on the selected trees
from candidate population
3.2. Local seeding of the trees
In the nature, during the seeding procedure of the
trees, some seeds fall just near the trees, and after
some time, they will grow. Now, the competition
between the near-young trees starts, and those
Update the best tree trees with better growing conditions such as
enough sunlight and better location survive. The
local seeding of the trees is simulated regarding
this natural event. This operator is performed on
the trees with the age '0', and adds some neighbors
Stop condition? of each tree to the forest. The number of seeds
No
that fall on the land near the trees and then turn
into the trees as neighbors is considered as an
Yes FOA parameter named the local seeding changes
or LSC. The value of this parameter used to solve
End different optimization problems depends on the
dimension of the problem domain. In FOA, for
Figure 10. Flowchart of FOA. problems with a dimension bigger than 5, 2/10 is
recommended for LSC, and for dimensions less
than 5, the LSC value will be 1. In local seeding

41
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

operator, at first, a variable of a tree with the age rate, and should be defined at the beginning of the
'0' is selected. After that, a random number named algorithm.
r is generated in the range of(−∆x, ∆x). Then the The steps of global seeding of the trees are
value for the selected variable is added to r. ∆x is defined as follows. Regarding the transfer rate
a small value, and it is smaller than the related parameter, the trees from the candidate population
variable's upper limit. Now, a new tree with age are selected. Then some variables of each tree are
'0' is added to the forest. selected randomly. After that, the value for each
For simulating the local seeding operator to solve selected variable is changed with another
SRFLP, the swap operator is used to generate new randomly-generated value in the related variable
trees. In this operator, for each tree with age '0', range. As a result, a tree with age '0' is added to
two facilities are selected randomly, and their the forest. This operator performs a global search
locations are exchanged. With this operator and in the search space. During this stage, another
using the LSC parameter, many new trees are parameter that is the number of variables, whose
added to the forest. In the next stage of the values will be changed, is defined as the global
algorithm, a limitation on the number of trees seeding changes or GSCs.
should be considered as the operator of population During the global seeding operator in the
limiting. proposed algorithm to solve SRFLP, after
determination of GSCs, the swap operator is
3.3. Population limiting carried out. The number of repetition of the swap
To avoid infinite expansion of the forest, after operator is the number of GSCs. After producing
generating numerous trees in the stage of local new trees, these trees with age '0' are added to the
seeding of the trees, the number of trees should be forest.
limited. In FOA, the two parameters lifetime and
area limit restrict the population of the trees. At 3.5. Updating the best so far tree
first, the trees with an age more than the lifetime In this stage, after sorting trees according to the
parameter are removed, and will form the value of their fitness function, the tree with the
candidate population. best fitness function value is selected as the best
The second limitation is the area limit, in which one. After that, the age of this tree turns to '0'. As
after ranking the trees according to the fitness mentioned earlier, in this way, it will be possible
value, if the number of trees exceeds the for the best tree to locally optimize its location by
limitation of the forest, extra trees are omitted the local seeding operator.
from the forest and added to the candidate
population. The value for the area limit parameter 3.6. Stop conditions
is considered to be the same as the number of the In FOA, the three stopping conditions predefined
initial trees. After population limiting, the global number of iterations, reaching the specified level
seeding stage is performed on some percentage of of accuracy, and observance of no change in the
the candidate population (described in the next fitness value of the best tree are defined. In the
section). proposed forest optimization-based algorithm, the
number of iterations is considered as the stopping
3.4. Global seeding of trees condition. The pseudo-code of the proposed FOA
There are different kinds of trees, and numerous to solve SRFLP is described in algorithm 2.
animals and birds feed on the seeds and fruits of
these trees in the nature. Thus in spite of local 4. Comparison of cuckoo- and forest-
seeding of the trees, seeds are distributed in the optimization algorithms
entire forest, and consequently, the habitat of the Cuckoo optimization (CO) and forest optimization
trees expands. Also different natural processes (FO) algorithms are coded in C#. The algorithms
such as wind and water flow help distributing the are run on an Intel (R) core(TM) i5-3210 CPU
seeds in the entire forest widely. Using these @2.05 Gigahertz and 4.00Gigabytes ram under
natural processes, the global seeding stage is the Windows 8.1 operating system.
defined to simulate the distribution of the seeds of The performance of the two algorithms is
the trees in the forest. evaluated on several benchmark instances. These
The global seeding operator is performed on a instances are divided into two groups of instances
determined percentage of the candidate population with the number of facilities less than 30, and
from the previous stage. This percentage is instances with the number of facilities more than
another parameter of the algorithm named transfer 30. All the instances are available at
http://www.gerad.ca/files/sites/Anjos/flplib.html.

42
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

For instances with the number of facilities less are related to COA. Column 4 is the objective
than 30, the maximum number of iterations is function value of the achieved COA solution.
set150, and for the other group of instances, it is Column 5 shows the computational time of COA,
considered as 300. For each instance, each and column 6 calculates the gap between
algorithm is run 20 times. In the following objective function values of optimum solution and
section, the computational experience of the two achieved COA solution of the instances.
algorithms for the two groups of instances is As it can be seen in table 1, the proposed COA is
reported. able to achieve the best solutions reported in the
Algorithm2: Proposed FOA literature. The next three columns 7, 8, and 9 are
the results of applying FOA. In column 7, the
Input: number of facilities (n), flow matrix (F), facilities length
matrix (L), number of initial trees (N), maximum number of
value of objective function of the achieved FOA
iterations (maxiter), local seeding change (LSC), global seeding solution is shown. Column 8 shows the
change(GSC), transfer rate, life time, area limit. computational time of FOA, and column 9
Output: An approximation of an optimal solution to the SRFLP calculates the gap between objective function
instance. values of optimum solution and achieved FOA
Initialize forest with N individuals using section 2.1.
solution of the instances. This gap is calculated as
The age of each tree is initially set zero. follows:
achievedFOAsolution − optimumsolution
For i=0 to maxiter Gap =
optimumsolution
(7)
for each trees with age 0
for j=0 to LSC Regarding table 1, the average gap between
Perform local seeding on selected trees using section
3.2. objective function values of optimum solutions
end and achieved FOA solutions of all instances is
end 0.0017.
Increase the age of all trees by 1, except for the newly
generated trees in this stage. In the comparison of the two proposed algorithms,
Remove the trees with age bigger than lifetime parameter and it is notable that the proposed COA has a better
add them to the candidate population.
Sort trees according to their fitness value.
performance than FOA in both aspects of
Remove the extra trees that exceed the area limit parameter from achieving the best solution and computational
the end of forest and add them to the candidate population. time. Table 2 shows the mean and standard
Choose transfer rate percent of the candidate population.
For each selected tree deviation (SD) of objective function values of the
Perform global seeding using section 3.4. instances. As mentioned earlier, each instance is
Set the age 0 run 20 times. Figure 12 depicts and compares the
end
end computational times of the two algorithms.
Return the best tree as the result. In figure 12, the horizontal axis shows the
computational time, and the vertical axis depicts
4.1. Instances with number of facilities less the problem number.
than 30
Initially, the algorithms are tested on some 4.2. Instances with number of facilities more
instances with n ≤ 30 in the literature. The than 30
optimum solutions to this set of instances are The same computational results for the second set
known. Throughout the experiments, the of problems with the number of facilities more
following parameter values are used for FOA and than 30 are shown in table 3.
COA, respectively. According to this table, COA with the average
For COA, the initial population of cuckoos is set gap of 0.0013 has a better performance than FOA
30. The lower limit for egg-laying is considered 3, with the average gap of 0.0017. Also the
and the upper limit for egg-laying is set 5. In computational time of COA is less than FOA for
FOA, the number of initial trees is considered to all instances, and these results demonstrate a
n better performance of COA to solve SRFLP.
be 30, GSC= 1+⌊10⌋, LSC=5, and the transfer rate
Table 4 shows the mean and standard deviations
is set (0.9*n).
of the objective function values of instances of set
The results obtained for the application of COA
2. As mentioned earlier, each instance is run 20
and FOA are shown in table 1. In this table, the
times.
first column shows the problem number. The
Figure 13 shows and compares the difference
number of facilities of instances is presented in
between the computational times of the two
the second column. The next column is the
algorithms. In figure 13, the horizontal axis shows
objective function value of optimum solution of
the computational time, and the vertical axis
instances from the literature. Columns 4, 5, and 6
depicts two algorithms.

43
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

Table 1. Results for problem set1.

Problem Number of Optimal FOA FOA FOA COA COA COA
Number Facilities Value Result Time (Sec) Gap Result Time(Sec) Gap
1 4 78 78 0 0 78 0 0
2 4 638 638 0 0 638 0 0
3 8 801 801 0 0 801 0 0
4 8 2324.5 2324.5 0 0 2324.5 0 0
5 9 2469.5 2469.5 0 0 2469.5 0 0
6 9 4695.5 4695.5 0 0 4695.5 0 0
7 10 2781.5 2781.5 0 0 2781.5 0 0
8 11 6933.5 6933.5 0 0 6933.5 0 0
9 15 63.05 63.05 0 0 63.05 0 0
10 20 15549 15592 1 0.002765451 15549 1 0
11 25 4618 4625 4 0.001515808 4618 3 0
12 25 37116.5 37129.5 12 0.000350249 37116.5 9 0
13 25 24301 24325 12 0.000987614 24301 8 0
14 25 48291.5 48344 14 0.001087148 48291.5 9 0
15 25 15623 15633 14 0.000640082 15623 10 0
16 30 8247 8313 25 0.00800291 8247 18 0
17 30 21582.5 21725.5 25 0.006625738 21582.5 17 0
18 30 45449 45627 24 0.003916478 45449 17 0
19 30 56837.5 57013.5 25 0.003096547 56837.5 18 0
20 30 115268 116067 26 0.006931672 115268 17 0
Table 2. Mean and SD results for instances of set one in 20 times of run.
Problem number COA-Mean COA-SD FOA-Mean FOA-SD
1 78 0 78 0
2 638 0 638 0
3 801 0 801 0
4 2327.8 3.012474066 2331.7 5.380520421
5 2474.1 5.412947441 2475.8 6.66708332
6 4702.1 5.319774431 4704.7 5.826662853
7 2786.7 4.236744033 2789.7 6.300793601
8 6941.5 4.716990566 6943.5 5.979130372
9 63.05 0 63.05 0
10 15588 22.25982929 15607.4 14.27585374
11 4622.8 6.57267069 4633 9.082951062
12 37154.6 39.49272085 37171.8 38.6144403
13 24330.4 26.83840532 24358 35.46124645
14 48336.6 57.97995343 48388.4 45.04220243
15 15661.2 35.10982768 15684.6 29.16847613
16 8268.8 20.31501907 8328.6 13.50185172
17 21599.6 18.84608713 21764.8 32.21335127
18 45513 39.91240409 45713.4 79.10625766
19 56928.5 70.48049376 57128.3 78.31714372
20 115467.2 177.3899659 116325.4 264.8882783
Table 3. Results for problem set2.
Best
Problem Number of known FOA FOA FOA COA COA COA
Number Facilities value from Result Time (Sec) Gap Result Time(Sec) Gap
literature
1 40 107348.5 107886 39 0.005007056 107536.5 25 0.001751305
2 40 97693 97693 36 0 97693 24 0
3 40 78589.5 78689.5 37 0.001272435 78686.5 25 0.001234262
4 40 76669 77295 35 0.008164969 77191 23 0.006808488
5 40 103009 103127 35 0.001145531 103018 22 8.7371E-05
6 60 1477834 1500899 110 0.015607301 1477834 69 0
7 60 648337.5 649097.5 114 0.001172229 648792 69 0.000701024
8 60 841792 851120 109 0.011081122 841792 68 0
9 60 398468 398710 120 0.000607326 398480 69 3.01153E-05
10 60 318805 320678 114 0.005875065 318805 68 0
11 70 1518993.5 1530590 187 0.007634332 1530191 97 0.007371658
12 70 1441028 1443709 189 0.001860477 1441467 100 0.000304644
13 70 1518993.5 1527075.5 188 0.005320628 1520453.5 101 0.000961163
14 70 968796 969880 187 0.001118915 969464 98 0.000689516

44
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

15 70 4218017.5 4227125.5 190 0.002159308 4218797.5 101 0.000184921

Table 4. Mean and SD results for instances of set two in 20 times of run.
Problem number COA-Mean COA-SD FOA-Mean FOA-SD
N40_1 107791.8 237.6574741 107994.4 226.8453217
N40_2 97957.2 331.4131862 97997.2 414.8670872
N40_3 78997.1 414.9904216 79077.9 289.8569475
N40_4 77356.4 156.4106774 77598 310.6565628
N40_5 103483.6 427.486608 103765.4 586.7161153
N60_1 1478014 402.4922359 1501299 547.7225575
N60_2 648976.1 694.2137999 649159.4 652.3557695
N60_3 842162.4 828.2395789 851520 894.427191
N60_4 398775.8 264.9607518 399198 681.5937206
N60_5 318944.2 311.2606625 320707 884.421845
N70_1 1530418.2 260.7799843 1531022.8 276.3099712
N70_2 1441707 328.6335345 1444283.6 413.8892362
N70_3 1521250.7 711.4087785 1528030.2 962.3980856
N70_4 969872.4 704.0914714 970038.8 614.6276921
N70_5 4218891.9 544.4941926 4227575.3 640.7808317

Figure 12. Comparison of computational times of two CO- Figure 13. Comparison of computational times of two CO-
and FO-based algorithms for instances with number of and FO-based algorithms for instances with number of
facilities less than 30. facilities more than 30.

5. Conclusion the performance of these two algorithms, two

This paper considers the single row facility layout groups of instances with the number of facilities
problem (SRFLP). Some exact and heuristic less and more than 30 are used. The
methods have been developed in the recent years computational results obtained for the two
to solve this problem but, as this problem is NP- algorithms show that COA has a better
hard, finding optimum solutions for the large performance than FOA in both aspects of
instances of this problem is not possible within a achieving the best solution and computational
reasonable time. Nowadays different meta- time.
heuristic algorithms are introduced in the
literature to solve SRFLP but we still need to have Reference
a fast algorithm that can obtain a near-optimum [1] Kothari, R. & Ghosh, D. (2012). The single row
solution for large instances of the problem. facility layout problem: state of the art, OPSEARCH,
Cuckoo optimization (CO) and forest optimization vol. 49, no. 4, pp. 442-462.
(FO) algorithms are two meta-heuristic algorithms [2] Amaral, A. R. S. (2006). On the exact solution of a
that have been introduced recently. facility layout problem, European Journal of
In the literature, these algorithms have not been Operational Research, vol. 173, no. 2, pp.508–518.
applied to solve SRFLP yet. In this paper, the [3] Garey, M. R., Johnson, D. S. &Stockmeyer, L.
performance of COA and FOA to solve SRFLP is (1974). Some simplified NP-complete problems,
compared. As COA and FOA are defined to solve Proceedings of the 6th Annual ACM Symposium on
continuous optimization problems in the Theory of Computing, New York, United States, pp.
literature, to solve SRFLP that is a discrete 47–63.
problem, at first, a cuckoo-based algorithm and a
forest-based algorithm are proposed. Then to test

45
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

[4] Simmons, D. M. (1969). One-dimensional space [19] Alfi, A. (2014). Swarm-based structure-specified
allocation: an ordering algorithm, Operations Research, controller design for bilateral transparent teleoperation
vol. 17, no. 5, pp. 812–826. systems via µ synthesis, IMA Journal of Mathematical
[5] Heragu, S. S. & Kusiak, A. (1988). Machine layout control and Information. vol. 31, pp. 111–136.
problem in flexible manufacturing systems, Operations
[20] Alfi, A. & Fateh, M. M. (2011). Identification of
Research, vol. 36, no. 2, pp. 258–268.
nonlinear systems using modified particle swarm
[6] FPicard, J. C. & Queyranne, M. (1981). On the one- optimization: A hydraulic suspension system, Journal
dimensional space allocation problem, Operations of Vehicle System Dynamics, vol. 46, no. 6, pp. 871-
Research, vol. 29, no. 2, pp. 371–391. 887.
[7] Anjos, M. F., Kennings, A. & Vannelli, A. (2005). [21] Alfi, A., Kalat, A. & Khooban, M. H. (2014).
A semidefinite optimization approach for the single Adaptive fuzzy sliding mode control for
row layout problem with unequal dimensions, Discrete synchronization of uncertain non-identical chaotic
Optimization, vol. 2, no. 2, pp. 113–122. systems using bacterial foraging optimization, Journal
of Intelligent and Fuzzy Systems, vol. 26, pp. 2567–
[8] Díaz, J., Petit, J. & Serna, M. (2002). A survey of 2576.
graph layout problems, ACM Computing Surveys, vol.
34, no. 3, pp. 313–356. [22] Alfi, A. & Khosravi, A. (2012). Constrained
nonlinear optimal control via a hybrid BA-SD,
[9] Petit, J. (2003). Experiments on the minimum linear International Journal on Engineering, vol. 25, no. 3, pp.
arrangement problem, Journal of Experimental 197-204.
Algorithmics, vol. 8, Article 2.3.
[23] Alfi, A. (2011). PSO with adaptive mutation and
[10] Heragu, S. S. & Kusiak, A. (1991). Efficient inertia weight and its application in parameter
models for the facility layout problem, European estimation of dynamic systems, ActaAutomatica, vol.
Journal of Operational Research, vol. 53, pp. 1–13. 37, no. 5, pp. 541-549.
[11] Anjos, M. F. & Vannelli, A. (2008). Computing [24] Khooban, M. H, Alfi, A., Nazeri, D. & Abadi, M,
globally optimal solutions for single-row layout (2013). Teaching-learning-based optimal interval type-
problems using semidefinite programming and cutting 2 fuzzy PID controller design: A nonholonomic
planes, INFORMS Journal on Computing, vol. 20, no. wheeled mobile robots, Robotica, vol. 31, no.7, pp.
4, pp. 611–617. 1059-1071.
[12]Amaral, A. R. S. (2009). A new lower bound for [25] Shahamat, H. & Pouyan, A. A. (2015).Feature
the single row facility layout problem, Discrete selection using genetic algorithm for classification of
Applied Mathematics. vol. 157, no. 1, pp. 183–190. schizophrenia using fMRI data, Journal of AI and Data
[13] Anjos, M. F. & Yen, G. (2009). Provably near- Mining. Vol. 3, no. 1, pp. 30-37.
optimal solutions for very large single-row facility [26] Neghabat, F. (1974). An efficient equipment
layout problems, Optimization Methods and Software. layout algorithm, Operations Research, vol. 22, pp.
vol. 24, no. 4–5, pp. 805–817. 622–628.
[14] Hungerländer, P. & Rendl, F. (2011). Semidefinite [27] Drezner, Z. (1987). A heuristic procedure for the
relaxations of ordering problems, Mathematical layout of a large number of facilities, Management
Programming, Series B, Vol. 140, no.1, pp. 77-97. Science, vol.7, no. 33, pp. 907–915.
[15] Mousavi, Y. & Alfi, A. (2015). A memetic [28] Heragu, S. S. & Kusiak, A. (1991). Efficient
algorithm applied to trajectory control by tuning of models for the facility layout problems, European
fractional order proportional-integral-derivative Journal of Operational Research, vol. 53, pp. 1–13.
controllers, Applied Soft Computing, vol. 36, pp.599–
617. [29] Kumar, K. R., Hadjinicola, G. C. & Lin, T. L.
(1995). A heuristic procedure for the single row facility
[16] Arab, A, & Alfi A. (2015).An adaptive gradient layout problem, European Journal of Operational
descent-based local search in memetic algorithm for Research, vol.87, no.1, pp. 65-73.
solving engineering optimization problems,
Information Sciences, vol. 299, pp. 117–142. [30] Djellab, H. & Gourgand, M. (2001). A new
heuristic procedure for the single-row facility layout
[17] Alfi, A & Fateh, M. M. (2011). Intelligent problem, International Journal of Computer Integrated
identification and control using improved fuzzy Manufacturing, vol. 14, pp. 270–280.
particle swarm optimization, Expert Systems with
Applications, vol. 38, pp.12312-12317. [31] Hungerländer, P. & Rendl, F. (2011). A
computational study for the single-row facility layout
[18] Alfi, A, & Modares, H. (2011). System problem, Technical Report, Alpen-Adria-Universität
identification and control using adaptive particle Klagenfurt, Austria.
swarm optimization, Journal of Applied Mathematical
[32] Romero, D. & anchez-Flores, A. S. (1990).
Modeling, vol. 35, pp.1210-1221.
Methods for the one-dimensional space allocation

46
Maadi et al. / Journal of AI and Data Mining, Vol 4, No 1, 2016.

problem, Computers & Operations Research, vol. 17, problem. In 4th IEEE conference on automation
pp. 465-473. science and engineering. Washington, DC, United
[33] Heragu, S. S. & Alfa, A. S. (1992). Experimental States.
analysis of simulated annealing based algorithms for
[46] Rajabioun, R. (2011). Cuckoo optimization
the facility layout problem, European Journal of
algorithm, Applied soft computing, vol. 11, no .8, pp.
Operational Research, vol. 57, pp. 190–202.
5508-5518.
[34] Gomes de Alvarenga, A., Negreiros-Gomes, F. J.
[47] Mahmoudi, S. & Lotfi, S. (2015). Modified
& Mestria, M. (2000). Meta-heuristic methods for a
cuckoo optimization algorithm (MCOA) to solve graph
class of the facility layout problem, Journal of
coloring problem. Applied soft computing, vol. 33, pp.
Intelligent Manufacturing, vol. 11, pp. 421–430.
48-64.
[35] Solimanpur, M., Vrat, P. & Shankar, R. (2005).An
[48] Shokri-Ghaleh, H. & Alfi, A. (2014). A
ant algorithm for the single row layout problem in
comparison between optimization algorithms applied
flexible manufacturing systems, Computers and
to synchronization of bilateral teleoperation systems
Operations Research, vol. 32, pp. 583–598.
against time delay and modeling uncertainties, Applied
[36] Datta, D., Amaral, A. R. S. & Figueira, J. R. Soft Computing, vol. 24, pp. 447-456.
(2011). Single row facility layout problem using a
[49] Khajeh, M. & Golzary, A. R. (2014). Synthesis of
permutation based genetic algorithm. European Journal
zinc oxide nanoparticles-chitosan for extraction of
of Operational Research., vol. 213, no.2, pp. 388–394.
methyl orange from water samples: Cuckoo
[37] Ficko, M., Brezocnik, M. & Balic, J. (2004), optimization algorithm-artificial neural network,
Designing the layout of single- and multiple-rows Spectrochimicaacta. Part A, Molecular and
flexible manufacturing system by genetic algorithms, biomolecular spectroscopy, vol. 131, pp.189-194.
Journal of Material Processing Technology, vol. 157-
[50] Stryczek, R. (2014). A meta-heuristic for fast
158, pp. 150–158.
machining error compensation, Journal of Intelligent
[38] Kumar, S., Asokan, P., Kumanan, S. & Varma, B. Manufacturing. pp. 1-12, doi: 10.1007/s10845-014-
(2008). Scatter search algorithm for single row layout 0945-0.
problem in fms. Advances in Production Engineering
[51] Ahirwal, M. K. Kumar, A. & Singh. G. K. (2013).
& Management, vol. 3, no. 4, pp. 193–204.
EEG/ERP adaptive noise canceller design with
[39] Kothari, R. & Ghosh, D. (2014). A scatter search Controlled Search Space (CSS) approach in cuckoo
algorithm for the single row facility layout problem, and other optimization algorithms, IEEE/ACM
Journal of Heuristics, vol. 20, pp.125–142. Transactions on Computational Biology and
Bioinformatics (TCBB), vol. 10, pp. 1491-1504.
[40] Samarghandi, H. & Eshghi, K. (2010). An
efficient tabu algorithm for the single row facility [52] Abolpour, B. & Mohebbi, A. (2013). Estimation
layout problem, European Journal of Operational of the compressive strength of 28-day-old concrete by
Research, vol. 205, no. 1, pp. 98–105. use of an adaptive cuckoo–fuzzy logic model, Research
on Chemical Intermediates, vol. 39, pp. 4001-4009.
[41] Samarghandi, H., Taabayan, P. & Jahantigh, F. F.
(2010). A particle swarm optimization for the single [53] Teimouri, R. & Sohrabpoor, H. (2013).
row facility layout problem, Computers and Operations Application of adaptive neuro-fuzzy inference system
Research, vol.58, no. 4, pp. 529–534. and cuckoo optimization algorithm for analyzing
electro chemical machining process, Frontiers of
[42] Akbari, M. & Maadi, M. (2011). Imperialist
Mechanical Engineering, vol. 8, pp.429-442.
competitive algorithm for solving single row facility
layout problem, 4th International Conference of Iranian [54] Mellal, M. A. & Williams, E. J. (2014). Parameter
Operations Research Society, Rasht, Guilan. optimization of advanced machining processes using
cuckoo optimization algorithm and hoopoe heuristic,
[43] Kunlei, L., Chaoyong, Z., Liang, G. & Xinyu, S.
Journal of Intelligent Manufacturing, pp. 1-16.doi:
(2011). Single row layout problem using an imperialist
10.1007/s10845-014-0925-4.
competitive algorithm, Proceedings of the 41st
International Conference on Computers & Industrial [55] Shokri-Ghaleh, H. & Alfi, A. (2014). Optimal
Engineering. Los Angeles, United States, pp. 578-586. synchronization of teleoperation systems via cuckoo
optimization algorithm. Nonlinear Dynamics, vol. 78,
[44] Braglia, M. (1996). Optimization of a simulated
no. 4, pp. 2359-2376.
annealing-based heuristic for single row machine
layout problem by genetic algorithm, International [56] Ghaemi M. & F .Derakhshani, M.(2014). Forest
Transactions in Operational Research, vol. 1, no. 3, pp. Optimization Algorithm, Expert Systems with
37–49. Applications, vol. 41, no. 15, pp. 6676-6687.
[45] Teo, Y. T. & Ponnambalam, S. G. (2008). A
hybrid ACO/PSO heuristic to solve single row layout

47
 ‫نشرهی هوش مصنوعی و داده کاوی‬

‫کاربرد دو الگوریتم جدید بهینهسازی فاخته و بهینهسازی جنگل برای حل مسألهی چیدمان تسهیالت‬
‫تکردیفه‬

‫‪2‬‬
‫منصوره معادی‪ ،1‬محمد جاویدنیا‪ *،2‬و مسعود قاسمی‬

‫‪1‬گروه مهندسی صنایع‪ ،‬دانشگاه دامغان‪ ،‬دامغان‪ ،‬ایران‪.‬‬

‫‪ 2‬گروه مهندسی کامپیوتر‪ ،‬دانشگاه دامغان‪ ،‬دامغان‪ ،‬ایران‪.‬‬

‫ارسال ‪5302/30/03‬؛ پذیرش‪5302/00/03‬‬

‫چکیده‪:‬‬

‫برانگیه‬ ‫امروزه با توجه به پیچیدگی ذاتی مسائل بهینهسازی واقعی‪ ،‬همواره توسعهی یک الگوریتم برای حل این مسائل بهه ننهوای یهک مو هول هال‬
‫مطرح بوده است‪ .‬مسألهی یدمای تسهیالت تک ردیفه‪ ،‬یک مسألهی ‪ NP-hard‬برای یدمای تعدادی تسهیالت مستطیلی شکل با طول متغیر بهر روی‬
‫یک سطر و با هدف حداقل سازی مجمول وزیدار فاصلهی بین هر جفت از تسهیالت است‪ .‬در این مقاله برای اولین بار دو الگوریتم بهینهسهازی فاتتهه و‬
‫بهینهسازی جنگل برای حل مسألهی یدمای تسهیالت تک ردیفه به کار گرفته شده و با یکدیگر مقایسه شدهاند‪ .‬نملگرهای هر دو الگوریتم با توجه بهه‬
‫مقایسهه‬ ‫ویژگیهای مسألهی فوق تغییر دادهشده و با مسأله وفق داده شدهاند و نتایج حاصل برای دو گروه از مثالهای معیار موجود در ادبیهات پهژوه‬
‫شدهاند‪ .‬در گروه اول تعداد تسهیالت کو کتر مساوی ‪ 03‬و در گروه دوم تعداد تسهیالت ب رگتر از ‪ 03‬است‪ .‬نتایج حاصل نشایدهنهدهی نملکهرد بهتهر‬
‫الگوریتم پیشنهادی مبتنی بر بهینهسازی فاتته نسبت به الگوریتم پیشنهادی مبتنی بر بهینهسازی جنگل‪ ،‬در هر دو گروه از مثالها هه از نرهر یهافتن‬
‫بهترین جواب و ه از نرر زمای محاسباتی الگوریتم است‪.‬‬

‫کلمات کلیدی‪ :‬مسألهی یدمای تسهیالت‪ ،‬مسألهی یدمای تسهیالت تک ردیفه‪ ،‬الگوریتم بهینهسازی فاتته‪ ،‬الگوریتم بهینهسازی جنگل‪.‬‬