This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2017.DOI

An Improved Atom Search Optimization

with Cellular Automata, a Lévy Flight
and an Adaptive Weight Strategy
FIRST A. PU SUN1 ,SECOND B. YONG ZHANG1 , THIRD C. JIALUN LIU1 ,AND FOURTH JIE BI1
1
School of Electronic and Information Engineering, University of Science and Technology Liaoning, Anshan, 114051, PR China (e-mail:892009384@qq.com)
1
School of Electronic and Information Engineering, University of Science and Technology Liaoning, Anshan, 114051, PR China (e-mail:zy9091@163.com)
1
School of Electronic and Information Engineering, University of Science and Technology Liaoning, Anshan, 114051, PR China (e-mail:liujialun0702@163.com)
1
School of Electronic and Information Engineering, University of Science and Technology Liaoning, Anshan, 114051, PR China (e-mail:1393723758@qq.com)
Corresponding author: Second B. YONG ZHANG (e-mail: zy9091@163.com).
This work was supported by the National Science Foundation of China under Grant No. 61473054.

ABSTRACT Atom search optimization (ASO) is a newly developed metaheuristic algorithm inspired
by molecular basis dynamics. The paramount challenge in ASO is that it is prone to stagnation in local
optima and premature convergence. To solve these issues, this paper uses a cellular automata structure, a
Lévy flight and adaptive weight strategies to help ASO balance exploration and exploitation; the proposed
method is named CALFASO. First, the cellular automata structure provides more information exchange for
the population and enriches the diversity of the population. Second, a Lévy flight has the characteristics of
a random walk, its short-step long-term search can improve the global search capability of the algorithm,
and its short-term long-step mutations can help the algorithm jump out of local optima. Third, adaptive
weights can effectively improve convergence performance. Finally, the proposed CALFASO was evaluated
on the CEC2017 test set and real-world engineering problems and compared with classical and novel
evolutionary algorithms to verify its performance. The experimental results and statistical analysis show that
the performance of the proposed CALFASO algorithm is better than that of the other selected algorithms.

INDEX TERMS Atom search optimization, Evolutionary algorithms, Cellular automata, Lévy flight

I. INTRODUCTION onfly algorithm (DA) [14], water wave optimization (WWO)

[15], cuckoo search (CS) [16], [17], and moth flame opti-
Over the past 30 years, metaheuristic algorithms have re- mization (MFO) [18].
ceived extensive attention and rapid development due to their
simplicity, efficiency and cleverness [1]. Different sources The PSO algorithm mimics the foraging behavior of flocks
of inspiration, optimization mechanisms and convergence of birds proposed by Eberhart and Kennedy [6] in 1995. DE
performance contribute to the design of different versions [19] has mainly been used to solve real-number optimiza-
of metaheuristic algorithms for solving real-world problems tion problems. The GA [10] is a computational model that
[2]–[4]. From the perspective of population, metaheuristic simulates the natural selection and genetic mechanism of
algorithms are roughly divided into two categories: (1) single Darwin’s biological evolution and was proposed by Holland
solution-based algorithms such as simulated annealing (SA) [10] in 1975. The inspiration for the GWO [11] is the social
[5], which starts from an initial solution and performs op- levels and hunting mechanism of wolves. The GSA [12] is
timization in the search space by simulating the annealing an optimization algorithm based on the interaction of gravity
process of solid matter in physics; and (2) population-based and mass. The NMR [13] algorithm mimics the mating
metaheuristics, where currently popular algorithms include patterns of NMRs in nature. The multiobjective dragonfly
particle swarm optimization (PSO) [6]–[8], differential evo- algorithm [14] is utilized to obtain nondominated Pareto
lution (DE) [9], the genetic algorithm (GA) [10], the grey optimal solutions, and the WWO [15] algorithm is used to
wolf optimizer (GWO) [11], and the gravitational search optimize connection weights in neural networks. Salgotra et
algorithm (GSA) [12]. Other recently introduced algorithms al. proposed a new CS [16], [17] algorithm to improve the
include the naked mole rat (NMR) algorithm [13], the drag- performance of CS. MFO [18] is a recent nature-inspired

VOLUME 4, 2016 1

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

algorithm motivated by the transverse orientation of moths convergence due to attraction, and the randomness of the LF
in nature. In summary, it can be seen that the inspiration and strategy increases the ability to jump, preventing the atoms
mechanism of metaheuristic algorithms based on populations from agglomerating in the later stage and falling into a local
are different, but they have obvious advantages compared optimum. Meanwhile, adaptive weights are used to balance
with those of solution-based algorithms: i) Information ex- exploration and exploitation. To the best of the authors’
change between populations, mutual learning, and common knowledge, there is no study proposed in the literature to
growth can improve the convergence performance and op- improve the ASO with a CA structure, LFs and adaptive
timization ability of the algorithm. ii) Collaboration among weights. The main contributions and motivation of this paper
populations balances exploration and exploitation better. can be summarized as follows:
Atom search optimization (ASO) [20] is a novel physical 1) Cellular automata (CA) are used to enhance information
metaheuristic algorithm that is inspired by molecular dynam- exchange and enrich population diversity. The CA structure
ics. ASO mimics the interaction between atoms (attraction can provide more effective information exchange and prevent
and repulsion) resulting from the Lennard-Jones potential the decrease in exploration ability caused by excessive atom
and the constraint forces resulting from the bond-length attraction. In this paper, we use six classic CA structures and
potential. In ASO, the position of each atom in the search choose the best structure for comparison with the selected
space represents the solution measured by the mass. The algorithm based on the sensitivity test.
heavier the mass is, the better the solution is, and vice 2) The LF strategy can help an atom generate more random
versa. Light atoms are good for exploration, and heavy atoms operations during the iteration process and help the atom
are good for exploitation. Through a series of benchmark jump out of local optima. Research has shown that the
function tests and successful application in the dispersion LF strategy is applicable to many metaheuristic algorithms
coefficient estimation of groundwater, ASO has shown strong [25]–[27]. The Lévy index β for controlling stability is a
competitiveness and performance. Some efforts on ASO have key parameter, and changes in the parameter β will affect
been made in terms of real-world problems and theoretical the distribution of the LF. According to related research
research. An ASO algorithm-based hybrid antenna array with [28], [29], when β ∈ [1, 2], the LF will obtain the best
receive beamforming was utilized to control sidelobe level performance. Therefore, we performed constant and random
and steering [21] in 2019. The ASO algorithm and sine- number sensitivity tests for β to ensure the stability of the
cosine algorithm (SCA) were integrated for automatic data algorithm. Through the sensitivity tests of the CA and LF
clustering as proposed by Mohamed Abd Elaziz et al. [22] in strategies, the best structure of the CA and parameters β
2019. Ahmed M. Agwa et al. [23] used the ASO algorithm of the LF were selected, and the comparison test for the
for the steady-state modeling of fuel cells in 2019. Baran algorithms was performed.
[24] proposed a chaotic atom search optimization algorithm 3) The adaptive weight is a linearly decreasing function; in
(ChASO) to help the ASO algorithm balance exploration the early stages of the iteration, global search is guaranteed
and exploitation, which was successfully applied to DC mo- to occur, and in the later stages of the iteration, the stability
tor speed control. However, similar to other metaheuristic of the atom is maintained, and small-scale exploration is
algorithms, although the original ASO algorithm is simple performed. Therefore, the adaptive weight is used to help the
and easy to implement, it suffers from two main drawbacks: population transition from exploration to exploitation during
premature convergence and falling into local optima. In ASO, the iterative process.
in the early stage of the iteration, there is a wide range 4) The proposed CALFASO algorithm was tested on the
of mutual attraction between atoms. Attraction will cause CEC2017 [30]–[32] test set and engineering problems, and
atoms to tend to aggregate, lose the ability to explore the it was compared with classic, novel and high-performing
entire search space, and converge prematurely. At the end variants of the other algorithms. Meanwhile, to obtain the
of the iteration, the weak repulsive force cannot truly help best CA structure and LF parameters, sensitivity analysis
the atoms jump out of local optima, and each atom will was performed. To more clearly show the contribution of
also be attracted by the best atom (with the best fitness), the different methods proposed, a synergy analysis of each
causing the atoms to gather near the best atom, which causes strategy was performed.
it to fall into a local optimum. Premature convergence affects 5) Cellular automata (CA), a Lévy Flight (LF) and adaptive
exploration, and falling into local optima affects exploitation. weights are applied to improve the performance of the ASO
Therefore, balancing exploration and exploitation is one of algorithm. The experimental results and statistical analysis
the core problems of the algorithm. Exploration emphasizes also verify the correctness of the hypothesis. Detailed statis-
global search—conducting extensive searches through the tical analysis are given in this paper based on qualitative anal-
entire search space—with the goal of finding more promising ysis, quantitative analysis, convergence preference, pairwise
optimal areas. Exploitation emphasizes local search and finds comparative analysis (Wilcoxon signed rank test) and the
optimal solutions in promising areas. Therefore, the motiva- Friedman test. The statistical analysis more clearly validates
tion of our proposed method is to help balance the atomic the effectiveness of the improvement.
interactions in the ASO algorithm. The CA structure is used The paper is structured as follows. In Section II, the origi-
to provide good information exchange to prevent premature nal ASO algorithm is briefly introduced. The detailed presen-
2 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

tation of the proposed CALFASO algorithm is described in difference vector. rij is the Euclidean distance between the
Section III. Benchmark functions, parameter settings, sensi- ith and the jth atoms.
tivity tests and engineering problems are provided in Section
IV. The experimental results and statistical analysis will be t − 1 3 − 20t
η(t) = χ(1 − ) e T (4)
discussed in Section V. Conclusions and prospects for future T
research will be given in Section VI.
where χ is the depth weight and is set to 50 in this
paper. T is the maximum number of iterations. Therefore,
II. ATOM SEARCH OPTIMIZATION (ASO)
there are two main factors affecting the interaction force
In ASO [20], the position of each atom in the feasible
F : η(t) and h. The functional behavior of F , with differ-
solution space represents a solution and is described by its
ent η(η= 1, 5, 10, 30, 50) corresponding to h values ranging
mass. Heavier atoms have smaller acceleration and repre-
from 0.9 to 2, is illustrated in Fig. 1. The scaled distance
sent better solutions. Lighter atoms have greater acceleration
between two atoms is defined by 5.
and represent worse solutions. There are interaction forces
between atoms: attraction and repulsion. In the iterative  rij (t)
process, repulsion can effectively prevent the premature con-  hmin , σ(t) < hmin

rij (t) rij (t)
σ(t) , hmin ≤ σ(t) ≤ hmax
vergence of an atom, and certain attraction can guide the hij (t) = (5)
atom’s movement. In atomic systems, there are three main rij (t)

hmax , σ(t) > hmax

mathematical representations: interaction forces, geometric
constraints, and atomic motion. Through atomic interaction, where hmin and hmax denote the lower and upper bounds
attraction and repulsion balance exploration and exploitation of the scaled distance h, respectively, and are expressed as 6.
well.

hmin = g0 + g(t)
A. INTERACTION FORCE (6)
hmax = u
The L-J potential between atoms is the source of atomic
interaction forces, and the sum of the forces acting on the where g0 is the lowest limit, set to 1.1, and u is the upper
ith atom is described as (1). limit, set to 1.24. g(t) is the drift factor, which can make the
X algorithm drift from exploration to exploitation and is defined
Fid (t) = randj Fijd (t) (1) by 7.
j∈Kbest
 
where randj [0,1] is a random number. Kbest is a collec- π t
tion containing the K best atoms, which consist of the first K g(t) = 0.1 × sin × (7)
2 T
atoms with the best fitness function values. During the early
iterations, to perform more exploration, each atom needs to The length scale σ(t) in 5 denotes the collision diameter
interact with as many atoms with better fitness values as pos- and is defined by 8.
sible, which are its K neighbors. During the final iterations, P
to perform more exploitation, each atom needs to interact
xij (t)

with as few atoms with better fitness values as possible as j∈Kbest
σ(t) = xij (t), (8)

K(t)

its K neighbors. Therefore, K gradually decreases as the



iteration progresses and is described by 2. 2

r
t
K(t) = N − (N − 2) × (2) B. GEOMETRIC CONSTRAINTS
T
Geometric constraints in molecular dynamics appear as co-
where N is the size of the atomic population, t is the valent bond connections in atomic motion. To solve the
current iteration and T is the maximum number of iterations. optimization problem, assume that each atom is connected
Fijd (t) is the interaction force applied to the ith atom by the to the atom with the best fitness value, so the constraint force
jth atom at the tth iteration to solve the optimization problem can be described by (9).
and is described by (3).
Gdi (t) = λ(t)(xdbest (t) − xdi (t)) (9)
h i ~r
−13 −7 ij
Fij (t) = −η(t) 2(hij (t)) − (hij (t)) (3) where xbest (t) is the position of the best atom at the tth
rij
iteration. The best atom is the atom with the best fitness. The
where η(t) is the depth function to adjust the repulsion Lagrangian multiplier is defined by (10).
region or attraction region and can be expressed by (4).
rij 20t
hij (t) = σ(t) is the ratio of the distance between two atoms λ(t) = θe− T (10)
to the length scale, which in this paper is called the scaled
distance between two atoms. ~rij = ~xj − ~xi is the position where θ is the multiplier weight and is set to 0.2.
VOLUME 4, 2016 3

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

space. vid ∈ [−vmax , vmax ], where vmax is set by the user.

xdi ∈ [xdmin , xdmax ], where xdmin and xdmax are the lower and
upper boundaries of the dth dimension, respectively. adi is the
acceleration of each atom.
For a more detailed explanation of ASO, one can refer to
[20].

III. THE PROPOSED CALFASO ALGORITHM

The validity of ASO was tested qualitatively and quantita-
tively by a series of benchmark functions and has been suc-
cessfully applied to hydrogeological parameter estimation.
However, the ASO algorithm still faces the difficulties of
premature convergence and local optimization. In this paper,
to solve these issues, three strategies will help ASO balance
exploration and exploitation: a cellular automata (CA) struc-
ture, a Lévy Flight and adaptive weights.
CA have become a more popular tool in scientific research
and are widely applied in a variety of fields. The innovative
rules of CA give it a strong vitality. The most well-known
CA rule, the “game of life”, was conceived by Conway [34]
in the late 1960s. Some efforts regarding metaheuristics with
a topological structure have been made in recent years and
provide a theoretical basis for such metaheuristics. Shi et
al. proposed cellular PSO [35] by incorporating a cellular
topological structure into PSO to improve the performance of
FIGURE 1. The interaction force F versus the scaled distance (h) with PSO. Yi [36] et al. proposed a modified harmony search al-
different depth values (η= 1, 5, 10, 30 and 50) [20].
gorithm with cellular local search to improve the exploitation
ability. Alba et al. proposed a cellular GA [37] by combining
C. ATOMIC MOTION a GA with cellular automata to enhance the performance
Atomic systems conform to the laws of classical mechanics of the original GA. The LF concept was introduced to the
[33]. With the joint contributions of interaction forces and swarm intelligence algorithm to improve performance and
geometric constraints, the equation for the acceleration of make the swarms more realistic. It has been verified that
each atom is defined according to (11). Lévy-based movements can be regarded as optimal searching
(hunting) approaches for foragers and hunters in nondestruc-
Fid (t) Gd (t) 3 − 20t
adi (t) = md (t)
+ mid (t) = −χ (1 − t−1 T ) e
T tive foraging circumstances [38]. Yang and Deb [39], [40]
i i
P randj [2×(hij (t)) −13 −7
−(hij (t)) ] employed an LF distribution to generate new search agents in
× mi (t) (11) the cuckoo search (CS) algorithm. LF is also used to improve
j∈Kbest
(xd (t)−xdi (t)) 20t (x
d
(t)−xd (t)) the performance of other optimization algorithms, such as
· k~xji (t),~xj (t)k + θe− T bestmi (t) i PSO [25], [29], the FA [41], and the ABC algorithm [42].
2

where mi (t) is the mass of each atom and can be calculated Similarly, adaptive weight is a very widely used and simple
by the expressions (12) and (13). method. The most classic is the weight factor in PSO [6] to
F iti (t)−F itbest (t)
help the algorithm transition from exploration to exploita-
− F it tion in iteration. An example of recent success is hetero-
Mi (t) = e worst (t)−F itbest (t) (12)
geneous comprehensive learning particle swarm optimiza-
Mi (t)
mi (t) = PN (13) tion (HCLPSO) [43], which sets different adaptive weights
j=1 Mj (t) and helps different subpopulations complete their tasks. The
where F itbest (t) and F itworst (t) are the best and worst three strategies have strong theoretical support, which has
fitness values at the tth iteration, respectively. prompted us to use them to improve the performance of the
Therefore, the velocity and position can be expressed as original ASO.
(14) and (15).
A. CELLULAR AUTOMATA (CA) STRUCTURE
vid (t + 1) = randdi vid (t) + adi (t) (14) CA were first proposed by Von Neumann and Ulam in
1966. In contrast to general dynamics models, CA are not
xdi (t + 1) = xdi (t) + vid (t + 1) (15)
determined by strictly defined physical equations or functions
where i = 1, 2, · · · , N and N is the size of the popu- but are constituted by a series of rules constructed by models.
lation. d = 1, 2, · · · , D, D is the dimension of the search CA are characterized by discrete time, space, and state. Each
4 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

variable takes only a limited number of states, and the rules of

state change are local in time and space. An excellent book,
A New Kind of Science, written by Wolfram [44], introduced
the development and application of CA in detail and attracted
widespread attention from scientific researchers.
With the deepening of research, CA performed well in
many fields, such as physics, computer science and traffic
[45]–[47]. CA are a group of cells distributed under a specific
grid in which each cell updates its state based on the state of
its neighbors. In this paper, to combine CA with the ASO
algorithm, five important elements in CA are redefined: the
cell, cell state, cell space, neighborhood and transition rule.
Cell: Cells originally represented equally spaced points on
a two-dimensional grid. In this paper, each cell represents
an atom. The size of the cell is determined based on the
population. FIGURE 2. The neighborhood structure of cellular automata. Taking L5 as an
example, construct a 10 by 5 grid, put 50 individuals in the grid, and label each
Cell state: The cell state is reflected by two aspects: the individual. The red callouts represent the center cells, and the blue callouts
current state and next state. The current state will change to represent the neighbors surrounding a center cell.
the next state according to the conversion transition rule.
Neighborhood: Neighbors are a group of cells around a
center cell. In this paper, neighbors are other atoms connected
by a center atom. To show the structure of different neighbors
more clearly, the following definitions are established. The
neighbors in the lattice grid are defined in terms of direction
and radius. Two common structural labels are Ln and Cn.
Ln indicates that the center cell has n − 1 nearest neighbors
surrounding it in specific directions (top, bottom, left, right).
Cn represents that the center cell has n − 1 nearest neighbors
surrounding it in specific directions (top, bottom, left, right,
diagonal). This article uses six classic structures [48], as
shown in Fig. 3.
The distribution of atoms in the search space is random.
To determine the neighbor positions of each atom in the CA,
a schematic diagram of the CA neighbor structure under the FIGURE 3. Different neighborhood structures of cellular automata (L5, L9, C9,
C13, C21 and C25).
L5 structure is drawn. The L5 structure is used to construct
neighbors from the top, bottom, left, and right directions, as
shown in Fig. 2. With a population of 50, we can construct transition rule is defined as (16).
a 10 by 5 or 5 by 10 grid, put all the atoms into the grid
f itness(Sit (Xi )), ...
 
according to a certain number from 1 to 50, and construct t+1
f itness(Si (Xi )) = min
their corresponding neighbors according to the rules. For f itness(Sit (Xi+r1 )
atoms at the edge of the lattice, such as the atom numbered (16)
one (shown in red), its neighbor numbers are 46, 6, 5, and where f itness (Sit (Xit )) represents the fitness of the cur-
2 (shown in blue); for atoms inside the grid, such as the rent solution Xi . Therefore, we can obtain the best neighbor
atom numbered 18 (shown in red), its neighbors are the in the CA structure through (16) to update the state of the
atoms numbered 13, 23, 17, and 19 (shown in blue), and center cell. Therefore, through the CA structure, each atom
they are positioned directly on the top, bottom, left and right, will be updated according to the states of the best neighbors
respectively. around it. The equation of velocity can be described by (17).
Transition rule: The transition rule is an important basis
for determining how each atom changes from the current vid (t + 1) = randdi vid (t) + randdi (Sid − xdi (t)) + adi (17)
state to the next state. The next state of the center cell is
determined by the current state of its neighbors. The state
of the neighbors’ best position Xi in the current atom is B. LÉVY FLIGHT
defined as Sit (Xn ). To ensure the effectiveness and speed of The LF mechanism is a strategy based on the random walks
information exchange, we update according to the neighbors of living organisms. Random walks are characterized by short
with the best state among the surrounding neighbors. The steps for long-term search and occasional long-step jumping,
VOLUME 4, 2016 5

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

and long-step jumping is characterized by directional vari-

ability. The LF’s special mechanism enables organisms to 350

balance exploration and exploitation [42] [25]. The short-step 300

search ensures that the population performs a global search 250

in a wider space. Occasionally, long-step mutations can help 200

the population jump out of local optima. This randomness 150
avoids the premature aggregation of atoms in the interaction. 100
Lévy motion is regarded as a type of non-Gaussian random
50
procedure whose random steps are determined based on a
0
Lévy stable distribution [49]. From Fig. 4, we can conclude
-50
that the trajectory of the LF shows a majority of short-
step walks and a few long-step mutations, which effectively -100

prevents the atoms from stagnation due to falling into local -150
-400 -350 -300 -250 -200 -150 -100 -50 0 50 100
optima during the movement and helps atoms perform global
search on a wider area. The equation of an LF distribution
FIGURE 4. Trajectory of the Lévy flight (the number of steps is 500, the
based on a Fourier transform can be described by (18) [42]. number of dimensions is 3, and the random number β = 1.4; the random walk
h i is characterized by short steps for long-term search and occasional long-step
β jumping).
F (k) = exp −α|k| ) (18)

where the skewness and scale factor α ∈ [−1, 1]. Deb and
adjusted. The adaptive weight parameters are set according
Lee’s [50] [51] research found that different β values affect
to classical PSO [6]. In this paper, the adaptive weight w is
the distribution of an LF. The larger β is, the smaller the step
defined by (23).
size, and vice versa. Therefore, to obtain better parameters
and ensure the stability of the Lévy distribution, the Man- w = 0.9 − 0.5 ∗ t/T (23)
tegna technique [52] is used to obtain the LF parameters. The
step size is described by (19). Therefore, through analysis of the three strategies, we
find that they can help atoms jump out of local optima,
→
prevent premature convergence, and balance exploration and
u →
rand(1, Dim) ⊕ levi(β) ∼ 0.01 ( (t) − X α (t)) (19)
−β X
exploitation; the velocity is defined by (24).
v
where parameters u and v are obtained by the normal vid (t + 1) = w ∗ randdi vid (t). ∗ LF (β)
(24)
distribution described by (20), where σu is defined as (21). + randdi (Sid − xdi (t)) + adi

u ∼ N (0, σu2 ), v ∼ N (0, σv2 ) (20) D. FLOWCHART OF THE CALFASO ALGORITHM

In summary, the proposed CALFASO algorithm is a com-
" # β1 bination of three strategies: cellular automata, a Lévy flight
Γ(1 + β) sin( πβ
2 ) and adaptive weights. To more clearly describe the flow of the
σu = , σv = 1 (21)
Γ( 1+β
β−1
proposed CALFASO algorithm and show the improvement to
2 )β × 2
2
the ASO algorithm achieved by the three proposed strategies,
where Γ represents the gamma function. According to the the flowchart of the CALFASO algorithm is shown in Fig.
relevant literature, the optimal value range for the parameter 5. Fig. 5 clearly illustrates the structure of the proposed
β is [1, 2]. To obtain the best value of β, we executed a CALFASO algorithm. The structure of the proposed algo-
sensitivity test on β. The details of the sensitivity test for rithm is mainly composed of two parts. The main sequential
β are described in Section IV-B. Therefore, through the structure consists of the mechanism equations. The condi-
random walk strategy of the LF, the atomic group can balance tional structure uses the maximum number of calculations
exploration and exploitation better. The equation of velocity to define the termination condition. The CA structure helps
is described by (22). an atom generate new velocity components to obtain a better
velocity direction. The LF’s random walk strategy guarantees
vid (t + 1) = randdi vid (t). ∗ LF (β)
(22) jumping and avoids local optima. Adaptive weights balance
+ randdi (Sid − xdi (t)) + adi
the transition from exploration to exploitation throughout
the iteration. As shown in Fig. 5, the purpose of the three
C. ADAPTIVE WEIGHT improvement strategies we propose is mainly to enhance the
In the early stages of the iteration, the atomic group will atom’s performance by changing its velocity.
search in a wider range, that is, perform global search, and Through the three strategies, we balance exploration and
in the later stages of the iteration, local search is mainly exploitation better. First, the basic parameters are initialized.
performed. To ensure the atomic movement trend goes from Second, the algorithm enters an iterative process with the
oscillating to stable, an adaptive weight is introduced to be maximum number of calculations defining the termination
6 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

and composite functions f20 − f30 . Function 2 is removed

due to its instability in high dimensions. A unimodal function
Start has only one global optimal value, which is appropriate for
testing the exploitation performance of the algorithm. Simple
Initialize Parameters: Population Size, Allowable
Boundary, Dimensionality, Maximum number of
multimodal functions have multiple local optima, which is
evaluations appropriate for testing the algorithm’s ability to avoid local
optima. A hybrid function usually consists of unimodal and
Initialize Position X and Velocity V
Set the Fitbest = Inf
multimodal functions, and it can verify the comprehensive
ability of the algorithm to avoid premature convergence.
Composite functions consist of a number of variables, each
Maximum number of Yes of which represents a different function, which usually in-
evaluations satisfied
clude basic and hybrid functions. Variable uncertainty and
No complexity are closer to real-world problems, and this type
Calculate the fitness value Fiti of function can more accurately judge the performance of the
algorithm [12].

If Fiti < Fitbest

No
B. PARAMETER SETTINGS
The details of the nine algorithms compared with the pro-
Yes
posed CALFASO are shown in Table 2. The parameters
Fitbest = Fiti , Xbest =Xi of each selected algorithm are set according to the sugges-
tions of the corresponding references. To ensure the validity
Calculate the mass using (12) and (13)
and fairness of the test, all algorithms have the following
parameter settings: The population size is set to 50. The
Define its K neighbors using (2) maximum number of evaluations is 300,000. Each time the
fitness value is calculated, it is recorded as an evaluation
Calculate the interaction force Fi using (1) number to ensure the complexity of the calculation, and each
algorithm runs independently 30 times. To test the reliability
Calculate the constraint force Gi using (9) of the algorithm, the CEC2017 test set was performed on 3
dimensions: 30, 50, and 100. To obtain the best CA structure
Calculate the acceleration using (11) and LF parameter β, sensitivity tests [53] of CA and β
were performed in this subsection. As shown in Table 3,
Calculate the best CA structure, LF parameter, adaptive we tested different CA structures on 30 dimensions, and the
weight, apply to (24) to update velocity
results showed that the L5 structure of CA achieved the best
performance and ultimately ranked first. As shown in Table
Update the position using (15) 4, four different types of functions were selected for the
sensitivity analysis of β: the unimodal function f3 , simple
Output the best solution Xbest
multimodal function f6 , hybrid function f15 and composite
function f22 . According to the relevant references [3], 10
End different parameters were selected for testing: constants from
1.2 to 2.0 and random numbers [1, 2]. The average result
of the minimum fitness (Mean) is used as an indicator, and
FIGURE 5. The flow chart of the proposed CALFASO algorithm.
the average value of the minimum fitness (Best Number) is
recorded. The final parameter selection is performed accord-
ing to the average results. From Table 4, when β = 1.4,
the best performance is achieved. Through the sensitivity test
condition. Finally, when the maximum number of evaluations
for the CA structure and β of the LF, the L5 structure and
is reached, the optimal solution is output.
β = 1.4 are selected as the final parameters in this paper.
Additionally, to better demonstrate the effectiveness of our
IV. BENCHMARK FUNCTIONS, PARAMETER SETTINGS proposed method, a synergy analysis was performed, and the
AND ENGINEERING PROBLEMS original ASO algorithm and different strategies were tested
A. BENCHMARK FUNCTIONS separately and combined for analysis. The results of the
To extensively and effectively verify the performance of the synergy analysis are shown in Table 5. We selected the best
proposed CALFASO, the CEC2017 test set was used [30]– CA structure (L5) and LF parameters for the synergy analysis
[32]. Table 1 lists the functions in the test set. These functions based on the sensitivity analysis. As shown in Table 5, when
are of four main types: unimodal functions f1 − f3 , simple each strategy works independently, the CA structure achieves
multimodal functions f4 − f10 , hybrid functions f11 − f20 , the best results, followed by w and LF. When two strategies
VOLUME 4, 2016 7

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

shear stress, surge frequency and outside diameter. The three

main design variables are the wire diameter 1 ≤ x1 ≤ 70,
the mean coil diameter 0.6 ≤ x2 ≤ 3 and the number of
active coils 0.207 ≤ x3 ≤ 0.5. The target optimal value is
0.0126652. The problem can be expressed as 26.
2

 min f (x) = (x33 + 2)x2 x1

 x x3
 g1 (x)=1 − 71785x 4 ≤ 0

 2
 1
4x22 −x1 x2
s.t
 g2 (x) = + 1 −1≤0
12566(x2 x1 −x41 ) 5108x1
3
(26)
140.45x1
g3 (x) = 1 − x2 x3 ≤ 0


FIGURE 6. Compound gear train. The number of teeth on gears A, B, D and 

 2
F. [54].
g4 (x) = x11.5
+x2
−1≤0


V. EXPERIMENTAL RESULTS AND STATISTICAL

are combined, CA-LF achieves the best results, followed
ANALYSIS
by CA-w and LF-w. When three strategies are combined,
In this section, to comprehensively evaluate the performance
we obtain the best results in all collaborative experiments.
of the proposed CALFASO algorithm, we conducted qualita-
Therefore, we find that the CA structure contributes the most
tive and quantitative analysis of the experimental results. Ad-
to the improvement of algorithm performance. At the same
ditionally, drawing a convergence curve to intuitively observe
time, the combination of different strategies has a synergistic
the trajectory of an atom is of great significance for analyzing
effect, and they can enhance each other. This is of great
global search and local search. To ensure the precision of the
significance for improving the performance of the algorithm.
results, statistical analysis was performed. The experimental
results and statistical analysis show that the performance of
C. CALFASO FOR ENGINEERING PROBLEMS
the proposed CALFASO algorithm is better than that of the
In this section, we will evaluate the performance of the other selected algorithms in this paper.
proposed CALFASO on two real-world engineering prob-
lems: an unconstrained problem, gear train design, and a
A. QUALITATIVE RESULTS OF CALFASO
constrained problem, tension/compression spring design. En-
In this section, four different functions are randomly selected
gineering problems test the stability of the algorithm. An
from all benchmark functions to verify the quality of the
excellent algorithm should achieve good results in both func-
proposed CALFASO. They are the unimodal function f1 , the
tion tests and real-world problems. To ensure the validity
simple multimodal function f9 , and the composite functions
and fairness of the test, all algorithms have the following
f22 and f27 ; different types of functions can evaluate the
parameter settings: The population size is set to 50. The
performance of the proposed CALFASO algorithm from
maximum number of evaluations is 300,000. Each time the
multiple perspectives. To simplify the problem, the dimen-
fitness value is calculated, it is recorded as an evaluation
sion is set to 2, the number of iterations is set to 150, and
number to ensure the complexity of the calculation, and each
the whole process is performed with 4 atoms. There are four
algorithm runs independently 30 times.
main qualitative indicators to evaluate the performance of the
proposed CALFASO: the search process, trajectory curve,
1) Gear Train Design
convergence curve, and trajectory curve of acceleration.
For the gear train design in Fig. 6, the aim is to produce a
As shown in Figs. 7-10, the search process intuitively
gear ratio as close as possible to 1/6.931 [54]. The goal of this
describes the search history of the positions of the four atoms
problem is to minimize the square of the difference between
during the iterative process. Throughout the search space,
the desired gear ratio and the gear ratio of the current design.
sparse areas represent global exploration and dense areas
There are four main variables x1 (Td ), x2 (Tb ), x3 (Ta ), and
represent local search. The entire population executes the
x4 (Tf ), where Td , Tb , Ta , and Tf are the number of teeth on
process from exploration to exploitation. There is more atom
gears A, B, D and F, respectively. For each gear, the minimum
search history near the global optimal solution. Extensive
number of teeth is 12, the maximum is 60, and the range is
global search helps populations explore more promising ar-
x1, x2, x3, x4 ∈ [12, 60]. The target optimal value is 2.7e −
eas.
12. Therefore, the objective function can be expressed as 25.
Figs. 7-10 show the trajectory curves of four different
M inf (x) = (1/6.931 − x1 x2 /x3 x4 )2 (25) functions in the first and second dimensions. Although the
functions are different, they have similar characteristics. Each
atom executes a large-scale oscillating process at the begin-
2) Tension/compression spring design ning of the iteration, which means that in the early global
For the tension/compression spring design, the goal of this search, after the update, the atom’s movement gradually
problem is to minimize the weight of a tension/compression stabilizes, indicating that the atom is looking for a global
spring while obeying constraints on the minimum deflection, optimal value in a promising region.
8 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

In Figs. 7-10, the convergence curves of different functions the composite functions f21 − f30 , the proposed CALFASO
are shown. A convergence curve can clearly show the con- algorithm achieves the best average results on functions
vergence rate and convergence accuracy. For the unimodal f21 − f24 and f29 . The ASO algorithms achieved the best
function f1 , simple multimodal function f9 , and composite average results on functions f22 and f26 . SPSO2011, VS,
function f22 , after a short period of exploration, the conver- DSA and BMA achieved the best average results on functions
gence curve drops rapidly. For the composite function f27 , f28 , f25 , f27 , f30 , respectively. The proposed CALFASO
the convergence curve drops slowly, indicating that more algorithm has a total rank of 83.5, and it ultimately ranks
local optimal operations are being performed. Therefore, first, with significantly better performance than other selected
we can conclude that complex functions execute a longer algorithms. It is worth noting that the proposed CALFASO
convergence process to ensure the accuracy of convergence. algorithm has more advantages in solving complex func-
Different algorithms have their own key factors, and acceler- tions because it benefits from the combination of the CA
ation is a core parameter in this paper. It can be seen from structure’s information exchange and the LF strategy’s long
the trajectory of acceleration that the acceleration tends to and short jumps. These strategies significantly improve the
stabilize after a brief small-scale oscillation, which is of great algorithm’s convergence performance and its ability to avoid
significance for balancing exploration and exploitation. local optima.
Through the above qualitative analysis, the four qualitative Tables 7-8 present the results for 50 and 100 dimen-
parameters show the performance of the proposed CAL- sions, respectively. On 50 and 100 dimensions, the proposed
FASO algorithm from different perspectives. In summary, the CALFASO algorithm has 13 and 15 functions ranked first,
proposed CALFASO algorithm has good convergence and respectively. For 50 dimensions, its total rank is 65, and
optimal trajectory, and it balances exploration and exploita- its final rank is 1. On 100 dimensions, its total rank is 65,
tion well. and its final rank is 1. The experimental results show that
with the increase of dimensions, the performance of the
B. RESULTS OF DIFFERENT ALGORITHMS ON proposed CALFASO algorithm is better than that of the other
CEC2017 BENCHMARK FUNCTIONS algorithms, which proves its stability and anti-interference
In this section, the proposed CALFASO algorithm is com- to a certain extent. In summary, the proposed CALFASO
pared with 9 other selected algorithms on the CEC2017 test algorithm shows better performance on the CEC2017 test
set. To ensure the validity and fairness of the results, the set, balancing the exploration and exploitation better than the
following parameters were set: The population size was 50, other algorithms.
the maximum number of evaluations was 300,000, and the
number of independent runs was 30. We used 3 dimensions C. CONVERGENCE PREFERENCE OF THE
(30, 50, and 100) to test the stability of the algorithm. To ALGORITHMS
observe the experimental results more clearly, each function There is no optimization algorithm that can best perform
is ranked according to the best average result, and the ranking for all different types of problems, hence the performance
of each function is arithmetically combined to obtain the total evaluation of the proposed algorithm should come from
rank and the final rank. different perspectives by using multifarious benchmark func-
Table 6 shows the test results on 30 dimensions. The tions. Therefore, in this section, different types of functions
proposed CALFASO algorithm ranks first on 13 functions, on different dimensions are randomly selected, and their
and its performance is significantly better than that of other convergence processes are constructed to evaluate the perfor-
algorithms. For the unimodal functions, SPSO2011, VS and mance of the proposed CALFASO algorithm. The maximum
the BMA achieved the best average results on function f3 . number of evaluations FES is 300,000. To save calculation
The BMA also achieved the best average results on function costs, the result is calculated every 100 times. The conver-
f1 . The proposed CALFASO algorithm obtained satisfactory gence graph intuitively shows the convergence performance
results on both functions f1 and f3 . For the simple multi- of an algorithm and is an important metric for evaluating the
modal functions f4 − f10 , the performance of the proposed algorithm. As shown in Fig. 11, we selected the convergence
CALFASO algorithm on functions f5 − f10 is better than that curves of different functions on different dimensions. Dif-
of the other selected algorithms. The BMA achieved the best ferent types of functions can comprehensively evaluate the
average result on function f4 . CLPSO and the DSA achieved performance of the algorithm. From Fig. 11, we can intu-
the best average result on function f6 , and the DSA and ASO itively see that the proposed CALFASO algorithm has good
achieved the best average result on function f9 . For hybrid convergence performance and obtains satisfactory results in
functions f11 − f20 , the proposed CALFASO algorithm both convergence rate and accuracy. On 30 dimensions, the
ranks first on functions f11 and f17 . The BMA achieved the proposed CALFASO performs a better convergence than the
best average result on functions f12 − f15 and f18 − f19 . others on functions f5 , f8 , f10 , f21 , f24 . On 50 dimensions,
CLPSO achieved the best average result on functions f16 the proposed CALFASO performs a better convergence than
and f20 . The proposed CALFASO algorithm has certain the others on functions f5 , f20 , f24 , f26 . For function f1 ,
advantages in solving simple multimodal functions, and the the GSA is superior the others. On 100 dimensions, the
BMA has certain advantages in solving hybrid functions. For proposed CALFASO performs a better convergence than the
VOLUME 4, 2016 9

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

FIGURE 7. Qualitative results of function f1 for four qualitative indicators: the search process, trajectory curve, convergence curve, and trajectory curve of acceleration.

FIGURE 8. Qualitative results of function f9 for four qualitative indicators: the search process, trajectory curve, convergence curve, and trajectory curve of acceleration.

others on functions f3 , f8 , f10 , f13 , f21 . Figs. 11 show to its competitors on the benchmark functions, including
that from the perspective of the convergence performance, unimodal, multimodal, hybrid and composition functions.
the proposed CALFASO is highly competitive compared

10 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

FIGURE 9. Qualitative results of function f22 for four qualitative indicators: search process, trajectory curve, convergence curve, and trajectory curve of acceleration.

FIGURE 10. Qualitative results of function f27 for four qualitative indicators: search process, trajectory curve, convergence curve, and trajectory curve of acceleration.

D. COMPARISON OF STATISTICAL RESULTS GIVEN BY tical results given by different algorithms for the gear train
DIFFERENT ENGINEERING PROBLEMS design—except for the OSA and BOA, all the algorithms
In this section, the results of two engineering problems are can obtain the target value results. As seen from Table 12—
discussed. As shown in Table 10—the comparison of statis-
VOLUME 4, 2016 11

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

(a) f5 -30D (b) f8 -30D (c) f10 -30D

(d) f21 -30D (e) f24 -30D (f) f1 -50D

(g) f5 -50D (h) f20 -50D (i) f24 -50D

(j) f26 -50D (k) f3 -100D (l) f8 -100D

(m) f10 -100D (n) f13 -100D (o) f21 -100D

(p) legend for algorithms

12 VOLUME 4, 2016
FIGURE 11. Comparison of convergence curves of different functions on different dimensions.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

a comparison of the statistical results given by different al- respectively. Table 14 shows that in the Friedman test, the
gorithms for the tension/compression spring design—the VS proposed CALFASO algorithm achieves the best results in
algorithm obtains the best average result. Among the results the different dimensions and has obvious advantages over the
of the optimal values, the proposed CALFASO algorithm selected algorithms.
is closest to the target value. In summary, the proposed
CALFASO algorithm is still competitive on unconstrained VI. CONCLUSIONS AND PROSPECTS
and constrained engineering problems. In this paper, the CA structure, Lévy flight and adaptive
weight are used to help ASO balance exploration and ex-
E. STATISTICAL ANALYSIS ploitation; the new method is named CALFASO. To verify
Statistical analysis is a scientific, accurate, objective evalua- the performance of the proposed CALFASO algorithm, the
tion method, and it is of great significance for the methodical CEC2017 test set and two real-world engineering problems
verification of experimental results [55]. This paper uses non- were used for testing. To analyze the experimental results
parametric statistical tests to analyze the significant differ- for all aspects, quantitative and qualitative analyses were
ences between the proposed CALFASO algorithm and other performed, and statistical analysis further verified the sta-
selected algorithms. For pairwise comparison of the problem- bility and accuracy of the algorithm. In summary, the per-
solving success of evolutionary algorithms, a problem-based formance of the proposed CALFASO algorithm is superior
or multi-problem-based statistical comparison method can be to that of other selected algorithms in this paper. The CA
used [56], [57]. strategy provides more information exchange for the atoms
In this paper, the Wilcoxon signed rank test (WSRT) was and enriches the diversity of the population. The LF strategy
used to perform a pairwise comparison test, where the sig- effectively helps the population to explore and avoid falling
nificance level was α = 0.05. WSRT is a nonparametric test into local optima. Adaptive weight effectively improves the
that can be used to test which algorithm has better statistical convergence performance of the algorithm. The conclusion
performance when multiple algorithms are compared on dif- shows that the three strategies have a good effect on ASO
ferent benchmark functions. ‘+’ represents that the proposed performance.
CALFASO outperforms the comparison algorithm at a 95% It is worth noting that the strategy proposed in this paper
significance level (α = 0.05). ‘=’ indicates that there is no still has some possibility for improvement, such as devel-
significant difference between the proposed CALFASO algo- oping a new CA structure, as the inherent CA structure
rithm and the comparison algorithm. ‘−’ indicates that the limits the population’s communication opportunities. At the
proposed CALFASO algorithm shows poorer performance same time, the LF trajectory of each individual may be
than the comparison algorithm. The last line gives a summary more conducive to the search of the population. At the
of the three comparisons (+/ = /−). In this paper, the same time, nonlinear weights may have advantages over
average result of each algorithm running independently 30 linear weights. The proposed CALFASO algorithm, with
times is used for the WRST. To comprehensively analyze the the help of CA, an LF, and adaptive weights, solves the
results of the statistical analysis, the WRST was performed single-objective optimization problem of CEC2017 well and
on 3 dimensions (30, 50, 100). As shown in Table 9, 11 and has good performance in engineering cases. However, at the
13, the performance of the proposed CALFASO algorithm same time, there are more thought-provoking observations.
in the pairwise comparison is significantly better than that The concept of the neighbor structure provided by cellular
of the other selected algorithms. The proposed CALFASO automata is applied in the ASO algorithm. This approach
performs well compared to the classical algorithm, emerging only helps to increase information exchange and give atoms
algorithm and excellent algorithm variants. It is particularly a better movement trend. Analysis of the chaotic behavior of
worth noting that the three proposed strategies have greatly cellular automata may be more important. At the same time,
improved the performance of the original ASO algorithm. In the parameter settings in a LF are analyzed manually. There
the comparison of the 30, 50, and 100 dimensions, the WRST may be a large number of parameters for different strategies,
results of the proposed CALFASO and the original ASO and automatic parameter determination is more promising
algorithm are 14/10/5, 18/8/3, and 21/7/1, respectively. than manual determination. From the experimental results
It is clear from the results that the performance of the and analysis, CA and the LF have made great contributions to
proposed CALFASO is even better in higher dimensions. improve the performance of the algorithm, and their ease of
As the number of dimensions increases, the advantages be- implementation is their advantage. All methods boil down to
come more apparent. Table 14 shows the mean rank of all the ingenuity of mathematics, making their analysis and inte-
algorithms in this paper according to the Friedman test [58] gration attractive. At the same time, real improvement should
on all test functions, and the highest rank is shown in bold. not be only measured by test functions and engineering cases.
Additionally, the corresponding p-values are listed in the last After improving the performance of the algorithm, it is worth
row. The Friedman statistic is distributed according to chi- exploring its potential to solve complex real-world problems.
square with 7 degrees of freedom, and the p-values of the dif- In the future, our work will start from the following prin-
ferent dimensions calculated by the Friedman test are 7.68E- ciples: (1) Research on the ASO algorithm can be extended
11, 7.38E-11, and 4.23E-11 for 30, 50, and 100 dimensions, to multi-objective optimization problems. (2) The ASO al-
VOLUME 4, 2016 13

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

gorithm should be applied to neural network optimization [20] W. Zhao, L. Wang, and Z. Zhang, “Atom search optimization and its
problems, such as BP and RBF neural networks. (3) More im- application to solve a hydrogeologic parameter estimation problem,”
Knowledge-Based Systems, vol. 163, pp. 283–304, 2019.
provements can be made to the CA structure and LFs. (4) Dif- [21] M. A. Almagboul, F. Shu, Y. Qian, X. Zhou, J. Wang, and J. Hu, “Atom
ferent versions of ASO applications should be developed for search optimization algorithm based hybrid antenna array receive beam-
real-world problems. (5) Stability and convergence analysis forming to control sidelobe level and steering the null,” AEU-International
Journal of Electronics and Communications, vol. 111, p. 152854, 2019.
are difficult points of evolutionary algorithms. Convergence [22] M. A. Elaziz, N. Nabil, A. A. Ewees, and S. Lu, “Automatic data clustering
analysis of ASO algorithm will be studied in the future. based on hybrid atom search optimization and sine-cosine algorithm,” in
2019 IEEE Congress on Evolutionary Computation (CEC). IEEE, 2019,
pp. 2315–2322.
ACKNOWLEDGMENT [23] A. M. Agwa, A. A. El-Fergany, and G. M. Sarhan, “Steady-state modeling
The authors thank the anonymous reviewers for providing of fuel cells based on atom search optimizer,” Energies, vol. 12, no. 10, p.
valuable comments to improve this paper. 1884, 2019.
[24] B. Hekimoğlu, “Optimal tuning of fractional order pid controller for dc
motor speed control via chaotic atom search optimization algorithm,”
REFERENCES IEEE Access, vol. 7, pp. 38 100–38 114, 2019.
[25] H. Haklı and H. Uğuz, “A novel particle swarm optimization algorithm
[1] X. Yang, “Introduction to mathematical optimization,” From Linear Pro-
with levy flight,” Applied Soft Computing, vol. 23, pp. 333–345, 2014.
gramming to Metaheuristics, 2008.
[26] J.-H. Lin, C.-W. Chou, C.-H. Yang, H.-L. Tsai et al., “A chaotic levy flight
[2] Y. Zhou, J. Wang, Y. Zhou, Z. Qiu, Z. Bi, and Y. Cai, “Differential
bat algorithm for parameter estimation in nonlinear dynamic biological
evolution with guiding archive for global numerical optimization,” Applied
systems,” Computer and Information Technology, vol. 2, no. 2, pp. 56–63,
Soft Computing, vol. 43, pp. 424–440, 2016.
2012.
[3] A. A. Heidari, R. A. Abbaspour, and A. R. Jordehi, “An efficient chaotic
[27] A. F. Kamaruzaman, A. M. Zain, S. M. Yusuf, and A. Udin, “Levy
water cycle algorithm for optimization tasks,” Neural Computing and
flight algorithm for optimization problems-a literature review,” in Applied
Applications, vol. 28, no. 1, pp. 57–85, 2017.
Mechanics and Materials, vol. 421. Trans Tech Publ, 2013, pp. 496–501.
[4] ——, “Gaussian bare-bones water cycle algorithm for optimal reactive
[28] M. M. Zaw and E. E. Mon, “Web document clustering using cuckoo
power dispatch in electrical power systems,” Applied Soft Computing,
search clustering algorithm based on levy flight,” International Journal of
vol. 57, pp. 657–671, 2017.
Innovation and Applied Studies, vol. 4, no. 1, pp. 182–188, 2013.
[5] S. Kirkpatrick, C. D. Gelatt, and M. P. Vecchi, “Optimization by simulated
annealing,” science, vol. 220, no. 4598, pp. 671–680, 1983. [29] R. Jensi and G. W. Jiji, “An enhanced particle swarm optimization with
levy flight for global optimization,” Applied Soft Computing, vol. 43, pp.
[6] R. Eberhart and J. Kennedy, “A new optimizer using particle swarm
248–261, 2016.
theory,” in MHS’95. Proceedings of the Sixth International Symposium
on Micro Machine and Human Science. IEEE, 1995, pp. 39–43. [30] P. J. B. N.H.Awad, M.Z.Ali, “Problem definitions and evaluation criteria
for the cec 2017 special session and competition on single objective real-
[7] A. R. Jordehi, “Binary particle swarm optimisation with quadratic transfer
parameter numerical optimization,” International Journal of Intelligent
function: A new binary optimisation algorithm for optimal scheduling of
Computing and Cybernetics, 2016.
appliances in smart homes,” Applied Soft Computing, vol. 78, pp. 465–
480, 2019. [31] R. Salgotra, U. Singh, and S. Saha, “Improved cuckoo search with better
[8] B. Jana, S. Mitra, and S. Acharyya, “Repository and mutation based search capabilities for solving cec2017 benchmark problems,” in 2018
particle swarm optimization (rmpso): A new pso variant applied to recon- IEEE Congress on Evolutionary Computation (CEC). IEEE, 2018, pp.
struction of gene regulatory network,” Applied Soft Computing, vol. 74, 1–7.
pp. 330–355, 2019. [32] R. Salgotra, U. Singh, S. Saha, and A. Nagar, “New improved salshade-
[9] R. Storn and K. Price, “Differential evolution-a simple and efficient cnepsin algorithm with adaptive parameters,” in 2019 IEEE Congress on
heuristic for global optimization over continuous spaces,” Journal of global Evolutionary Computation (CEC). IEEE, 2019, pp. 3150–3156.
optimization, vol. 11, no. 4, pp. 341–359, 1997. [33] P. A. Davidson, “An introduction to magnetohydrodynamics,” 2002.
[10] J. H. Holland, “Adaptation in natural and artificial systems. 1975,” Ann [34] J. Conway, “The game of life,” Scientific American, vol. 223, no. 4, p. 4,
Arbor, MI: University of Michigan Press and, 1992. 1970.
[11] S. Mirjalili, S. M. Mirjalili, and A. Lewis, “Grey wolf optimizer,” Ad- [35] Y. Shi, H. Liu, L. Gao, and G. Zhang, “Cellular particle swarm optimiza-
vances in engineering software, vol. 69, pp. 46–61, 2014. tion,” Information Sciences, vol. 181, no. 20, pp. 4460–4493, 2011.
[12] E. Rashedi, H. Nezamabadi-Pour, and S. Saryazdi, “Gsa: a gravitational [36] J. Yi, L. Gao, X. Li, and J. Gao, “An efficient modified harmony search
search algorithm,” Information sciences, vol. 179, no. 13, pp. 2232–2248, algorithm with intersect mutation operator and cellular local search for
2009. continuous function optimization problems,” Applied Intelligence, vol. 44,
[13] R. Salgotra and U. Singh, “The naked mole-rat algorithm,” Neural Com- no. 3, pp. 725–753, 2016.
puting and Applications, vol. 31, no. 12, pp. 8837–8857, 2019. [37] E. Alba and B. Dorronsoro, “The exploration/exploitation tradeoff in
[14] S. Khalilpourazari and S. Khalilpourazary, “Optimization of time, cost dynamic cellular genetic algorithms,” IEEE transactions on evolutionary
and surface roughness in grinding process using a robust multi-objective computation, vol. 9, no. 2, pp. 126–142, 2005.
dragonfly algorithm,” Neural Computing and Applications, pp. 1–12, [38] D. W. Sims, E. J. Southall, N. E. Humphries, G. C. Hays, C. J. Bradshaw,
2018. J. W. Pitchford, A. James, M. Z. Ahmed, A. S. Brierley, M. A. Hindell
[15] I. Aljarah, H. Faris, and S. Mirjalili, “Optimizing connection weights in et al., “Scaling laws of marine predator search behaviour,” Nature, vol.
neural networks using the whale optimization algorithm,” Soft Computing, 451, no. 7182, p. 1098, 2008.
vol. 22, no. 1, pp. 1–15, 2018. [39] X.-S. Yang and S. Deb, “Cuckoo search via lévy flights,” in 2009 World
[16] R. Salgotra, U. Singh, and S. Saha, “New cuckoo search algorithms with Congress on Nature & Biologically Inspired Computing (NaBIC). IEEE,
enhanced exploration and exploitation properties,” Expert Systems with 2009, pp. 210–214.
Applications, vol. 95, pp. 384–420, 2018. [40] A. K. Bhateja, A. Bhateja, S. Chaudhury, and P. Saxena, “Cryptanalysis of
[17] M. A. El Aziz and A. E. Hassanien, “Modified cuckoo search algorithm vigenere cipher using cuckoo search,” Applied Soft Computing, vol. 26,
with rough sets for feature selection,” Neural Computing and Applications, pp. 315–324, 2015.
vol. 29, no. 4, pp. 925–934, 2018. [41] G. Kalantzis, C. Shang, Y. Lei, and T. Leventouri, “Investigations of a gpu-
[18] K. Kaur, U. Singh, and R. Salgotra, “An enhanced moth flame optimiza- based levy-firefly algorithm for constrained optimization of radiation ther-
tion,” Neural Computing and Applications, pp. 1–35, 2018. apy treatment planning,” Swarm and Evolutionary Computation, vol. 26,
[19] A. K. Qin, V. L. Huang, and P. N. Suganthan, “Differential evolution pp. 191–201, 2016.
algorithm with strategy adaptation for global numerical optimization,” [42] W. A. Hussein, S. Sahran, and S. N. H. S. Abdullah, “Patch-levy-based
IEEE transactions on Evolutionary Computation, vol. 13, no. 2, pp. 398– initialization algorithm for bees algorithm,” Applied Soft Computing,
417, 2008. vol. 23, pp. 104–121, 2014.

14 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

[43] N. Lynn and P. N. Suganthan, “Heterogeneous comprehensive learning FIRST A. AUTHOR was born in Baoding, Hebei
particle swarm optimization with enhanced exploration and exploitation,” Province, China, in 1995. He received a B.E.
Swarm and Evolutionary Computation, vol. 24, pp. 11–24, 2015. degree in electrical engineering and automation
[44] S. Wolfram, A new kind of science. Wolfram media Champaign, IL, from Yan Shan University in 2018. He is currently
2002, vol. 5. pursuing an M.S. degree in control science and
[45] ——, Cellular automata and complexity: collected papers. CRC Press, engineering from the School of Electronic and
2018. Information Engineering, University of Science
[46] M. Guidolin, A. S. Chen, B. Ghimire, E. C. Keedwell, S. Djordjević, and
and Technology Liaoning, Anshan, China.
D. A. Savić, “A weighted cellular automata 2d inundation model for rapid
flood analysis,” Environmental Modelling & Software, vol. 84, pp. 378–
394, 2016.
[47] R. Alonso-Sanz, “Cellular automata with memory,” Cellular Automata: A
Volume in the Encyclopedia of Complexity and Systems Science, Second
Edition, pp. 153–183, 2018.
[48] C. Lu, L. Gao, and J. Yi, “Grey wolf optimizer with cellular topological
structure,” Expert Systems with Applications, vol. 107, pp. 89–114, 2018.
[49] G. M. Viswanathan, V. Afanasyev, S. Buldyrev, E. Murphy, P. Prince,
and H. E. Stanley, “Lévy flight search patterns of wandering albatrosses,”
Nature, vol. 381, no. 6581, p. 413, 1996.
[50] X.-S. Yang and S. Deb, “Multiobjective cuckoo search for design opti- SECOND B. AUTHOR was born in Anshan,
mization,” Computers & Operations Research, vol. 40, no. 6, pp. 1616– Liaoning, China. He received B.S. and M.S. de-
1624, 2013.
grees in control science and engineering from the
[51] C.-Y. Lee and X. Yao, “Evolutionary algorithms with adaptive lévy muta-
University of Science and Technology Liaoning,
tions,” in Proceedings of the 2001 Congress on Evolutionary Computation
(IEEE Cat. No. 01TH8546), vol. 1. IEEE, 2001, pp. 568–575. Anshan, in 1987 and 1990, respectively, and a
[52] R. N. Mantegna, “Fast, accurate algorithm for numerical simulation of levy Ph.D. in control science from Dalian University
stable stochastic processes,” Physical Review E, vol. 49, no. 5, p. 4677, of Technology, China, in 2005. He is currently a
1994. Professor and a graduate student supervisor in the
[53] R. Paulavičius, Y. D. Sergeyev, D. E. Kvasov, and J. Žilinskas, “Globally- School of Electronic and Information Engineering,
biased disimpl algorithm for expensive global optimization,” Journal of University of Science and Technology Liaoning,
Global Optimization, vol. 59, no. 2-3, pp. 545–567, 2014. Anshan, China. His main research area of interest is the modeling of complex
[54] E. Sandgren, “Nonlinear integer and discrete programming in mechanical industry processes and intelligent control.
design optimization,” Journal of Mechanical Design, vol. 112, no. 2, pp.
223–229, 1990.
[55] S. García, A. Fernández, J. Luengo, and F. Herrera, “Advanced non-
parametric tests for multiple comparisons in the design of experiments
in computational intelligence and data mining: Experimental analysis of
power,” Information Sciences, vol. 180, no. 10, pp. 2044–2064, 2010.
[56] J. Derrac, S. García, D. Molina, and F. Herrera, “A practical tutorial on
the use of nonparametric statistical tests as a methodology for comparing
evolutionary and swarm intelligence algorithms,” Swarm and Evolutionary
Computation, vol. 1, no. 1, pp. 3–18, 2011. THIRD C. AUTHOR was born in Chengde, Hebei
[57] P. Civicioglu, “Backtracking search optimization algorithm for numerical Province, China, in 1996. He received a B.E.
optimization problems,” Applied Mathematics and Computation, vol. 219, degree in automation from Liaoning University
no. 15, pp. 8121–8144, 2013. of Science and Technology in 2018. He is cur-
[58] H. Liu, G. Ding, and B. Wang, “Bare-bones particle swarm optimization rently pursuing an M.S. degree in control science
with disruption operator,” Applied Mathematics and Computation, vol. and engineering at the School of Electronics and
238, pp. 106–122, 2014. Information Engineering, Liaoning University of
[59] M. Zambrano-Bigiarini, M. Clerc, and R. Rojas, “Standard particle swarm Science and Technology Liaoning, Anshan, China.
optimisation 2011 at cec-2013: A baseline for future pso improvements,”
in 2013 IEEE Congress on Evolutionary Computation. IEEE, 2013, pp.
2337–2344.
[60] J. J. Liang, A. K. Qin, P. N. Suganthan, and S. Baskar, “Comprehensive
learning particle swarm optimizer for global optimization of multimodal
functions,” IEEE transactions on evolutionary computation, vol. 10, no. 3,
pp. 281–295, 2006.
[61] B. Doğan and T. Ölmez, “A new metaheuristic for numerical function
optimization: Vortex search algorithm,” Information Sciences, vol. 293,
pp. 125–145, 2015.
[62] P. Civicioglu, “Transforming geocentric cartesian coordinates to geodetic FOURTH D. AUTHOR was born in Zibo, Shang-
coordinates by using differential search algorithm,” Computers & Geo- dong Province, China, in 1995. He received a B.E.
sciences, vol. 46, pp. 229–247, 2012.
degree in automation engineering from the Uni-
[63] Q. Zhang, R. Wang, J. Yang, A. Lewis, F. Chiclana, and S. Yang, “Biology
versity of Jinan Quancheng College in 2019. He is
migration algorithm: a new nature-inspired heuristic methodology for
global optimization,” Soft Computing, pp. 1–26, 2018. currently pursuing an M.S. degree in control sci-
[64] M. Jain, S. Maurya, A. Rani, and V. Singh, “Owl search algorithm: a ence and engineering at the School of Electronics
novel nature-inspired heuristic paradigm for global optimization,” Journal and Information Engineering, Liaoning University
of Intelligent & Fuzzy Systems, vol. 34, no. 3, pp. 1573–1582, 2018. of Science and Technology Liaoning, Anshan,
[65] S. Arora and S. Singh, “Butterfly optimization algorithm: a novel approach China.
for global optimization,” Soft Computing, vol. 23, no. 3, pp. 715–734,
2019.

VOLUME 4, 2016 15

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 1. CEC2017 benchmark functions.

No. Function Formulation Range fopt

Shifted and Rotated
D
f1 (x) = x21 + 106 x2i
P
f1 Bent Cigar Function [-100,100] 100
i=2
D D D
x2i + ( 0.5xi )2 + ( 0.5xi )4
P P P
f3 Zakharov Function f3 (x) = [-100,100] 300
i=1 i=1 i=1
D−1
2
(100(x2i − xi+1 ) + (xi − 1)2 )
P
f4 Rosenbrock’s Function f4 (x) = [-100,100] 400
i=1
D
(x2i − 10 cos(2πxi ) + 10)
P
f5 Rastrigin’s Function f5 (x) = [-100,100] 500
i=1 √
(sin2 ( x2 +y 2 )−0.5)
f6 Expanded Schaffer’s F6 Function g(x, y) = 0.5 + 2 2 2 [-100,100] 600
(1+0.001(x +y ))
f6 (x) = g(x1 , x2 ) + g(x2 , x3 ) + ... + g(xD−1 , xD ) + g(xD , x1 )
D D D
(xi − µ0 )2 , dD + s (xi − µ1 )2 ) + 10(D −
P P P
f7 (x) = min( cos(2πzi ))
i=1 r i=1 i=1

f7 Lunacek Bi-Rastrigin Function µ2

0 −d , s 1 [-100,100] 700
µ0 = 2.5, µ1 = − s= 1 − 2√D+20−8.2 ,d = 1
10(x−o)
y= 100 , xi = 2sign(xi ∗)yi + µ0 , f or i = 1, 2, ..., D
z = ∧100 (x − µ0 )
D
(zi2 − 10cos(2πzi ) + 10) + f13 ∗
P
f8 (x) =
i=1 
5.12(x−o) xi , if |xi | ≤ 0.5
x = M1 , yi = f or i = 1, 2,..., D
100 round(2xi )/2, if |xi | > 0.5
z = M1 ∧10 M2 Tasy
0.2
(Tosz (y))
i−1
α
f8 Noncontinuous Rastrigin’s Function where∧ : λii = α 2(D−1) , i = 1, 2,..., D [-100,100] 800
√
β 1+β i−1 xi
Tasy : if, xi > 0, xi = xi D−1 i = 1, 2,...,D [4]
, f or
Tosz = f or, xi = sign(xi )exp(xi + 0.049(sin(c1 xi )(+ sin(c2 xi ))), f or i = 1andD [4]
−1, if xi < 0
log(|xi |), if, xi 6= 0 , sign(x ) =
n
where xi = i 0, if xi = 0
0, otherwise 1, otherwise
10, if xi > 0 7.9, if xi > 0
n n
c1 = 5.5, otherwise , andc 2 = 3.1, otherwise
D−1
f9 (x) = sin2 (πw1 ) + (wi − 1)2 [1 + 10sin2 (πwi + 1)] + (wD − 1)2 [1 + sin2 (2πwd )]
P
f9 Levy Function i=1 [-100,100] 900
xi −1
W herewi =1+ 4 , ∀i= 1, ..., D
D
P
f10 (x) = 418.9829 × D − g(zi ), zi = xi + 4.209687462275036e + 002
i=1
1/2

f10 Schwefel’s Function  zi sin(|zi |
 ), if |zi | < 500 [-100,100] 1000
p (zi −500)2
g(zi ) = (500 − mod(zi , 500))sin( |500 − mod(zi − 500)|) − 10000D , if zi > 500
(zi +500)2

 p
(mod(|zi | , 500) − 500) sin( |mod(zi − 500)| − 500) − 10000D , if zi < −500
f11 Hybrid Function 1 (N=3) [-100,100] 1100
f12 Hybrid Function 2 (N=3) [-100,100] 1200
f13 Hybrid Function 3 (N=3) [-100,100] 1300
f14 Hybrid Function 4 (N=4) [-100,100] 1400
f15 Hybrid Function 5 (N=4) [-100,100] 1500
f16 Hybrid Function 6 (N=4) [-100,100] 1600
f17 Hybrid Function 6 (N=5) [-100,100] 1700
f18 Hybrid Function 6 (N=5) [-100,100] 1800
f19 Hybrid Function 6 (N=5) [-100,100] 1900
f20 Hybrid Function 6 (N=6) [-100,100] 2000
f21 Composition Function 1 (N=3) [-100,100] 2100
f22 Composition Function 2 (N=3) [-100,100] 2200
f23 Composition Function 3 (N=4) [-100,100] 2300
f24 Composition Function 4 (N=4) [-100,100] 2400
f25 Composition Function 5 (N=5) [-100,100] 2500
f26 Composition Function 6 (N=5) [-100,100] 2600
f27 Composition Function 7 (N=6) [-100,100] 2700
f28 Composition Function 8 (N=6) [-100,100] 2800
f29 Composition Function 9 (N=3) [-100,100] 2900
f30 Composition Function 10 (N=3) [-100,100] 3000

TABLE 2. Algorithms used in the comparisons.

Algorithms Reference Year Parameter settings

SPSO2011 [59] 2011 w = 1/(2 ∗ log(2)), c1 = c2 = 0.5 + log(2)
t
CLPSO [60] 2006 learning probability P c = 0.05 + 0.45 ∗ ee10−1
−1
VS [61] 2015 x = 0.1 for gammaincinv function
DSA [62] 2012 p1 = p2 = 0.3 ∗ rand
GSA [12] 2009 Rnorm = 2
BMA [63] 2018 Switch probability P r = 0.2, maximum number of cycles=80
OSA [64] 2018 βmax = 1.9, βmin = 0, β = βmax − t ∗ (βmax − βmin )/T
BOA [65] 2019 Sensor Modalityc = 0.01 Switch Probability =0.8
ASO [20] 2019 Depth weight=50 Multiplier weight=0.2
CALFASO present present L5 structure of CA, Lévy Flight β = 1.4, w = 0.9 − 0.5 ∗ t/T , Depth weight=50 Multiplier weight=0.2

16 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 3. Sensitivity test for the CA.

L5 L9 C9 C13 C21 C25

f1 Mean 1.8045E+03 2.1702E+03 2.6350E+03 2.3068E+03 2.6178E+03 2.8650E+03
Std 1.8681E+03 3.3939E+03 2.8861E+03 2.9229E+03 3.1947E+03 2.9192E+03
Best 1.1083E+02 1.0011E+02 1.0048E+02 1.0002E+02 1.0263E+02 1.0021E+02
Rank 1 2 5 3 4 6
f3 Mean 1.5442E+04 8.7694E+03 8.8957E+03 6.6406E+03 3.6390E+03 3.4704E+03
Std 5.0751E+03 4.1085E+03 4.5106E+03 2.6432E+03 2.2014E+03 2.0832E+03
Best 8.1098E+03 3.4186E+03 2.4188E+03 2.1695E+03 9.9031E+02 6.1652E+02
Rank 6 4 5 3 2 1
f4 Mean 5.0200E+02 5.0209E+02 4.9754E+02 4.9864E+02 5.1137E+02 5.0635E+02
Std 2.0996E+01 3.0764E+01 2.7840E+01 3.2031E+01 2.9157E+01 2.9430E+01
Best 4.6401E+02 4.0428E+02 4.0028E+02 4.0003E+02 4.4124E+02 4.0941E+02
Rank 3 4 1 2 6 5
f5 Mean 5.3642E+02 5.4607E+02 5.4443E+02 5.4936E+02 5.5289E+02 5.6465E+02
Std 1.0708E+01 1.0964E+01 1.0887E+01 1.3172E+01 1.3998E+01 1.4959E+01
Best 5.2090E+02 5.2786E+02 5.2587E+02 5.2189E+02 5.2985E+02 5.3681E+02
Rank 1 3 2 4 5 6
f6 Mean 6.0004E+02 6.0000E+02 6.0001E+02 6.0001E+02 6.0000E+02 6.0001E+02
Std 4.5581E-02 8.8060E-03 1.1051E-02 1.6489E-02 1.0900E-02 1.7915E-02
Best 6.0001E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02
Rank 6 1 5 3 1 4
f7 Mean 7.7193E+02 7.8190E+02 7.7959E+02 7.8611E+02 8.0098E+02 8.0531E+02
Std 1.1031E+01 1.2399E+01 1.0664E+01 1.3798E+01 1.9983E+01 1.8711E+01
Best 7.5515E+02 7.6118E+02 7.5422E+02 7.5952E+02 7.6975E+02 7.7698E+02
Rank 1 3 2 4 5 6
f8 Mean 8.3735E+02 8.4406E+02 8.3882E+02 8.4907E+02 8.5815E+02 8.5761E+02
Std 8.4166E+00 1.1776E+01 1.0241E+01 1.4368E+01 1.6911E+01 1.5680E+01
Best 8.2487E+02 8.2487E+02 8.1492E+02 8.2288E+02 8.2885E+02 8.3383E+02
Rank 1 3 2 4 6 5
f9 Mean 9.1800E+02 9.3169E+02 9.2824E+02 9.5432E+02 1.0239E+03 1.0239E+03
Std 1.2650E+01 2.0152E+01 2.7443E+01 5.0439E+01 1.1842E+02 1.0608E+02
Best 9.0209E+02 9.0209E+02 9.0381E+02 9.0900E+02 9.1172E+02 9.0999E+02
Rank 1 3 2 4 5 6
f10 Mean 3.6568E+03 3.6308E+03 3.7430E+03 3.7624E+03 3.7836E+03 3.8444E+03
Std 4.9758E+02 5.0833E+02 5.2278E+02 5.2746E+02 5.4675E+02 4.5790E+02
Best 2.8230E+03 2.6340E+03 2.2941E+03 2.5175E+03 2.8441E+03 2.9259E+03
Rank 2 1 3 4 5 6
f11 Mean 1.1550E+03 1.1542E+03 1.1614E+03 1.1528E+03 1.1544E+03 1.1632E+03
Std 3.2981E+01 2.9545E+01 3.3344E+01 2.7849E+01 2.8718E+01 3.3218E+01
Best 1.1166E+03 1.1095E+03 1.1156E+03 1.1177E+03 1.1149E+03 1.1171E+03
Rank 4 2 5 1 3 6
f12 Mean 507800.9651 439583.787 547259.3561 456739.9835 458063.8399 528070.1545
Std 278301.4353 321786.4734 837432.8995 413766.8082 396815.3743 612898.7694
Best 34539.97716 57300.98855 39747.92694 32904.03944 24532.95626 42353.48471
Rank 4 1 6 2 3 5
f13 Mean 8596.905714 7817.73523 8615.519574 8466.668317 12725.88569 10791.05112
Std 4308.237099 5055.245877 5388.646688 6011.680745 12200.40192 10217.03504
Best 1832.893405 1499.27019 1382.731463 1459.578071 1420.582248 1361.27947
Rank 3 1 4 2 6 5
f14 Mean 13313.20298 17104.24975 13913.06968 18436.15562 18763.46108 21305.76192
Std 20498.94519 33164.77827 15942.32314 22745.84153 18290.83269 27721.14405
Best 1850.544817 1827.08797 2134.646828 2810.411117 1561.250169 1616.325975
Rank 1 3 2 4 5 6
f15 Mean 4492.845543 5776.002201 4247.410513 4876.190481 7074.451798 8075.025878
Std 4692.36356 4629.210259 2897.399005 4031.988651 5784.247314 6718.974978
Best 1527.245301 1531.782623 1524.872825 1522.225618 1545.828869 1615.464302
Rank 2 4 1 3 5 6
f16 Mean 2342.538747 2350.904858 2364.399187 2367.79119 2427.553076 2474.204022
Std 255.9592393 258.7827726 269.8280859 294.5859936 254.3404457 296.0786687
Best 1821.416968 1955.181881 1737.869432 1733.416301 1737.580393 2034.843898
Rank 1 2 3 4 5 6
f17 Mean 1894.507767 1924.808964 1863.888764 1914.869273 2019.63447 1977.96096
Std 151.9528658 166.6363252 118.0628418 140.6206422 183.1323558 189.7261226
Best 1722.324484 1721.306429 1733.379673 1726.908286 1725.36293 1735.818087
Rank 2 4 1 3 6 5
f18 Mean 316583.7963 269686.9079 229233.2469 301953.7855 306757.4534 322764.0306
Std 276928.6177 205064.4112 151159.5531 266830.657 186965.0652 243668.0018
Best 49978.65104 27346.70292 34222.07447 45226.36202 64050.76396 32485.77925
Rank 5 2 1 3 4 6
f19 Mean 6010.958999 6513.84547 6994.093547 9534.295791 8223.975644 10533.62211
Std 4500.779725 5845.855347 6085.586706 8057.869195 6881.55073 9804.568663
Best 1928.904518 1927.783958 1920.420188 1960.624478 1913.14575 1945.773106
Rank 1 2 3 5 4 6
f20 Mean 2230.325162 2300.401431 2231.790085 2270.170983 2329.838786 2303.09082
Std 128.5432782 164.8078599 115.3881978 143.2243767 159.1560608 146.3143231
Best 2018.365441 2015.572017 2034.633155 2029.804858 2026.312868 2028.914196
Rank 1 4 2 3 6 5
f21 Mean 2336.554358 2341.057255 2343.725017 2344.585488 2358.879338 2356.90952
Std 8.686371499 8.528060738 12.77583505 12.9143033 14.43734918 14.50910262
Best 2321.67469 2323.227888 2320.957073 2318.904935 2334.899196 2329.139787
Rank 1 2 3 4 6 5
f22 Mean 2300.364369 2300.831872 2300.264931 2300.530099 2300.709862 2300.768673
Std 0.9974625 1.386679388 0.810389549 1.278017653 1.312238001 1.424090295
Best 2300.002097 2300.000208 2300.000148 2300.000052 2300.00001 2300.000002
Rank 2 6 1 3 4 5
f23 Mean 2699.043096 2704.185851 2701.54882 2712.498762 2722.097003 2726.150797
Std 15.14845967 13.57421999 16.6264738 17.77279906 22.67896385 20.83929968
Best 2666.944503 2669.873939 2667.784864 2674.687958 2680.41323 2685.383453
Rank 1 3 2 4 5 6
f24 Mean 2864.052434 2874.50953 2873.901363 2884.492696 2893.819381 2895.77655
Std 16.34504881 15.69302476 19.05375262 15.60609078 21.74539444 23.65895506
Best 2832.961132 2845.303067 2837.529441 2850.490008 2848.550872 2857.336751
Rank 1 3 2 4 5 6
f25 Mean 2890.403782 2895.020435 2891.96296 2893.354819 2899.120417 2896.401778
Std 5.374235072 10.48516305 7.634303056 8.379958562 15.09061792 12.48956586
Best 2887.52584 2887.679478 2887.49274 2887.859812 2887.656897 2884.289011
Rank 1 4 2 3 6 5
f26 Mean 4109.866469 4220.566372 4131.767736 4350.271443 4156.398414 4329.779751
Std 315.3964993 240.0540477 403.2348674 236.2099647 668.0412164 436.4135702
Best 2800.021297 2900.003529 2800.001984 3729.524539 2800.000067 2800.000034
Rank 1 4 2 6 3 5
f27 Mean 3240.249253 3241.950533 3240.188035 3245.548791 3245.237036 3249.083487
Std 12.34384163 14.89683381 12.84474869 14.48224521 15.56165001 15.98984756
Best 3216.972215 3197.734774 3214.641721 3215.966397 3217.341277 3217.994279
Rank 2 3 1 5 4 6
f28 Mean 3211.191777 3214.275719 3215.400413 3214.614202 3214.630934 3217.251309
Std 14.74716229 16.2210124 16.75351926 17.96843486 31.25274719 38.49438113
Best 3195.115534 3194.972907 3189.629625 3180.827194 3109.466897 3100.092662
Rank 1 2 5 3 4 6
f29 Mean 3491.133702 3539.322827 3536.832248 3535.031562 3592.525083 3650.919923
Std 132.2396305 154.2917926 153.3679117 145.4431429 130.9400245 168.389412
Best 3345.496917 3354.919308 3344.703501 3257.291868 3364.6478 3352.963373
Rank 1 4 3 2 5 6
f30 Mean 9401.491942 11761.82477 10345.16287 10641.20578 12866.89558 16260.21962
Std 3203.900784 8893.635087 4298.68424 3787.559271 5288.129836 10724.21481
Best 6021.350063 6192.478539 6482.3162 6741.126181 6137.32726 6279.162278
Rank 1 4 2 3 5 6
Total Rank 58 84 78 98 133 157
Final Rank 1 3 2 4 5 6

VOLUME 4, 2016 17

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 4. Sensitivity test for the LF (different β values).

β 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 random[1,2]

f3 Mean 1.8840E+04 2.3356E+04 1.5322E+04 1.3611E+04 1.3583E+04 1.3394E+04 1.4375E+04 1.4282E+04 1.5385E+04 12762.48275
Std 5.7744E+03 8.2765E+03 5.7776E+03 5.9086E+03 5.0632E+03 5.3619E+03 4.5968E+03 5.6737E+03 6.0394E+03 4888.285688
Best 5.2751E+03 8.3691E+03 6.4679E+03 4.6440E+03 4.1495E+03 4.3088E+03 8.2899E+03 6.6028E+03 3.4534E+03 3975.666278
Rank 9 10 7 4 3 2 6 5 8 1
f6 Mean 6.0022E+02 6.0108E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0201E+02 600.0032452
Std 1.8175E-01 4.2339E-01 5.8801E-04 1.3032E-02 1.3296E-02 5.5240E-03 1.7372E-04 1.1356E-03 1.5883E+00 0.007038692
Best 6.0004E+02 6.0055E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0013E+02 600.0000076
Rank 8 9 1 1 1 1 1 1 10 6
f15 Mean 3875.558296 3941.069318 3316.150155 4066.567271 4178.678408 3956.232469 4719.03418 3534.176537 9891.356969 4773.331948
Std 3810.561517 2698.913491 2427.489603 2865.672898 2960.702645 2383.845859 4681.526168 2737.673167 5380.828231 3818.932927
Best 1526.865413 1531.19085 1524.175047 1524.730705 1550.525405 1514.160399 1570.75542 1519.573669 2730.87103 1515.333676
Rank 3 4 1 6 7 5 8 2 10 9
f22 Mean 2301.199791 2312.351616 2300.049067 2300.197239 2300.049067 2300.098263 2300.049067 2300.049195 2300.590984 2300.245469
Std 1.241919634 2.025597928 0.346957867 0.675612793 0.34695786 0.486273967 0.346957867 0.347863325 1.391044391 0.743868136
Best 2300.194185 2304.039758 2300 2300.000005 2300 2300 2300 2300 2300 2300
Rank 9 10 1 6 3 5 2 4 8 7
Best Number 0 0 3 1 1 1 1 1 0 1
Average Rank 7.25 8.25 2.5 4.25 3.5 3.25 4.25 3 9 5.75

18 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 5. Synergy analysis of the different strategies.

CA LF w CA-LF CA-w LF-w CA-LF-w

f1 Mean 2.3639E+03 2.4660E+04 1.7116E+03 2.5024E+03 1.3817E+03 1.7789E+03 1.6789E+03
Std 1.5746E+03 2.7685E+04 2.1540E+03 1.5448E+03 3.5724E+02 1.1124E+03 1.9920E+03
Best 1.0015E+02 1.2747E+03 1.0001E+02 2.2672E+02 3.2144E+02 1.1905E+02 1.0267E+02
Rank 5 7 3 6 1 4 2
f3 Mean 3.0000E+02 2.0789E+04 6.5651E+02 3.0008E+02 3.0000E+02 1.7331E+04 3.0001E+02
Std 3.5009E-14 4.6079E+03 6.5778E+02 2.6748E-01 1.8283E-14 4.5710E+03 1.0490E-02
Best 3.0000E+02 1.3792E+04 3.1308E+02 3.0000E+02 3.0000E+02 1.0846E+04 3.0000E+02
Rank 1 7 5 4 2 6 3
f4 Mean 5.1139E+02 5.1856E+02 4.7813E+02 4.9121E+02 5.5490E+02 5.1545E+02 4.9947E+02
Std 8.4512E+01 1.1187E+01 5.4004E+01 4.0683E+01 5.5688E+01 8.4944E+00 4.4275E+01
Best 4.0400E+02 4.8955E+02 4.0007E+02 4.0944E+02 4.0400E+02 4.7200E+02 4.0004E+02
Rank 4 6 1 2 7 5 3
f5 Mean 5.3001E+02 5.5054E+02 5.3787E+02 5.2779E+02 5.3114E+02 5.4491E+02 5.2070E+02
Std 7.6406E+00 8.3878E+00 1.0038E+01 6.5632E+00 7.1181E+00 1.2320E+01 4.2002E+00
Best 5.1691E+02 5.3749E+02 5.1492E+02 5.1691E+02 5.2189E+02 5.2579E+02 5.1293E+02
Rank 3 7 5 2 4 6 1
f6 Mean 6.0001E+02 6.0072E+02 6.0010E+02 6.0000E+02 6.0000E+02 6.0013E+02 6.0000E+02
Std 2.7665E-02 5.0400E-01 2.7626E-01 2.4705E-12 1.0222E-07 1.0262E-01 1.3518E-13
Best 6.0000E+02 6.0008E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02 6.0000E+02
Rank 4 7 5 2 3 6 1
f7 Mean 7.4308E+02 7.4711E+02 7.5247E+02 7.5137E+02 7.4020E+02 7.4338E+02 7.4671E+02
Std 3.8224E+00 4.0043E+00 8.3207E+00 5.6633E+00 2.3878E+00 2.5616E+00 3.9169E+00
Best 7.3745E+02 7.3916E+02 7.4240E+02 7.4420E+02 7.3896E+02 7.3787E+02 7.4015E+02
Rank 2 5 7 6 1 3 4
f8 Mean 8.2305E+02 8.3907E+02 8.3303E+02 8.2982E+02 8.2603E+02 8.3685E+02 8.2225E+02
Std 6.8240E+00 7.2553E+00 1.0440E+01 5.9381E+00 4.6307E+00 7.8102E+00 5.2680E+00
Best 8.1492E+02 8.2986E+02 8.1890E+02 8.1990E+02 8.1691E+02 8.3084E+02 8.1492E+02
Rank 2 7 5 4 3 6 1
f9 Mean 9.0000E+02 9.0034E+02 9.0000E+02 9.0004E+02 9.0000E+02 9.0000E+02 9.0000E+02
Std 0.0000E+00 3.0613E-01 5.5855E-14 1.6645E-01 0.0000E+00 8.4444E-14 1.6346E-02
Best 9.0000E+02 9.0006E+02 9.0000E+02 9.0000E+02 9.0000E+02 9.0000E+02 9.0000E+02
Rank 1 7 4 6 2 3 5
f10 Mean 3.9071E+03 3.6383E+03 3.6285E+03 3.1481E+03 4.1553E+03 3.2938E+03 3.2868E+03
Std 8.5314E+02 5.7521E+02 5.5718E+02 3.7579E+02 4.8841E+02 5.7712E+02 4.9312E+02
Best 2.5605E+03 2.4256E+03 2.6525E+03 2.4113E+03 3.0176E+03 2.5900E+03 2.2374E+03
Rank 6 5 4 1 7 3 2
f11 Mean 1.1650E+03 1.2102E+03 1.1696E+03 1.1169E+03 1.1609E+03 1.1537E+03 1.1378E+03
Std 3.3815E+01 3.2449E+01 3.2030E+01 1.1409E+01 2.8440E+01 2.7511E+01 2.8780E+01
Best 1.1239E+03 1.1582E+03 1.1251E+03 1.1096E+03 1.1289E+03 1.1297E+03 1.1091E+03
Rank 5 7 6 1 4 3 2
f12 Mean 2.5094E+04 8.0278E+05 8.3128E+04 8.0678E+04 9.4222E+03 3.5567E+05 40697.89338
Std 7.4181E+04 4.7952E+05 4.7854E+04 1.5773E+05 3.2038E+03 1.6735E+05 95671.74544
Best 6.3736E+03 2.5020E+05 2.1967E+04 1.8977E+04 5.4139E+03 8.2298E+04 7896.956563
Rank 2 7 5 4 1 6 3
f13 Mean 8.0573E+03 8.1092E+03 8.6147E+03 5.7343E+03 8.8162E+03 8.6110E+03 8876.89155
Std 3.9859E+03 5.4027E+03 5.1629E+03 4.4253E+03 2.9860E+03 4.0692E+03 4507.824548
Best 2.3808E+03 2.1794E+03 1.3338E+03 3.1194E+03 4.7032E+03 2.6543E+03 3304.205699
Rank 2 3 5 1 6 4 7
f14 Mean 1.8261E+03 8.3174E+04 8.8696E+03 6.7661E+03 1.8135E+03 1.8219E+04 6289.928134
Std 5.0103E+02 6.1893E+04 6.2404E+03 4.2790E+03 2.8338E+02 8.0187E+03 6209.301584
Best 1.4470E+03 8.6811E+03 1.5466E+03 1.4587E+03 1.5292E+03 3.0285E+03 1528.83058
Rank 2 7 5 4 1 6 3
f15 Mean 3.0053E+03 3.5409E+03 4.7711E+03 5.4491E+03 2.5787E+03 6.1329E+03 4383.931234
Std 1.9655E+03 2.8149E+03 3.2604E+03 3.9005E+03 2.3027E+03 3.5625E+03 2982.877421
Best 1.5044E+03 1.5233E+03 1.5143E+03 1.7098E+03 1.5077E+03 1.5266E+03 1507.745972
Rank 2 3 5 6 1 7 4
f16 Mean 2.2287E+03 2.5617E+03 2.3501E+03 2.1523E+03 2.0738E+03 2.3086E+03 2204.551831
Std 1.8187E+02 2.3901E+02 2.7010E+02 8.8969E+01 1.1601E+02 1.4364E+02 172.7773924
Best 1.9134E+03 2.0850E+03 1.8441E+03 2.0215E+03 1.9807E+03 2.0712E+03 1938.40386
Rank 4 7 6 2 1 5 3
f17 Mean 1.8518E+03 1.9152E+03 1.8833E+03 1.7693E+03 1.8477E+03 1.8599E+03 1787.117606
Std 9.7567E+01 1.4287E+02 1.2405E+02 6.3027E+01 5.7467E+01 1.0272E+02 86.546883
Best 1.7357E+03 1.7506E+03 1.7443E+03 1.7186E+03 1.7435E+03 1.7287E+03 1711.3394
Rank 4 7 6 1 3 5 2
f18 Mean 2.4652E+04 1.0960E+06 8.5197E+04 1.2612E+05 4.5718E+04 5.1376E+05 89615.33442
Std 7.4966E+03 1.3546E+06 4.1808E+04 6.4146E+04 1.6761E+04 5.0625E+05 66162.75947
Best 9.3277E+03 8.0476E+04 2.5239E+04 2.4024E+04 8.8696E+03 7.6995E+04 26147.26221
Rank 1 7 3 5 2 6 4
f19 Mean 6.3147E+03 5.8761E+03 7.0477E+03 4.5705E+03 3.5763E+03 7.1948E+03 6512.982298
Std 5.0554E+03 3.5451E+03 5.6556E+03 2.1993E+03 2.3886E+03 5.1934E+03 6425.458596
Best 2.2468E+03 1.9400E+03 1.9637E+03 1.9853E+03 1.9279E+03 2.1377E+03 1920.977052
Rank 4 3 6 2 1 7 5
f20 Mean 2.2338E+03 2.2514E+03 2.2443E+03 2.1850E+03 2.2480E+03 2.2783E+03 2196.053767
Std 1.3613E+02 1.1797E+02 9.8214E+01 8.5591E+01 9.0727E+01 1.4000E+02 113.1340032
Best 2.1471E+03 2.1546E+03 2.0821E+03 2.0172E+03 2.1467E+03 2.1541E+03 2029.195886
Rank 3 6 4 1 5 7 2
f21 Mean 2.3295E+03 2.3465E+03 2.3429E+03 2.3231E+03 2.3405E+03 2.3389E+03 2317.859493
Std 7.1725E+00 7.2328E+00 1.5290E+01 3.4512E+00 8.2383E+00 6.3649E+00 23.1692393
Best 2.3128E+03 2.3372E+03 2.3200E+03 2.3123E+03 2.3200E+03 2.3242E+03 2200
Rank 3 7 6 2 5 4 1
f22 Mean 2.3000E+03 2.3018E+03 2.4653E+03 2.3000E+03 2.3000E+03 2.3000E+03 2300
Std 9.9916E-13 9.3337E-01 9.0545E+02 3.3286E-11 1.2554E-12 1.1506E-11 1.02383E-12
Best 2.3000E+03 2.3006E+03 2.3000E+03 2.3000E+03 2.3000E+03 2.3000E+03 2300
Rank 1 6 7 5 2 4 3
f23 Mean 2.7138E+03 2.7033E+03 2.6988E+03 2.6817E+03 2.7084E+03 2.6897E+03 2670.228782
Std 2.3018E+01 1.6178E+01 2.4299E+01 1.0680E+01 1.3420E+01 1.4465E+01 13.57317699
Best 2.6664E+03 2.6803E+03 2.6531E+03 2.6596E+03 2.6976E+03 2.6546E+03 2647.050445
Rank 7 5 4 2 6 3 1
f24 Mean 2.8502E+03 2.8357E+03 2.8369E+03 2.8373E+03 2.8579E+03 2.8226E+03 2835.354439
Std 1.6277E+01 2.3240E+01 1.4301E+01 1.0267E+01 1.5567E+01 3.5202E+01 7.760136514
Best 2.8275E+03 2.7849E+03 2.8122E+03 2.8182E+03 2.8228E+03 2.7000E+03 2817.89306
Rank 6 3 4 5 7 1 2
f25 Mean 2.9098E+03 2.9003E+03 2.8882E+03 2.8881E+03 2.8993E+03 2.8885E+03 2889.398024
Std 1.2296E+01 1.1871E+01 6.7617E+00 4.4165E-01 8.7032E+00 5.2911E+00 3.931306791
Best 2.8943E+03 2.8903E+03 2.8868E+03 2.8873E+03 2.8939E+03 2.8869E+03 2887.594298
Rank 7 6 2 1 5 3 4
f26 Mean 4.0252E+03 3.7637E+03 3.3320E+03 3.7454E+03 4.0673E+03 3.6546E+03 3784.810858
Std 3.6005E+02 6.7219E+02 5.1818E+02 2.4249E+02 1.2915E+02 6.4532E+02 117.3587905
Best 2.8000E+03 2.8070E+03 2.8000E+03 2.9000E+03 3.8710E+03 2.8000E+03 3559.353884
Rank 6 4 1 3 7 2 5
f27 Mean 3.3590E+03 3.2700E+03 3.2440E+03 3.2372E+03 3.3418E+03 3.2598E+03 3235.548896
Std 3.5923E+01 1.6116E+01 2.0844E+01 7.8539E+00 2.9268E+01 1.1506E+01 7.878746139
Best 3.2798E+03 3.2326E+03 3.2141E+03 3.2198E+03 3.2936E+03 3.2374E+03 3214.013112
Rank 7 5 3 2 6 4 1
f28 Mean 3.1547E+03 3.2245E+03 3.1630E+03 3.1691E+03 3.2279E+03 3.2097E+03 3173.610357
Std 7.8074E+01 1.2859E+01 6.0587E+01 5.8196E+01 8.9298E+01 1.1052E+01 57.99070724
Best 3.1000E+03 3.2062E+03 3.1000E+03 3.1000E+03 3.1000E+03 3.1953E+03 3100
Rank 1 6 2 3 7 5 4
f29 Mean 3.4413E+03 3.5297E+03 3.4865E+03 3.3645E+03 3.4118E+03 3.3987E+03 3361.467447
Std 1.3233E+02 1.6246E+02 1.4273E+02 6.3216E+01 8.0782E+01 5.5121E+01 50.27807923
Best 3.3499E+03 3.3790E+03 3.3328E+03 3.3140E+03 3.3618E+03 3.3640E+03 3315.248384
Rank 5 7 6 2 4 3 1
f30 Mean 1.8274E+04 9.0547E+03 1.0336E+04 7.8049E+03 1.7119E+04 8.8389E+03 7917.882751
Std 9.3417E+03 1.3397E+03 3.4306E+03 7.5317E+02 4.6057E+03 1.6642E+03 927.8588573
Best 8.1222E+03 6.8227E+03 6.7050E+03 6.2390E+03 9.4013E+03 7.3399E+03 6373.418561
Rank 7 4 5 1 6 3 2
Total Rank 107 168 130 86 110 130 81
Final Rank 3 7 5 2 4 6 1

VOLUME 4, 2016 19

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 6. Comparison of the CEC2017 test functions: 30 dimensions.

SPSO2011 CLPSO VS DSA GSA BMA OSA BOA ASO CALFASO

f1 Mean 1.9426E+03 1.1055E+02 2.3531E+03 1.0117E+02 1.5857E+03 1.0000E+02 4.7942E+10 5.0843E+10 2.8309E+03 1.6789E+03
Std 1.4619E+03 1.1740E+01 3.4786E+03 1.8156E+00 9.6918E+02 4.2293E-03 5.3644E+09 2.4512E+09 3.1513E+03 1.9920E+03
Best 1.0038E+02 1.0112E+02 1.0004E+02 1.0000E+02 1.1525E+02 1.0000E+02 4.2054E+10 4.7244E+10 1.2012E+02 1.0267E+02
Rank 6 3 7 2 4 1 9 10 8 5
f3 Mean 3.0000E+02 1.5860E+04 3.0000E+02 3.5403E+04 6.7853E+04 3.0000E+02 8.3587E+04 8.4636E+04 4.0690E+02 3.0001E+02
Std 4.2554E-04 3.4045E+03 3.0477E-10 6.1564E+03 8.2471E+03 1.3268E-13 1.1628E+04 4.1405E+03 1.0579E+02 1.0490E-02
Best 3.0000E+02 1.0564E+04 3.0000E+02 2.4290E+04 5.9341E+04 3.0000E+02 6.8440E+04 7.4017E+04 3.0000E+02 3.0000E+02
Rank 2 6 2 7 8 2 9 10 5 4
f4 Mean 4.5755E+02 4.7808E+02 4.9692E+02 4.9261E+02 5.2964E+02 4.3754E+02 9.9343E+03 1.8850E+04 4.9854E+02 4.9947E+02
Std 3.6145E+01 1.1152E+01 1.1739E+01 1.0884E+01 1.1507E+01 2.9036E+01 1.6442E+03 2.4801E+03 3.3077E+01 4.4275E+01
Best 4.0014E+02 4.3831E+02 4.6901E+02 4.7632E+02 5.0087E+02 4.0000E+02 6.8146E+03 1.6478E+04 4.0004E+02 4.0004E+02
Rank 2 3 5 4 8 1 9 10 6 7
f5 Mean 5.4006E+02 5.5550E+02 5.9001E+02 5.5843E+02 7.0901E+02 6.7162E+02 9.2662E+02 8.7741E+02 5.3094E+02 5.2070E+02
Std 1.9820E+01 4.6410E+00 2.3963E+01 7.0339E+00 1.4905E+01 5.5716E+01 1.7817E+01 1.5787E+01 8.6713E+00 4.2002E+00
Best 5.2288E+02 5.4842E+02 5.6368E+02 5.4138E+02 6.7312E+02 5.5572E+02 8.7731E+02 8.4866E+02 5.1691E+02 5.1293E+02
Rank 3 4 6 5 8 7 10 9 2 1
f6 Mean 6.0568E+02 6.0000E+02 6.1066E+02 6.0000E+02 6.4015E+02 6.0836E+02 6.9175E+02 6.7135E+02 6.0002E+02 6.0000E+02
Std 1.8987E+00 7.3844E-12 8.1066E+00 4.7206E-14 5.4088E+00 5.8872E+00 8.3926E+00 7.2294E+00 4.7153E-02 1.3518E-13
Best 6.0214E+02 6.0000E+02 6.0247E+02 6.0000E+02 6.3286E+02 6.0124E+02 6.7764E+02 6.6273E+02 6.0000E+02 6.0000E+02
Rank 5 2 7 2 8 6 10 9 4 2
f7 Mean 7.9056E+02 8.1118E+02 8.3758E+02 7.8842E+02 7.4816E+02 9.6983E+02 1.4513E+03 1.3566E+03 7.5315E+02 7.4671E+02
Std 1.1394E+01 8.1093E+00 3.3381E+01 1.1023E+01 2.8105E+00 5.4447E+01 5.1389E+01 4.8177E+01 6.2409E+00 3.9169E+00
Best 7.6153E+02 7.9329E+02 7.7817E+02 7.6776E+02 7.4153E+02 8.0226E+02 1.2962E+03 1.2401E+03 7.4460E+02 7.4015E+02
Rank 5 6 7 4 2 8 10 9 3 1
f8 Mean 8.4442E+02 8.6230E+02 8.8693E+02 8.6485E+02 9.4258E+02 9.6674E+02 1.1516E+03 1.1031E+03 8.2836E+02 8.2225E+02
Std 8.2993E+00 5.2922E+00 2.1951E+01 9.6401E+00 1.5361E+01 7.9469E+01 2.5344E+01 9.8498E+00 9.7056E+00 5.2680E+00
Best 8.3558E+02 8.5125E+02 8.5671E+02 8.4656E+02 9.2536E+02 8.4278E+02 1.1045E+03 1.0850E+03 8.1592E+02 8.1492E+02
Rank 3 4 6 5 7 8 10 9 2 1
f9 Mean 9.3710E+02 9.1540E+02 1.7347E+03 9.0000E+02 1.3973E+03 1.4494E+03 1.1225E+04 9.3847E+03 9.0000E+02 9.0000E+02
Std 3.4627E+01 1.2805E+01 8.5485E+02 4.3362E-03 2.6100E+02 5.4920E+02 1.7674E+03 6.5695E+02 9.9020E-14 1.6346E-02
Best 9.0490E+02 9.0265E+02 9.1415E+02 9.0000E+02 9.0000E+02 9.3608E+02 8.2313E+03 8.7546E+03 9.0000E+02 9.0000E+02
Rank 5 4 8 2 6 7 10 9 2 2
f10 Mean 5.8178E+03 3.9222E+03 3.9450E+03 3.7796E+03 4.7705E+03 7.0927E+03 8.2648E+03 8.4823E+03 3.9156E+03 3.2868E+03
Std 5.1543E+02 3.2308E+02 5.0059E+02 2.0103E+02 5.0319E+02 1.2369E+03 5.2503E+02 4.9933E+02 6.4577E+02 4.9312E+02
Best 4.7762E+03 3.3599E+03 3.0011E+03 3.3940E+03 3.2615E+03 4.6455E+03 7.5796E+03 7.7172E+03 2.5439E+03 2.2374E+03
Rank 7 4 5 2 6 8 9 10 3 1
f11 Mean 1.2046E+03 1.1438E+03 1.2594E+03 1.1391E+03 1.2087E+03 1.2467E+03 9.4742E+03 7.2083E+03 1.1794E+03 1.1378E+03
Std 3.2623E+01 1.2333E+01 3.3950E+01 1.8080E+01 2.2155E+01 5.9419E+01 1.3782E+03 1.1891E+03 4.2510E+01 2.8780E+01
Best 1.1483E+03 1.1265E+03 1.1972E+03 1.1129E+03 1.1750E+03 1.1179E+03 6.9070E+03 4.8567E+03 1.1309E+03 1.1091E+03
Rank 5 3 8 2 6 7 10 9 4 1
f12 Mean 1.9961E+05 2.9121E+05 7.2970E+05 2.0267E+05 5.8154E+05 3.5345E+03 1.3586E+10 1.4536E+10 1.0656E+05 4.0698E+04
Std 9.5309E+04 1.0780E+05 7.7720E+05 1.2515E+05 1.9092E+05 1.4127E+03 1.1579E+09 2.8325E+09 7.6075E+04 9.5672E+04
Best 2.5251E+04 1.5121E+05 7.2952E+04 4.6168E+04 2.4447E+05 1.9016E+03 1.0571E+10 6.4389E+09 1.3212E+04 7.8970E+03
Rank 4 6 8 5 7 1 9 10 3 2
f13 Mean 5.0153E+04 1.9997E+03 5.7977E+04 1.4926E+04 3.1990E+04 1.8270E+03 5.1627E+09 1.1560E+10 6.8061E+03 8.8769E+03
Std 2.0063E+04 1.3199E+02 1.3831E+04 1.1298E+04 9.2749E+03 1.9620E+02 9.7627E+08 6.9048E+09 6.8050E+03 4.5078E+03
Best 2.9187E+04 1.7286E+03 3.4876E+04 1.4040E+03 1.2300E+04 1.5615E+03 2.2588E+09 2.1001E+09 1.7745E+03 3.3042E+03
Rank 7 2 8 5 6 1 9 10 3 4
f14 Mean 8.2249E+03 4.2260E+04 4.0773E+03 2.6440E+03 9.6866E+04 1.5115E+03 1.8545E+07 1.0529E+07 7.4708E+03 6.2899E+03
Std 5.7553E+03 3.1667E+04 1.7438E+03 1.5638E+03 3.2689E+04 9.9903E+00 1.1303E+07 7.9296E+06 3.1860E+03 6.2093E+03
Best 2.3890E+03 6.7166E+03 1.9071E+03 1.4815E+03 3.2311E+04 1.4918E+03 2.4766E+06 2.3975E+06 1.6246E+03 1.5288E+03
Rank 6 7 3 2 8 1 10 9 5 4
f15 Mean 3.4134E+04 1.7479E+03 5.0028E+04 3.3100E+03 7.1013E+03 1.6827E+03 7.5165E+08 7.4219E+08 3.7759E+03 4.3839E+03
Std 1.8347E+04 5.6388E+01 1.8743E+04 2.7112E+03 1.2354E+03 4.7810E+01 8.0467E+07 3.2077E+08 2.5389E+03 2.9829E+03
Best 1.2628E+04 1.6357E+03 1.8730E+04 1.5491E+03 4.8609E+03 1.6134E+03 6.0037E+08 8.5317E+07 1.5537E+03 1.5077E+03
Rank 7 2 8 3 6 1 10 9 4 5
f16 Mean 2.3213E+03 2.1018E+03 2.2112E+03 2.1976E+03 3.0976E+03 3.2210E+03 5.4986E+03 6.7942E+03 2.2552E+03 2.2046E+03
Std 2.1476E+02 5.6135E+01 2.9366E+02 1.5070E+02 2.2444E+02 3.4195E+02 3.8547E+02 8.7274E+02 2.5207E+02 1.7278E+02
Best 1.9419E+03 1.9295E+03 1.7110E+03 1.9261E+03 2.6319E+03 2.2651E+03 4.6179E+03 5.2520E+03 1.8458E+03 1.9384E+03
Rank 6 1 4 2 7 8 9 10 5 3
f17 Mean 1.8877E+03 1.7946E+03 1.9227E+03 1.8493E+03 2.7592E+03 2.2500E+03 6.4219E+03 1.1139E+04 1.8549E+03 1.7871E+03
Std 3.6845E+01 4.8019E+01 1.2726E+02 6.2447E+01 1.7669E+02 1.2619E+02 2.6523E+03 6.5756E+03 6.3442E+01 8.6547E+01
Best 1.7834E+03 1.7485E+03 1.7718E+03 1.7548E+03 2.2973E+03 1.9877E+03 4.1825E+03 2.8243E+03 1.7414E+03 1.7113E+03
Rank 5 2 6 3 8 7 9 10 4 1
f18 Mean 2.7423E+05 1.2589E+05 7.5493E+04 5.9223E+04 1.7664E+05 1.9012E+03 9.3371E+07 1.2064E+08 1.4873E+05 8.9615E+04
Std 1.9205E+05 3.9739E+04 4.9247E+04 2.3129E+04 8.3631E+04 2.0495E+01 1.5491E+08 1.0425E+08 5.2685E+04 6.6163E+04
Best 3.3389E+04 6.6802E+04 4.0514E+04 2.3051E+04 8.6858E+04 1.8658E+03 3.1289E+06 5.6284E+06 5.1650E+04 2.6147E+04
Rank 8 5 3 2 7 1 9 10 6 4
f19 Mean 2.0354E+05 2.0273E+03 1.7638E+05 4.9114E+03 4.8075E+03 1.9660E+03 2.6669E+08 8.0809E+08 5.1459E+03 6.5130E+03
Std 1.4108E+05 7.2511E+01 9.2239E+04 4.8485E+03 1.2632E+03 1.1467E+01 1.4587E+08 5.1856E+08 4.6667E+03 6.4255E+03
Best 3.2643E+04 1.9328E+03 4.7016E+03 1.9215E+03 3.3070E+03 1.9484E+03 1.4070E+08 3.4173E+08 1.9618E+03 1.9210E+03
Rank 8 2 7 4 3 1 9 10 5 6
f20 Mean 2.2251E+03 2.1486E+03 2.3694E+03 2.1613E+03 2.9159E+03 2.3972E+03 3.0639E+03 2.8454E+03 2.2157E+03 2.1961E+03
Std 6.0568E+01 5.2770E+01 7.4850E+01 8.1642E+01 2.7203E+02 1.1621E+02 1.4448E+02 4.7939E+01 7.5229E+01 1.1313E+02
Best 2.1780E+03 2.0329E+03 2.2944E+03 2.0498E+03 2.5250E+03 2.1508E+03 2.9348E+03 2.7995E+03 2.0549E+03 2.0292E+03
Rank 5 1 6 2 9 7 10 8 4 3
f21 Mean 2.3643E+03 2.3485E+03 2.4024E+03 2.3631E+03 2.5326E+03 2.4743E+03 2.7409E+03 2.4544E+03 2.3314E+03 2.3179E+03
Std 1.2561E+01 1.9378E+01 3.2340E+01 1.1508E+01 2.4738E+01 5.4033E+01 2.8725E+01 1.5678E+02 5.9500E+00 2.3169E+01
Best 2.3359E+03 2.2564E+03 2.3423E+03 2.3397E+03 2.4761E+03 2.3609E+03 2.6671E+03 2.3026E+03 2.3204E+03 2.2000E+03
Rank 5 3 6 4 9 8 10 7 2 1
f22 Mean 2.3002E+03 2.3296E+03 2.3002E+03 2.5042E+03 4.7912E+03 2.3005E+03 9.9150E+03 4.1767E+03 2.3000E+03 2.3000E+03
Std 7.4989E-01 1.0979E+01 7.6117E-01 7.7797E+02 2.5542E+03 1.1251E+00 6.3151E+02 7.1225E+02 1.3271E-12 1.0238E-12
Best 2.3000E+03 2.3082E+03 2.3000E+03 2.3000E+03 2.3000E+03 2.3000E+03 8.3244E+03 3.5762E+03 2.3000E+03 2.3000E+03
Rank 3 6 4 7 9 5 10 8 1.5 1.5
f23 Mean 2.7046E+03 2.7020E+03 2.7504E+03 2.7157E+03 3.3939E+03 2.8356E+03 3.5251E+03 3.3686E+03 2.6910E+03 2.6702E+03
Std 1.6003E+01 4.7604E+00 2.9502E+01 1.2273E+01 8.6796E+01 6.1290E+01 1.2041E+02 1.9558E+02 3.6397E+01 1.3573E+01
Best 2.6660E+03 2.6882E+03 2.6927E+03 2.6825E+03 3.1687E+03 2.7228E+03 3.3805E+03 3.1273E+03 2.6642E+03 2.6471E+03
Rank 4 3 6 5 9 7 10 8 2 1
f24 Mean 2.8700E+03 2.8913E+03 2.8856E+03 2.9086E+03 3.1594E+03 2.9957E+03 3.7316E+03 3.7909E+03 2.8862E+03 2.8354E+03
Std 1.1180E+01 1.2555E+01 2.2135E+01 1.4156E+01 6.8608E+01 4.9860E+01 1.9610E+02 1.9316E+02 6.6882E+01 7.7601E+00
Best 2.8359E+03 2.8785E+03 2.8546E+03 2.8796E+03 3.0309E+03 2.8925E+03 3.5418E+03 3.3610E+03 2.8095E+03 2.8179E+03
Rank 2 5 3 6 8 7 9 10 4 1
f25 Mean 2.9214E+03 2.8870E+03 2.8869E+03 2.8872E+03 2.9281E+03 2.8871E+03 4.3261E+03 5.5655E+03 2.8894E+03 2.8894E+03
Std 1.8883E+01 7.2416E-02 2.7781E+00 2.7452E-01 7.6266E+00 7.7768E-01 2.3311E+02 3.7518E+02 8.2721E+00 3.9313E+00
Best 2.8837E+03 2.8869E+03 2.8835E+03 2.8868E+03 2.9110E+03 2.8835E+03 4.0371E+03 4.8705E+03 2.8837E+03 2.8876E+03
Rank 7 2 1 4 8 3 9 10 5 6
f26 Mean 3.7522E+03 3.6893E+03 4.5587E+03 4.0443E+03 4.1512E+03 5.0754E+03 1.1437E+04 1.0894E+04 3.2931E+03 3.7848E+03
Std 7.2952E+02 4.4501E+02 2.9646E+02 5.3337E+02 1.4499E+03 7.8780E+02 6.7641E+02 1.2800E+03 5.6074E+02 1.1736E+02
Best 2.8000E+03 3.1246E+03 4.2424E+03 2.9002E+03 2.8000E+03 2.9000E+03 1.0191E+04 6.7726E+03 2.8000E+03 3.5594E+03
Rank 3 2 7 5 6 8 10 9 1 4
f27 Mean 3.2469E+03 3.2096E+03 3.2316E+03 3.2069E+03 3.8086E+03 3.2296E+03 4.8477E+03 3.8324E+03 3.2383E+03 3.2355E+03
Std 1.4859E+01 3.9895E+00 1.2737E+01 4.6645E+00 2.6232E+02 1.8087E+01 3.8111E+02 2.4594E+02 2.7991E+01 7.8787E+00
Best 3.2326E+03 3.2028E+03 3.2079E+03 3.1959E+03 3.5577E+03 3.2031E+03 3.7640E+03 3.5681E+03 3.2095E+03 3.2140E+03
Rank 7 2 4 1 8 3 10 9 6 5
f28 Mean 3.1072E+03 3.2089E+03 3.2135E+03 3.2126E+03 3.2113E+03 3.1583E+03 6.4373E+03 7.7863E+03 3.1262E+03 3.1736E+03
Std 2.7596E+01 3.0263E+00 2.3126E+01 6.6064E+00 4.3553E+01 7.1234E+01 9.5181E+02 4.2374E+02 4.8706E+01 5.7991E+01
Best 3.1000E+03 3.2058E+03 3.1508E+03 3.1999E+03 3.1648E+03 3.1000E+03 5.4825E+03 7.4975E+03 3.1000E+03 3.1000E+03
Rank 1 5 8 7 6 3 9 10 2 4
f29 Mean 3.5995E+03 3.4362E+03 3.6297E+03 3.4996E+03 4.2990E+03 4.2262E+03 7.0542E+03 1.1235E+04 3.5857E+03 3.3615E+03
Std 5.1602E+01 4.0443E+01 2.1525E+02 6.8414E+01 1.9674E+02 2.8834E+02 5.4516E+02 2.4009E+03 1.2811E+02 5.0278E+01
Best 3.4859E+03 3.3570E+03 3.4548E+03 3.4043E+03 3.9341E+03 3.6995E+03 5.5400E+03 6.8607E+03 3.3510E+03 3.3152E+03
Rank 5 2 6 3 8 7 9 10 4 1
f30 Mean 8.0203E+04 8.6245E+03 1.0017E+06 8.8405E+03 4.3001E+04 6.2791E+03 3.6257E+09 2.2349E+09 1.1142E+04 7.9179E+03
Std 4.5997E+04 8.0043E+02 5.1296E+05 2.3617E+03 6.4792E+03 1.4352E+03 1.0582E+09 6.8193E+08 3.2059E+03 9.2786E+02
Best 3.2120E+04 6.3308E+03 2.6081E+05 6.2534E+03 3.0899E+04 5.3453E+03 1.0781E+09 7.3103E+08 7.0342E+03 6.3734E+03
Rank 7 3 8 4 6 1 10 9 5 2
Total Rank 143 100 167 109 201 135 276 270 110.5 83.5
Final Rank 6 2 7 3 8 5 10 9 4 1

20 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 7. Comparison of the CEC2017 test functions: 50 dimensions.

SPSO2011 CLPSO VS DSA GSA BMA OSA BOA ASO CALFASO

f1 Mean 2.3185E+03 3.6300E+03 1.3618E+04 2.1101E+03 4.2090E+02 3.8197E+03 8.7642E+10 1.0153E+11 4.2268E+03 1.5693E+03
Std 2.5349E+03 2.3996E+03 9.5989E+03 2.2379E+03 5.6471E+02 3.2917E+03 9.8333E+09 5.8386E+09 3.7682E+03 1.5242E+03
Best 1.2928E+02 1.2445E+03 3.7418E+02 1.1449E+02 1.0274E+02 1.0002E+02 7.4624E+10 9.2062E+10 1.7968E+02 1.0074E+02
Rank 4 5 8 3 1 6 9 10 7 2
f3 Mean 9.6554E+03 9.1596E+04 3.0000E+02 1.3725E+05 1.6872E+05 3.1438E+02 2.4017E+05 1.9187E+05 1.5259E+04 4.5907E+03
Std 1.8766E+03 9.4480E+03 1.8344E-04 2.2761E+04 1.3339E+04 1.5933E+01 3.8893E+04 1.6781E+04 3.6922E+03 2.1481E+03
Best 6.6867E+03 7.7224E+04 3.0000E+02 9.7793E+04 1.4994E+05 3.0030E+02 1.8107E+05 1.5586E+05 1.0819E+04 1.6356E+03
Rank 4 6 1 7 8 2 10 9 5 3
f4 Mean 5.3012E+02 5.3065E+02 5.6742E+02 5.2497E+02 7.1141E+02 4.7975E+02 3.6722E+04 4.0399E+04 5.9235E+02 5.6708E+02
Std 6.2147E+01 1.1695E+01 4.1174E+01 1.3924E+01 1.4427E+02 3.4587E+01 6.5165E+03 3.7493E+03 5.8084E+01 5.2349E+01
Best 4.2853E+02 5.0304E+02 5.0721E+02 5.0296E+02 4.7024E+02 4.2851E+02 1.9413E+04 3.1158E+04 5.0401E+02 4.7152E+02
Rank 3 4 6 2 8 1 9 10 7 5
f5 Mean 6.7903E+02 6.8577E+02 6.7952E+02 6.6082E+02 8.3672E+02 7.9768E+02 1.1542E+03 1.1225E+03 5.7548E+02 5.5741E+02
Std 1.2582E+01 1.2273E+01 3.6970E+01 1.5415E+01 2.4951E+01 1.2151E+02 3.2787E+01 2.0311E+01 1.5298E+01 8.0456E+00
Best 6.5194E+02 6.4223E+02 6.2835E+02 6.3580E+02 7.9053E+02 6.0149E+02 1.1252E+03 1.0738E+03 5.4278E+02 5.3582E+02
Rank 4 6 5 3 8 7 10 9 2 1
f6 Mean 6.1271E+02 6.0000E+02 6.2151E+02 6.0000E+02 6.5171E+02 6.2739E+02 7.0565E+02 6.9005E+02 6.0003E+02 6.0000E+02
Std 1.9757E+00 1.0927E-07 8.4734E+00 4.1790E-11 4.5534E+00 6.3610E+00 3.2282E+00 3.4330E+00 6.6523E-02 3.0738E-13
Best 6.0816E+02 6.0000E+02 6.1067E+02 6.0000E+02 6.4143E+02 6.1966E+02 6.9352E+02 6.8516E+02 6.0000E+02 6.0000E+02
Rank 5 2 6 2 8 7 10 9 4 2
f7 Mean 9.4457E+02 9.8940E+02 1.0024E+03 9.1884E+02 8.7973E+02 1.2650E+03 2.0667E+03 1.9309E+03 7.9543E+02 7.9144E+02
Std 2.0082E+01 1.6140E+01 8.0623E+01 2.2136E+01 2.9275E+01 3.4042E+01 4.4957E+01 4.2184E+01 7.9903E+00 1.0272E+01
Best 8.8122E+02 9.6208E+02 8.5624E+02 8.6823E+02 8.4926E+02 1.1645E+03 1.9329E+03 1.8516E+03 7.8077E+02 7.7741E+02
Rank 5 6 7 4 3 8 10 9 2 1
f8 Mean 9.4376E+02 9.9060E+02 9.8713E+02 9.5632E+02 1.1577E+03 1.2781E+03 1.4677E+03 1.4756E+03 8.8441E+02 8.5751E+02
Std 2.5218E+01 1.1711E+01 4.2415E+01 1.8939E+01 3.3295E+01 4.7225E+01 3.8899E+01 1.4273E+01 1.5267E+01 1.0202E+01
Best 8.7064E+02 9.7029E+02 9.1740E+02 9.2525E+02 1.1174E+03 1.1814E+03 1.4047E+03 1.4366E+03 8.7263E+02 8.3980E+02
Rank 3 6 5 4 7 8 9 10 2 1
f9 Mean 9.9329E+02 1.5611E+03 5.3520E+03 9.6218E+02 7.7498E+03 1.3428E+04 3.7563E+04 3.4408E+04 9.0000E+02 9.0287E+02
Std 4.1301E+01 2.3732E+02 2.2495E+03 6.2952E+01 5.8811E+02 5.4894E+03 2.1016E+03 1.9285E+03 0.0000E+00 4.1103E+00
Best 9.3705E+02 1.3065E+03 2.9462E+03 9.0059E+02 6.4708E+03 1.8097E+03 3.6024E+04 2.9682E+04 9.0000E+02 9.0000E+02
Rank 4 5 6 3 7 8 10 9 1 2
f10 Mean 1.0017E+04 8.5934E+03 6.1674E+03 7.0551E+03 7.8475E+03 1.3655E+04 1.4928E+04 1.4895E+04 5.1106E+03 5.4960E+03
Std 7.3117E+02 2.4868E+02 7.6587E+02 3.6455E+02 5.4639E+02 1.7099E+03 1.0096E+03 2.8853E+02 1.1252E+03 7.6080E+02
Best 9.0780E+03 7.8227E+03 5.0507E+03 6.1113E+03 6.3873E+03 8.5048E+03 1.2920E+04 1.4527E+04 3.5589E+03 4.5455E+03
Rank 7 6 3 4 5 8 10 9 1 2
f11 Mean 1.2478E+03 1.2256E+03 1.2898E+03 1.1891E+03 1.4490E+03 1.3283E+03 2.3131E+04 2.5987E+04 1.2619E+03 1.1663E+03
Std 2.8485E+01 1.4179E+01 6.7417E+01 1.7742E+01 7.2972E+01 8.9573E+01 1.7114E+03 2.7341E+03 4.5367E+01 1.0146E+01
Best 1.2051E+03 1.1966E+03 1.1992E+03 1.1657E+03 1.3196E+03 1.2194E+03 1.7094E+04 2.0355E+04 1.1969E+03 1.1430E+03
Rank 4 3 6 2 8 7 9 10 5 1
f12 Mean 2.1745E+06 3.5961E+06 1.3381E+07 2.8634E+06 1.6695E+06 1.2853E+05 8.9934E+10 5.9517E+10 1.0234E+06 6.8850E+05
Std 9.4846E+05 1.2226E+06 1.0796E+07 1.2186E+06 5.6845E+05 5.9938E+04 2.1819E+10 9.7537E+09 2.5456E+05 1.7507E+05
Best 2.3398E+05 2.4297E+06 2.6306E+06 1.2466E+06 8.8002E+05 1.5108E+04 6.3015E+10 4.5157E+10 2.0695E+05 2.8942E+05
Rank 5 7 8 6 4 1 10 9 3 2
f13 Mean 5.5789E+04 5.9120E+03 1.2318E+05 8.2292E+03 2.6304E+04 6.1208E+03 4.9939E+10 3.9600E+10 2.5337E+03 2.8088E+03
Std 3.6003E+04 1.3521E+03 8.3382E+04 1.5354E+04 3.9200E+03 2.1551E+03 7.2157E+09 8.1286E+09 1.6349E+03 1.3368E+03
Best 3.1038E+04 3.0511E+03 3.8393E+04 1.6555E+03 1.7382E+04 3.6926E+03 4.2368E+10 2.4144E+10 1.6023E+03 1.4272E+03
Rank 7 3 8 5 6 4 10 9 1 2
f14 Mean 1.0827E+05 1.7080E+05 5.2347E+04 8.5329E+04 6.5374E+05 1.6410E+03 8.4720E+07 1.9763E+08 4.9473E+04 2.9609E+04
Std 5.7193E+04 1.6690E+05 2.5418E+04 8.0309E+04 1.1905E+06 2.1564E+01 7.7982E+07 8.8917E+07 1.8833E+04 1.7056E+04
Best 1.2957E+04 2.0207E+04 4.9604E+03 4.5757E+03 4.0701E+04 1.6169E+03 1.8126E+07 9.4995E+07 4.7379E+03 8.7512E+03
Rank 6 7 4 5 8 1 9 10 3 2
f15 Mean 1.8838E+04 2.2029E+03 4.6034E+04 7.2497E+03 1.5129E+04 2.2999E+03 7.2848E+09 7.6385E+09 5.9365E+03 3.4289E+03
Std 5.8785E+03 2.7292E+02 1.8814E+04 4.4797E+03 3.5503E+03 1.4457E+02 1.4095E+09 2.5331E+09 2.6074E+03 1.6759E+03
Best 7.8949E+03 1.8298E+03 2.0313E+04 1.5735E+03 4.7678E+03 2.0134E+03 6.5651E+09 3.5947E+09 1.6930E+03 1.5748E+03
Rank 7 1 8 5 6 2 9 10 4 3
f16 Mean 2.7362E+03 2.8225E+03 3.0410E+03 3.1475E+03 3.7253E+03 5.1605E+03 9.4976E+03 1.0734E+04 2.7219E+03 2.3228E+03
Std 2.7054E+02 1.6936E+02 3.9126E+02 2.1197E+02 3.4967E+02 3.2223E+02 1.5805E+03 7.6873E+02 3.3329E+02 2.8069E+02
Best 1.9855E+03 2.1991E+03 2.2850E+03 2.6057E+03 2.9377E+03 4.0204E+03 7.2015E+03 9.0995E+03 2.1780E+03 1.8480E+03
Rank 3 4 5 6 7 8 9 10 2 1
f17 Mean 2.5752E+03 2.5577E+03 2.8091E+03 2.6714E+03 3.6713E+03 4.0799E+03 1.0341E+04 2.3884E+04 2.6480E+03 2.5408E+03
Std 2.0983E+02 1.0970E+02 1.6929E+02 2.0302E+02 2.4923E+02 1.9673E+02 2.2801E+03 1.9220E+04 2.8659E+02 1.8979E+02
Best 2.1632E+03 2.2972E+03 2.3929E+03 2.1881E+03 2.7491E+03 3.8503E+03 6.5478E+03 7.1740E+03 2.2565E+03 2.0164E+03
Rank 3 2 6 5 7 8 9 10 4 1
f18 Mean 4.0555E+05 9.0968E+05 4.2995E+05 1.0041E+06 1.3112E+06 2.2636E+03 2.6198E+08 1.8839E+08 2.1585E+05 2.3743E+05
Std 3.3573E+05 2.7095E+05 1.1085E+05 6.6400E+05 9.2744E+05 7.1562E+01 6.4650E+07 1.0547E+08 4.9018E+05 7.1288E+05
Best 5.5358E+04 2.5730E+05 2.5090E+05 2.2129E+05 4.6811E+05 2.1725E+03 1.8923E+08 8.6237E+07 3.2906E+04 4.0200E+04
Rank 4 6 5 7 8 1 10 9 2 3
f19 Mean 8.2746E+04 2.2525E+03 6.4043E+05 1.5370E+04 1.5999E+04 2.1117E+03 2.5403E+09 3.2148E+09 1.8248E+04 1.6443E+04
Std 1.5484E+04 3.0137E+02 2.8245E+05 8.0363E+03 3.8807E+03 3.5979E+01 5.3576E+08 9.4184E+08 6.5448E+03 5.9479E+03
Best 2.0637E+04 2.0794E+03 9.7323E+04 4.2011E+03 1.1583E+04 2.0288E+03 1.8432E+09 1.9770E+09 3.0932E+03 4.7816E+03
Rank 7 2 8 3 4 1 9 10 6 5
f20 Mean 2.9604E+03 2.4574E+03 2.8122E+03 2.7774E+03 3.6987E+03 3.8478E+03 4.1945E+03 4.1145E+03 2.4954E+03 2.2935E+03
Std 2.3245E+02 1.1433E+02 1.6839E+02 1.5166E+02 2.8692E+02 2.0756E+02 1.8686E+02 1.4120E+02 3.0790E+02 2.3555E+02
Best 2.6976E+03 2.3612E+03 2.2857E+03 2.4001E+03 3.0919E+03 3.3815E+03 4.0154E+03 3.6743E+03 2.1148E+03 2.0470E+03
Rank 6 2 5 4 7 8 10 9 3 1
f21 Mean 2.4295E+03 2.4837E+03 2.5111E+03 2.4751E+03 2.7339E+03 2.6857E+03 3.2238E+03 3.0689E+03 2.3806E+03 2.3624E+03
Std 3.4514E+01 9.6471E+00 3.8142E+01 2.7544E+01 2.0934E+01 1.2097E+02 5.9203E+01 8.2102E+01 2.3040E+01 1.0370E+01
Best 2.3618E+03 2.4670E+03 2.4323E+03 2.4147E+03 2.7044E+03 2.4135E+03 3.1075E+03 2.8330E+03 2.3460E+03 2.3426E+03
Rank 3 5 6 4 8 7 10 9 2 1
f22 Mean 1.0584E+04 1.0231E+04 8.0882E+03 6.1253E+03 1.0857E+04 1.3589E+04 1.7167E+04 1.4074E+04 6.9873E+03 6.7136E+03
Std 2.9208E+03 1.2868E+03 8.9223E+02 3.4072E+03 5.5088E+02 3.0078E+03 7.9457E+02 3.2559E+03 1.1065E+03 1.3199E+03
Best 2.3000E+03 3.5165E+03 5.9867E+03 2.3000E+03 1.0192E+04 2.3000E+03 1.5404E+04 8.6037E+03 6.1134E+03 2.3000E+03
Rank 6 5 4 1 7 8 10 9 3 2
f23 Mean 2.9266E+03 2.9005E+03 2.9187E+03 2.9148E+03 4.2634E+03 3.1047E+03 4.4074E+03 4.1798E+03 2.8729E+03 2.7960E+03
Std 6.5598E+01 1.4289E+01 4.5154E+01 1.7667E+01 1.2406E+02 1.4438E+02 1.7256E+02 1.3520E+02 7.5229E+01 1.8585E+01
Best 2.7662E+03 2.8760E+03 2.8695E+03 2.8720E+03 3.9674E+03 2.9119E+03 4.0265E+03 3.8204E+03 2.7880E+03 2.7526E+03
Rank 6 3 5 4 9 7 10 8 2 1
f24 Mean 3.0278E+03 3.1304E+03 3.1308E+03 3.1088E+03 3.6769E+03 3.2599E+03 5.1065E+03 4.9718E+03 3.2576E+03 2.9413E+03
Std 3.0453E+01 1.4014E+01 4.5904E+01 2.5223E+01 4.3345E+01 1.2014E+02 3.4094E+02 2.7626E+02 3.3216E+02 2.0375E+01
Best 2.9741E+03 3.1084E+03 3.0293E+03 3.0547E+03 3.6048E+03 3.0645E+03 4.5799E+03 4.7824E+03 2.9993E+03 2.9145E+03
Rank 2 4 5 3 8 7 10 9 6 1
f25 Mean 3.0528E+03 3.0510E+03 3.0318E+03 3.0368E+03 3.3225E+03 3.0216E+03 1.2714E+04 1.5671E+04 3.0766E+03 3.0520E+03
Std 4.1947E+01 9.7126E+00 2.2364E+01 1.7193E+01 1.1524E+02 3.8106E+01 1.1095E+03 5.2185E+02 3.4639E+01 3.8818E+01
Best 2.9602E+03 3.0325E+03 2.9618E+03 3.0145E+03 3.1320E+03 2.9560E+03 1.0191E+04 1.4315E+04 3.0107E+03 2.9922E+03
Rank 6 4 2 3 8 1 9 10 7 5
f26 Mean 5.5160E+03 5.4581E+03 5.8288E+03 5.4944E+03 7.3722E+03 8.3705E+03 1.5360E+04 1.7743E+04 4.9456E+03 4.5119E+03
Std 7.8332E+02 1.1487E+02 4.0108E+02 2.6461E+02 2.3262E+03 1.5538E+03 1.1113E+03 3.5546E+02 4.2997E+02 1.1046E+02
Best 2.9000E+03 5.2160E+03 5.4700E+03 4.9788E+03 2.9000E+03 5.1484E+03 1.4427E+04 1.6442E+04 2.9000E+03 4.3200E+03
Rank 5 3 6 4 7 8 9 10 2 1
f27 Mean 3.5558E+03 3.3314E+03 3.3942E+03 3.3322E+03 5.9469E+03 3.4696E+03 8.2637E+03 5.7453E+03 3.7808E+03 3.3471E+03
Std 6.1418E+01 1.8787E+01 7.6548E+01 2.4879E+01 2.7418E+02 1.1595E+02 7.5914E+02 5.3318E+02 1.7812E+02 4.8287E+01
Best 3.5018E+03 3.2876E+03 3.3126E+03 3.2890E+03 5.3705E+03 3.3415E+03 6.8116E+03 5.0229E+03 3.5249E+03 3.2775E+03
Rank 6 1 4 2 9 5 10 8 7 3
f28 Mean 3.3017E+03 3.3187E+03 3.2861E+03 3.3195E+03 3.4447E+03 3.2990E+03 1.3964E+04 1.3623E+04 3.2851E+03 3.3285E+03
Std 1.8054E+01 1.5899E+01 2.6009E+01 1.4703E+01 7.3751E+01 2.1669E+01 1.8389E+03 1.0225E+03 2.7886E+01 3.9590E+01
Best 3.2591E+03 3.3022E+03 3.2589E+03 3.2846E+03 3.3833E+03 3.2588E+03 9.5234E+03 1.1117E+04 3.2635E+03 3.2663E+03
Rank 4 5 2 6 8 3 10 9 1 7
f29 Mean 4.0465E+03 3.5644E+03 4.4488E+03 3.8491E+03 5.3194E+03 5.6038E+03 6.6659E+04 2.7752E+05 3.8859E+03 3.5640E+03
Std 1.8575E+02 9.2016E+01 3.9732E+02 1.3028E+02 3.0420E+02 4.5959E+02 4.0509E+04 4.6120E+05 1.5411E+02 1.0556E+02
Best 3.8424E+03 3.4269E+03 3.7804E+03 3.6028E+03 4.6283E+03 4.4617E+03 3.8063E+04 6.9230E+04 3.4983E+03 3.2797E+03
Rank 5 2 6 3 7 8 9 10 4 1
f30 Mean 6.8222E+06 6.9627E+05 2.5981E+07 8.5203E+05 2.2766E+07 1.7540E+06 7.9786E+09 7.9353E+09 2.8753E+06 1.3094E+06
Std 4.9800E+05 3.8325E+04 4.1339E+06 6.0419E+04 3.8674E+06 9.5451E+05 1.8480E+09 1.8221E+09 4.4061E+05 1.4948E+05
Best 5.3973E+06 6.4742E+05 1.8786E+07 7.2624E+05 1.5162E+07 6.7421E+05 5.3349E+09 2.8281E+09 2.2394E+06 1.0802E+06
Rank 6 1 8 2 7 4 10 9 5 3
Total Rank 140 116 158 112 198 154 278 271 103 65
Final Rank 5 4 7 3 8 6 10 9 2 1

VOLUME 4, 2016 21

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 8. Comparison of the CEC2017 test functions: 100 dimensions.

SPSO2011 CLPSO VS DSA GSA BMA OSA BOA ASO CALFASO

f1 Mean 6.7068E+03 4.0058E+03 1.1722E+04 3.0800E+03 2.7902E+03 3.4028E+03 2.3696E+11 2.4757E+11 4.5076E+03 4.4481E+03
Std 7.3099E+03 2.6051E+03 9.9536E+03 2.1507E+03 2.3538E+03 4.6420E+03 1.5456E+10 1.7048E+10 5.4520E+03 2.5689E+03
Best 1.0436E+02 9.0239E+02 8.2570E+02 1.6587E+02 8.6901E+02 5.1252E+02 2.2136E+11 2.2249E+11 1.9998E+02 3.3569E+02
Rank 7 4 8 2 1 3 9 10 6 5
f3 Mean 2.1922E+05 4.5267E+05 8.7238E+04 4.4389E+05 3.5419E+05 8.8353E+04 3.6313E+05 3.2724E+05 2.1190E+05 1.3011E+05
Std 2.3507E+04 2.0650E+04 1.4353E+04 5.8526E+04 6.6589E+03 7.1626E+04 4.4865E+03 2.0034E+04 2.5929E+04 1.4688E+04
Best 1.7463E+05 4.2506E+05 6.1048E+04 3.1789E+05 3.4758E+05 3.9831E+04 3.5300E+05 3.1329E+05 1.5471E+05 1.0179E+05
Rank 5 10 1 9 7 2 8 6 4 3
f4 Mean 6.5499E+02 7.3101E+02 6.8076E+02 6.5054E+02 2.4042E+03 6.5629E+02 1.0923E+05 1.1625E+05 8.7223E+02 7.1567E+02
Std 4.5683E+01 3.2526E+01 4.2793E+01 3.7557E+01 1.0434E+03 2.5759E+01 1.6779E+04 6.4064E+03 1.1680E+02 3.9152E+01
Best 5.4100E+02 6.6241E+02 6.2996E+02 5.7990E+02 9.5600E+02 5.9559E+02 7.2884E+04 9.7419E+04 6.2151E+02 6.1929E+02
Rank 2 6 4 1 8 3 9 10 7 5
f5 Mean 1.0026E+03 1.2003E+03 1.0446E+03 1.0354E+03 1.3071E+03 1.3053E+03 2.1032E+03 2.0250E+03 8.0048E+02 6.8848E+02
Std 1.7267E+02 3.1740E+01 4.1472E+01 4.0059E+01 5.5773E+01 3.1371E+02 3.0304E+01 2.5209E+01 4.7068E+01 2.2396E+01
Best 6.9800E+02 1.1376E+03 9.4176E+02 9.7105E+02 1.2104E+03 1.0264E+03 2.0532E+03 1.9331E+03 7.1989E+02 6.5422E+02
Rank 3 6 5 4 8 7 10 9 2 1
f6 Mean 6.2114E+02 6.0000E+02 6.4852E+02 6.0000E+02 6.5853E+02 6.6571E+02 7.1478E+02 7.0466E+02 6.0278E+02 6.0000E+02
Std 4.1058E+00 8.7758E-05 6.1684E+00 5.5848E-06 1.5276E+00 9.1283E+00 1.8883E+00 2.2203E+00 1.2867E+00 9.2252E-03
Best 6.1402E+02 6.0000E+02 6.3050E+02 6.0000E+02 6.5506E+02 6.3372E+02 7.1032E+02 6.9735E+02 6.0021E+02 6.0000E+02
Rank 5 2 6 2 7 8 10 9 4 2
f7 Mean 1.3828E+03 1.5789E+03 1.5572E+03 1.3417E+03 1.9888E+03 2.4242E+03 4.0702E+03 3.9195E+03 9.4478E+02 9.5415E+02
Std 9.3432E+01 2.1406E+01 8.7604E+01 4.6526E+01 7.0526E+01 3.9858E+02 2.9575E+01 6.9159E+01 1.8066E+01 2.6872E+01
Best 1.0417E+03 1.5312E+03 1.4359E+03 1.2123E+03 1.8114E+03 1.5388E+03 3.9805E+03 3.7558E+03 9.1560E+02 9.1877E+02
Rank 4 6 5 3 7 8 10 9 1 2
f8 Mean 1.3109E+03 1.4935E+03 1.2811E+03 1.3462E+03 1.7730E+03 1.8445E+03 2.5877E+03 2.4693E+03 1.0758E+03 9.7491E+02
Std 1.3727E+02 2.1652E+01 8.3335E+01 4.6748E+01 3.7880E+01 2.3362E+02 3.7314E+01 3.1542E+01 1.8478E+01 1.5095E+01
Best 9.5521E+02 1.4250E+03 1.1532E+03 1.2420E+03 1.6855E+03 1.2696E+03 2.5426E+03 2.4162E+03 1.0348E+03 9.5123E+02
Rank 4 6 3 5 7 8 10 9 2 1
f9 Mean 1.3817E+03 1.6214E+04 1.8942E+04 7.7306E+03 2.0229E+04 4.0822E+04 7.7044E+04 7.9001E+04 9.7446E+02 1.3590E+03
Std 3.1801E+02 1.7076E+03 3.0165E+03 3.8074E+03 7.7695E+02 1.1303E+04 3.9796E+03 1.8377E+03 2.3703E+02 3.0481E+02
Best 9.7912E+02 1.1538E+04 1.3146E+04 2.3511E+03 1.9317E+04 3.0790E+04 6.8089E+04 7.1838E+04 9.0018E+02 1.0612E+03
Rank 3 5 6 4 7 8 9 10 1 2
f10 Mean 2.4586E+04 2.4264E+04 1.4054E+04 1.8587E+04 1.5636E+04 3.1303E+04 3.1908E+04 3.1214E+04 1.4235E+04 1.1971E+04
Std 1.2629E+03 4.4650E+02 9.0751E+02 4.7059E+02 5.3408E+02 4.5490E+02 7.9087E+02 9.9288E+02 2.3992E+03 1.3347E+03
Best 2.2732E+04 2.2498E+04 1.2072E+04 1.7719E+04 1.3523E+04 2.9552E+04 2.9394E+04 3.0405E+04 1.1118E+04 9.9943E+03
Rank 7 6 2 5 4 9 10 8 3 1
f11 Mean 2.1827E+03 8.6490E+03 2.6931E+03 1.4189E+04 7.6244E+04 2.8028E+03 3.4239E+05 2.3377E+05 3.2743E+03 1.8540E+03
Std 1.6246E+02 1.4832E+03 2.2098E+02 3.4409E+03 2.2112E+04 5.3675E+02 9.0314E+04 2.4477E+04 5.7792E+02 4.0446E+02
Best 1.9305E+03 7.0326E+03 2.2388E+03 4.7177E+03 4.1340E+04 2.1992E+03 2.4813E+05 2.0471E+05 1.9685E+03 1.4784E+03
Rank 2 6 3 7 8 4 10 9 5 1
f12 Mean 1.0101E+07 2.1354E+07 1.0124E+08 2.0587E+07 1.0546E+08 1.0855E+06 2.1676E+11 1.7566E+11 4.2319E+06 4.1753E+06
Std 5.0622E+06 3.3866E+06 4.5239E+07 6.5089E+06 1.3055E+08 9.0416E+05 1.8698E+10 1.7849E+10 8.2466E+06 5.0999E+06
Best 3.3167E+06 1.5210E+07 3.5820E+07 8.2724E+06 1.1467E+07 3.7124E+05 1.5587E+11 1.4542E+11 1.7008E+06 8.3034E+05
Rank 4 6 7 5 8 1 10 9 3 2
f13 Mean 4.0778E+04 1.0813E+04 7.9535E+04 5.3751E+03 2.8094E+04 1.0915E+04 4.2267E+10 4.1254E+10 8.2876E+03 3.3367E+03
Std 1.1646E+04 7.3087E+03 5.1392E+04 2.7759E+03 3.8862E+03 6.6848E+03 4.7092E+09 8.0883E+09 3.7625E+03 1.3509E+03
Best 1.9222E+04 4.5872E+03 2.4925E+04 1.8171E+03 2.0370E+04 4.3084E+03 2.9567E+10 2.7832E+10 6.1674E+03 1.6594E+03
Rank 7 4 8 2 6 5 10 9 3 1
f14 Mean 1.1671E+05 4.7858E+06 1.0017E+06 3.1015E+06 1.3111E+06 2.1121E+03 4.0423E+07 2.0253E+08 1.4492E+06 6.1161E+05
Std 2.9318E+04 1.4947E+06 7.4009E+05 1.3388E+06 3.0201E+05 9.4856E+01 1.1865E+07 5.1554E+07 8.9131E+05 1.1487E+06
Best 5.5687E+04 1.9475E+06 1.4746E+05 1.3085E+06 7.0883E+05 1.9265E+03 2.2691E+07 7.8264E+07 7.3344E+04 1.2774E+05
Rank 2 8 4 7 5 1 9 10 6 3
f15 Mean 3.2146E+04 4.8620E+03 5.9885E+04 2.7733E+03 1.2751E+04 4.0042E+03 2.0542E+10 2.1451E+10 2.5773E+03 2.6740E+03
Std 9.8034E+03 7.8242E+02 1.2288E+04 1.4675E+03 1.3640E+03 2.8567E+03 2.9144E+09 1.8256E+09 1.0759E+03 8.5828E+02
Best 1.8013E+04 3.2711E+03 3.8512E+04 1.7577E+03 1.1267E+04 2.3443E+03 1.6746E+10 1.8277E+10 1.7768E+03 1.6572E+03
Rank 7 5 8 3 6 4 9 10 1 2
f16 Mean 4.8923E+03 5.2795E+03 5.2195E+03 6.4114E+03 7.5073E+03 8.7887E+03 2.4245E+04 2.3773E+04 5.4557E+03 4.5253E+03
Std 6.4482E+02 2.2175E+02 5.2636E+02 3.4851E+02 5.8345E+02 2.7067E+03 2.6356E+03 8.5084E+02 9.5056E+02 3.1800E+02
Best 3.6270E+03 4.8481E+03 4.4833E+03 5.3901E+03 6.5167E+03 5.1258E+03 2.1328E+04 2.0862E+04 3.4001E+03 3.8476E+03
Rank 2 4 3 6 7 8 10 9 5 1
f17 Mean 4.2031E+03 3.8914E+03 4.6159E+03 5.1413E+03 4.9393E+03 6.8336E+03 1.0921E+07 1.2346E+07 3.8904E+03 3.9898E+03
Std 6.3350E+02 3.9784E+02 3.6102E+02 3.3261E+02 4.1630E+02 9.5576E+02 1.7177E+07 4.4438E+06 2.0383E+02 3.5721E+02
Best 3.2832E+03 3.6159E+03 3.9196E+03 4.1132E+03 4.4479E+03 5.7674E+03 9.5427E+04 5.1259E+06 3.5949E+03 3.4764E+03
Rank 4 2 5 7 6 8 9 10 1 3
f18 Mean 1.8801E+05 5.5632E+06 1.0649E+06 3.9859E+06 7.7170E+05 5.0571E+04 6.4679E+07 2.9213E+08 5.6498E+05 3.4943E+05
Std 5.7520E+04 1.4266E+06 6.4167E+05 1.4175E+06 1.1007E+05 1.3073E+04 3.5735E+07 1.0546E+08 1.9467E+05 9.0286E+04
Best 1.0355E+05 4.2493E+06 3.2364E+05 1.8912E+06 5.2229E+05 3.3170E+04 3.1944E+07 9.8400E+07 1.9372E+05 1.6027E+05
Rank 2 8 6 7 5 1 9 10 4 3
f19 Mean 3.6442E+05 3.8755E+03 2.8193E+06 3.2398E+03 1.2661E+04 4.1504E+03 2.4876E+10 2.1159E+10 4.0124E+03 2.8593E+03
Std 9.1334E+04 4.1224E+02 1.2563E+06 1.1990E+03 5.8774E+03 2.2527E+03 1.3088E+09 2.9372E+09 1.9445E+03 7.1232E+02
Best 1.8289E+05 2.8031E+03 8.6457E+05 2.0292E+03 4.7692E+03 2.4713E+03 2.1224E+10 1.6406E+10 2.1184E+03 1.9782E+03
Rank 7 3 8 2 6 5 10 9 4 1
f20 Mean 5.2244E+03 4.4138E+03 4.3708E+03 5.3973E+03 6.3563E+03 7.2917E+03 8.0430E+03 7.6968E+03 4.4766E+03 3.9420E+03
Std 2.4160E+02 1.2215E+02 4.5748E+02 2.6816E+02 2.1581E+02 3.3699E+02 5.3022E+02 1.3385E+02 5.7677E+02 3.4417E+02
Best 4.8432E+03 4.0596E+03 3.6516E+03 4.9846E+03 5.8814E+03 6.9379E+03 6.5205E+03 7.2882E+03 3.0331E+03 3.5905E+03
Rank 5 3 2 6 7 8 10 9 4 1
f21 Mean 2.7973E+03 3.0188E+03 2.8460E+03 2.8591E+03 3.6389E+03 3.2875E+03 4.8641E+03 4.7829E+03 2.7057E+03 2.5066E+03
Std 1.2040E+02 1.3860E+01 6.4015E+01 4.0170E+01 6.5395E+01 3.0521E+02 1.1844E+02 1.0307E+02 1.6299E+02 3.2462E+01
Best 2.5406E+03 2.9876E+03 2.7364E+03 2.7924E+03 3.5070E+03 2.7930E+03 4.6045E+03 4.5922E+03 2.5777E+03 2.4648E+03
Rank 3 6 4 5 8 7 10 9 2 1
f22 Mean 2.6727E+04 2.6911E+04 1.6800E+04 2.0597E+04 1.9982E+04 3.2633E+04 3.4748E+04 3.4853E+04 1.5872E+04 1.4943E+04
Std 8.5928E+02 2.2258E+02 1.6674E+03 5.3352E+02 5.9043E+02 1.1677E+03 7.1968E+02 4.1518E+02 2.5518E+03 1.5573E+03
Best 2.5315E+04 2.6337E+04 1.4773E+04 1.9309E+04 1.9457E+04 2.8306E+04 3.3073E+04 3.3518E+04 1.3314E+04 1.1866E+04
Rank 6 7 3 5 4 8 9 10 2 1
f23 Mean 3.2875E+03 3.2616E+03 3.4774E+03 3.1720E+03 6.7668E+03 4.0455E+03 7.1697E+03 5.9804E+03 4.6073E+03 3.1136E+03
Std 1.3079E+02 1.3749E+01 9.3944E+01 1.9138E+01 2.1466E+02 2.4297E+02 3.2999E+02 1.1835E+02 4.3921E+02 9.2363E+01
Best 3.0828E+03 3.2303E+03 3.2819E+03 3.1253E+03 6.2235E+03 3.4196E+03 6.2261E+03 5.7157E+03 3.9337E+03 3.0042E+03
Rank 4 3 5 2 9 6 10 8 7 1
f24 Mean 3.7258E+03 3.9309E+03 3.9776E+03 3.8093E+03 5.6501E+03 4.3477E+03 1.0450E+04 1.1479E+04 5.5130E+03 3.4807E+03
Std 1.0536E+02 2.1287E+01 1.1666E+02 4.9667E+01 3.1652E+02 2.3457E+02 7.7849E+02 1.1241E+03 1.1564E+03 5.6336E+01
Best 3.5188E+03 3.8705E+03 3.7914E+03 3.7050E+03 5.2362E+03 3.8762E+03 9.7358E+03 8.3207E+03 3.7694E+03 3.4107E+03
Rank 2 4 5 3 8 6 9 10 7 1
f25 Mean 3.3152E+03 3.3895E+03 3.3600E+03 3.3549E+03 4.7358E+03 3.2901E+03 2.2305E+04 2.7199E+04 3.6603E+03 3.3936E+03
Std 4.7656E+01 3.1444E+01 1.0210E+02 2.5704E+01 4.0539E+02 8.4110E+01 2.8552E+03 2.1157E+03 9.1201E+01 7.3903E+01
Best 3.2368E+03 3.2967E+03 3.2244E+03 3.2612E+03 4.1862E+03 3.1722E+03 2.0097E+04 2.5842E+04 3.4740E+03 3.2676E+03
Rank 2 5 4 3 8 1 9 10 7 6
f26 Mean 1.4960E+04 1.2388E+04 1.2169E+04 1.1253E+04 2.3454E+04 1.9079E+04 4.6361E+04 5.5681E+04 1.2856E+04 8.4999E+03
Std 4.3238E+03 2.7669E+02 8.5902E+02 3.6694E+02 1.6292E+03 2.8626E+03 2.7170E+03 2.2714E+03 3.6992E+03 4.5719E+02
Best 7.7665E+03 1.1877E+04 9.7479E+03 1.0209E+04 1.6059E+04 1.2433E+04 4.2003E+04 5.1744E+04 2.9000E+03 7.3060E+03
Rank 6 4 3 2 8 7 9 10 5 1
f27 Mean 4.0223E+03 3.5423E+03 3.6395E+03 3.4940E+03 1.1082E+04 3.6262E+03 1.5764E+04 1.2878E+04 4.3048E+03 3.5548E+03
Std 1.4156E+02 9.5226E+00 9.7873E+01 2.7437E+01 7.0665E+02 8.0621E+01 1.7099E+03 8.3816E+02 4.1854E+02 2.8623E+01
Best 3.7560E+03 3.5157E+03 3.4481E+03 3.4332E+03 8.6821E+03 3.4466E+03 1.0392E+04 1.1090E+04 3.9704E+03 3.5129E+03
Rank 6 2 5 1 8 4 10 9 7 3
f28 Mean 3.3939E+03 3.5318E+03 3.4494E+03 3.5207E+03 4.7153E+03 3.3737E+03 2.6490E+04 3.7380E+04 3.6404E+03 3.5529E+03
Std 4.7816E+01 2.8017E+01 3.1580E+01 2.1096E+01 4.8484E+02 3.7066E+01 1.2463E+03 8.6558E+02 1.1637E+02 5.5922E+01
Best 3.3318E+03 3.4972E+03 3.3917E+03 3.4865E+03 4.1483E+03 3.3452E+03 2.3441E+04 3.5166E+04 3.5072E+03 3.4264E+03
Rank 2 5 3 4 8 1 9 10 7 6
f29 Mean 7.5628E+03 5.9581E+03 6.9020E+03 6.3854E+03 8.9149E+03 1.1068E+04 3.6581E+05 1.1787E+06 6.0599E+03 5.4925E+03
Std 5.3242E+02 1.5006E+02 4.2128E+02 2.5362E+02 3.5597E+02 1.1606E+03 1.3814E+05 3.9646E+05 3.8116E+02 4.5759E+02
Best 6.7624E+03 5.5554E+03 6.6038E+03 5.9063E+03 8.0644E+03 5.9527E+03 9.3311E+04 2.6453E+05 5.5206E+03 4.6986E+03
Rank 6 2 5 4 7 8 9 10 3 1
f30 Mean 2.0537E+06 3.1793E+04 1.9726E+07 2.1573E+04 3.1703E+07 9.8280E+03 3.9170E+10 3.7327E+10 3.8297E+05 3.8801E+04
Std 6.3328E+05 8.7869E+03 1.0382E+07 9.0995E+03 3.4718E+07 3.2167E+03 3.0368E+09 3.8367E+09 1.8699E+06 2.1192E+04
Best 9.0240E+05 1.6935E+04 6.0348E+06 7.9494E+03 7.2909E+05 6.5806E+03 3.3405E+10 2.6056E+10 1.7683E+04 1.5367E+04
Rank 6 3 7 2 8 1 10 9 5 4
Total Rank 125 141 138 118 196 150 275 269 118 65
Final Rank 4 6 5 2 8 7 10 9 2 1

22 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 9. Determining the algorithm that statistically provides the best solution for each benchmark function by utilizing the two-sided Wilcoxon signed-rank test
(α = 0.05): 30 dimensions.

Fun. SPSO2011 vs CALFASO CLPSO vs CALFASO VS vs CALFASO DSA vs CALFASO GSA vs CALFASO
p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win
f1 0.41652148 275 190 = 5.75E-06 455 10 - 0.658296677 329 136 = 1.73004E-06 465 0 - 0.861198 275 190 =
f3 3.8753E-06 462 3 - 1.73E-06 0 465 + 1.72351E-06 465 0 - 1.7344E-06 0 465 + 1.72E-06 0 465 +
f4 0.00276527 450 15 - 0.049492 450 15 - 0.068713631 437 28 = 0.059835601 444 21 = 0.000125 165 300 +
f5 2.1565E-05 114 351 + 1.73E-06 0 465 + 1.73331E-06 0 465 + 1.7344E-06 0 465 + 1.72E-06 0 465 +
f6 1.5871E-06 0 465 + 1.55E-06 0 465 + 1.70299E-06 0 465 + 1 450 0 = 1.66E-06 0 465 +
f7 1.7333E-06 0 465 + 1.73E-06 0 465 + 1.7344E-06 0 465 + 1.7344E-06 0 465 + 0.120445 255 210 =
f8 1.9209E-06 30 435 + 1.73E-06 0 465 + 1.73222E-06 0 465 + 1.7344E-06 0 465 + 1.73E-06 0 465 +
f9 1.3635E-06 0 465 + 1.68E-06 0 465 + 1.69227E-06 0 465 + 9.93433E-05 30 435 + 2.03E-06 0 435 +
f10 1.7333E-06 0 465 + 3.72E-05 140 325 + 5.79245E-05 165 300 + 9.71105E-05 140 325 + 1.92E-06 30 435 +
f11 1.0246E-05 140 325 + 0.370923 255 210 = 1.73331E-06 0 465 + 0.813016851 329 136 = 2.12E-06 30 435 +
f12 1.972E-05 87 378 + 1.64E-05 30 435 + 2.35198E-06 30 435 + 2.84342E-05 30 435 + 1.73E-06 0 465 +
f13 1.7333E-06 0 465 + 1.73E-06 465 0 - 1.73004E-06 0 465 + 0.018518975 255 210 - 1.73E-06 0 465 +
f14 0.0752056 212 253 = 9.31E-06 59 406 + 0.308602058 360 105 = 0.000615641 437 28 - 1.73E-06 0 465 +
f15 1.7333E-06 0 465 + 4.07E-05 450 15 - 1.73331E-06 0 465 + 0.025637124 420 45 - 0.000453 189 276 +
f16 0.05708993 275 190 = 0.012453 374 91 - 0.68835149 345 120 = 0.958990172 360 105 = 1.73E-06 0 465 +
f17 3.3987E-05 87 378 + 0.544006 234 231 = 9.31059E-06 30 435 + 0.004114031 255 210 - 1.73E-06 0 465 +
f18 0.00020507 165 300 + 0.010444 234 231 - 0.370922658 360 105 = 0.042766688 410 55 - 0.000261 114 351 +
f19 1.7333E-06 0 465 + 1.02E-05 462 3 - 1.91613E-06 30 435 + 0.614314564 345 120 = 0.658322 294 171 =
f20 0.13589744 212 253 = 0.14704 374 91 = 1.97295E-05 114 351 + 0.503832969 329 136 = 1.73E-06 0 465 +
f21 1.7333E-06 0 465 + 2.84E-05 30 435 + 1.7344E-06 0 465 + 1.7344E-06 0 465 + 1.73E-06 0 465 +
f22 0.828125 459 6 = 1.5E-06 0 465 + 3.03484E-06 0 406 + 0.003464804 294 136 - 1.56E-06 0 465 +
f23 3.8799E-06 59 406 + 1.73E-06 0 465 + 1.73331E-06 0 465 + 1.7344E-06 0 465 + 1.73E-06 0 465 +
f24 1.7333E-06 0 465 + 1.73E-06 0 465 + 1.73331E-06 0 465 + 1.7344E-06 0 465 + 1.73E-06 0 465 +
f25 4.7265E-06 114 351 + 1.73E-06 465 0 - 2.59671E-05 464 1 - 1.7344E-06 465 0 - 1.73E-06 0 465 +
f26 0.7035637 275 190 = 0.115599 360 105 = 1.73331E-06 0 465 + 0.07864666 189 276 = 0.829009 294 171 =
f27 0.00119734 234 231 - 1.73E-06 465 0 - 0.158844482 387 78 = 1.7344E-06 465 0 - 1.73E-06 0 465 +
f28 0.00032005 399 3 - 0.057057 275 190 = 0.014766428 275 190 - 0.030009891 275 190 - 0.008723 294 171 -
f29 1.7344E-06 0 465 + 3.72E-05 114 351 + 1.73331E-06 0 465 + 3.18168E-06 87 378 + 1.73E-06 0 465 +
f30 1.73E-06 0 465 + 0.001962 189 276 + 1.7344E-06 0 465 + 0.245190309 329 136 = 1.73E-06 0 465 +
+/=/- 19/6/4 15/5/9 20/6/3 11/8/10 24/4/1
Fun BMA vs CALFASO OSA vs CALFASO BOA vs CALFASO ASO vs CALFASO
p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win
f1 1.73E-06 465 0 - 1.73E-06 0 465 + 1.73222E-06 0 465 + 0.093666919 212 253 =
f3 1.7116E-06 465 0 - 1.73E-06 0 465 + 1.71269E-06 0 465 + 1.91613E-06 30 435 +
f4 3.4036E-05 462 3 - 1.73E-06 0 465 + 1.73004E-06 0 465 + 0.14137899 399 66 =
f5 1.73E-06 0 465 + 1.73E-06 0 465 + 1.73222E-06 0 465 + 5.06967E-05 140 325 +
f6 1.7224E-06 0 465 + 1.56E-06 0 465 + 1.51654E-06 0 465 + 0.00012207 0 406 +
f7 1.7344E-06 0 465 + 1.73E-06 0 465 + 1.73331E-06 0 465 + 0.000283079 165 300 +
f8 1.7344E-06 0 465 + 1.73E-06 0 465 + 1.7344E-06 0 465 + 0.006434781 212 253 +
f9 1.7213E-06 0 465 + 1.64E-06 0 465 + 1.52926E-06 0 465 + 1 30 253 =
f10 1.7333E-06 0 465 + 1.73E-06 0 465 + 1.73331E-06 0 465 + 0.000358884 189 276 +
f11 1.7333E-06 0 465 + 1.73E-06 0 465 + 1.7344E-06 0 465 + 0.001832068 165 300 +
f12 1.7333E-06 465 0 - 1.73E-06 0 465 + 1.73331E-06 0 465 + 0.000160464 114 351 +
f13 1.7333E-06 465 0 - 1.73E-06 0 465 + 1.73004E-06 0 465 + 0.049498046 420 45 -
f14 1.7333E-06 465 0 - 1.73E-06 0 465 + 1.7344E-06 0 465 + 0.106394173 234 231 =
f15 1.7988E-05 450 15 - 1.73E-06 0 465 + 1.73331E-06 0 465 + 0.544006208 345 120 =
f16 1.9209E-06 30 435 + 1.73E-06 0 465 + 1.73331E-06 0 465 + 0.543995559 329 136 =
f17 1.7333E-06 0 465 + 1.73E-06 0 465 + 1.73004E-06 0 465 + 0.003378012 165 300 +
f18 1.7333E-06 465 0 - 1.73E-06 0 465 + 1.73331E-06 0 465 + 0.002765274 189 276 +
f19 2.352E-06 462 3 - 1.73E-06 0 465 + 1.73004E-06 0 465 + 0.271142526 399 66 =
f20 2.1619E-05 114 351 + 1.73E-06 0 465 + 1.7344E-06 0 465 + 0.171376389 212 253 =
f21 1.7333E-06 0 465 + 1.73E-06 0 465 + 2.59671E-05 189 276 + 1.23808E-05 140 325 +
f22 0.00086637 30 325 + 1.5E-06 0 465 + 1.19771E-06 0 465 + 0.5 87 120 =
f23 1.7333E-06 0 465 + 1.73E-06 0 465 + 1.73004E-06 0 465 + 0.001483498 165 300 +
f24 1.7344E-06 0 465 + 1.73E-06 0 465 + 1.7344E-06 0 465 + 0.004388579 255 210 -
f25 2.352E-06 464 1 - 1.73E-06 0 465 + 1.73331E-06 0 465 + 0.00183258 459 6 -
f26 4.2832E-06 30 435 + 1.73E-06 0 465 + 1.73004E-06 0 465 + 0.001592246 399 66 -
f27 0.04491324 437 28 - 1.73E-06 0 465 + 1.7344E-06 0 465 + 0.781258073 360 105 =
f28 0.30815392 329 136 = 1.72E-06 0 465 + 1.71269E-06 0 465 + 0.002091083 374 78 -
f29 1.7344E-06 0 465 + 1.73E-06 0 465 + 1.73331E-06 0 465 + 1.73331E-06 0 465 +
f30 6.8892E-05 459 6 - 1.73E-06 0 465 + 1.73331E-06 0 465 + 3.87306E-06 87 378 +
+/=/- 16/1/12 29/0/0 29/0/0 14/10/5

TABLE 10. Comparison of statistical results given by different algorithms for the gear train design.

Algorithms Mean Std Best

SPSO2011 8.1348038647E-12 9.1652296378E-12 2.7008571489E-12
CLPSO 4.0593438278E-12 5.1698867589E-12 2.7008571489E-12
VS 1.7270385497E-11 2.7847796849E-11 2.7008571489E-12
DSA 4.0040805335E-10 6.8370090965E-10 2.7008571489E-12
GSA 4.4250360701E-11 5.2125788104E-11 2.7008571489E-12
BMA 1.4512468982E-10 2.7903074602E-10 2.7008571489E-12
OSA 2.3859906128E-03 5.0300305134E-03 9.5208656548E-09
BOA 6.6610598600E-05 2.5085969853E-04 2.3576406580E-09
ASO 1.6007287889E-11 2.1557541408E-11 2.7008571489E-12
CALFASO 4.2302240515E-11 1.8036399248E-10 2.7008571489E-12

VOLUME 4, 2016 23

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 11. Determining the algorithm that statistically provides the best solution for each benchmark function by utilizing the two-sided Wilcoxon signed-rank test
(α = 0.05): 50 dimensions.

Fun. SPSO2011 vs CALFASO CLPSO vs CALFASO VS vs CALFASO DSA vs CALFASO GSA vs CALFASO
p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win
f1 0.42836897 294 171 = 0.000955 234 231 - 7.50959E-06 59 406 + 0.213297549 255 210 = 0.00921 399 66 -
f3 2.0952E-06 30 435 + 1.73E-06 0 465 + 1.72677E-06 465 0 - 1.73331E-06 0 465 + 1.73E-06 0 465 +
f4 0.01102822 399 66 - 0.001819 450 15 - 0.975373491 312 153 = 0.001483068 444 21 - 0.000114 140 325 +
f5 1.5263E-06 0 465 + 1.59E-06 0 465 + 1.69227E-06 0 465 + 1.72895E-06 0 465 + 1.69E-06 0 465 +
f6 1.1048E-06 0 465 + 1.57E-06 0 465 + 1.688E-06 0 465 + 1.70514E-06 0 465 + 1.49E-06 0 465 +
f7 1.6583E-06 0 465 + 1.71E-06 0 465 + 1.69655E-06 0 465 + 1.73222E-06 0 465 + 1.72E-06 0 465 +
f8 1.7181E-06 0 465 + 1.72E-06 0 465 + 1.72677E-06 0 465 + 1.73222E-06 0 465 + 1.73E-06 0 465 +
f9 1.7041E-06 0 465 + 1.72E-06 0 465 + 1.72677E-06 0 465 + 2.35198E-06 59 406 + 1.71E-06 0 465 +
f10 1.6966E-06 0 465 + 1.69E-06 0 465 + 0.001611521 275 190 - 2.34621E-06 30 435 + 1.63E-06 0 465 +
f11 1.6966E-06 0 465 + 1.66E-06 0 465 + 1.69227E-06 0 465 + 5.78983E-05 189 276 + 1.68E-06 0 465 +
f12 4.6613E-06 87 378 + 1.68E-06 0 465 + 1.7246E-06 0 465 + 1.73331E-06 0 465 + 1.73E-06 0 465 +
f13 1.703E-06 0 465 + 3.72E-06 87 378 + 1.67948E-06 0 465 + 0.000419255 140 325 + 1.66E-06 0 465 +
f14 8.9714E-06 114 351 + 1.63E-06 0 465 + 0.001577699 212 253 + 3.72426E-05 87 378 + 1.71E-06 0 465 +
f15 1.6912E-06 0 465 + 2.3E-05 450 15 - 1.69655E-06 0 465 + 0.000388111 212 253 + 1.68E-06 0 465 +
f16 7.9627E-05 140 325 + 5.04E-06 87 378 + 1.89945E-06 30 435 + 2.12664E-06 30 435 + 1.72E-06 0 465 +
f17 0.35949199 255 210 = 0.67315 329 136 = 3.36958E-05 59 406 + 0.013974564 212 253 + 1.66E-06 0 465 +
f18 0.00018425 87 378 + 2.85E-05 30 435 + 3.0586E-05 30 435 + 3.11081E-05 30 435 + 3.06E-05 30 435 +
f19 1.6998E-06 0 465 + 1.65E-06 465 0 - 1.71918E-06 0 465 + 0.428417818 374 91 = 0.64328 360 105 =
f20 1.8144E-06 30 435 + 0.003723 275 190 - 1.61469E-06 0 465 + 5.2135E-06 59 406 + 1.62E-06 0 465 +
f21 1.9477E-06 30 435 + 1.58E-06 0 465 + 1.68907E-06 0 465 + 1.7344E-06 0 465 + 1.7E-06 0 465 +
f22 9.5959E-05 87 378 + 1.68E-06 0 465 + 4.63968E-06 30 435 + 0.271142526 312 153 = 1.68E-06 0 465 +
f23 2.2723E-06 30 435 + 1.72E-06 0 465 + 1.70514E-06 0 465 + 1.7344E-06 0 465 + 1.65E-06 0 465 +
f24 1.6827E-06 0 465 + 1.65E-06 0 465 + 1.71809E-06 0 465 + 1.7344E-06 0 465 + 1.71E-06 0 465 +
f25 0.79675106 275 190 = 0.59954 420 45 = 0.040346382 420 45 - 0.082206466 410 55 = 1.66E-06 0 465 +
f26 0.00015766 59 406 + 1.72E-06 0 465 + 1.71918E-06 0 465 + 1.7344E-06 0 465 + 7.64E-06 59 406 +
f27 1.688E-06 0 465 + 0.279998 374 91 = 0.036736469 329 136 - 0.102010695 410 55 = 1.72E-06 0 465 +
f28 0.00256966 410 55 - 0.298509 374 91 = 0.000385629 420 45 - 0.289464319 345 120 = 1.71E-06 0 465 +
f29 1.6795E-06 0 465 + 0.177081 374 91 = 1.86995E-06 30 435 + 2.12532E-06 30 435 + 1.69E-06 0 465 +
f30 1.688E-06 0 465 + 1.65E-06 465 0 - 1.72895E-06 0 465 + 1.7344E-06 465 0 - 1.72E-06 0 465 +
+/=/- 24/3/2 18/5/6 23/1/5 21/6/2 27/1/1
Fun BMA vs CALFASO OSA vs CALFASO BOA vs CALFASO ASO vs CALFASO
p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win
f1 0.00237892 212 253 + 1.71E-06 0 465 + 1.67948E-06 0 465 + 0.002237559 212 253 +
f3 1.7268E-06 465 0 - 1.73E-06 0 465 + 1.73004E-06 0 465 + 1.73222E-06 0 465 +
f4 1.9006E-06 464 1 - 1.72E-06 0 465 + 1.71701E-06 0 465 + 0.101874773 275 190 =
f5 1.671E-06 0 465 + 1.69E-06 0 465 + 1.70945E-06 0 465 + 4.66666E-06 30 435 +
f6 1.4533E-06 0 465 + 1.62E-06 0 465 + 1.51849E-06 0 465 + 1.47889E-06 0 465 +
f7 1.7041E-06 0 465 + 1.72E-06 0 465 + 1.67312E-06 0 465 + 0.115357859 294 171 =
f8 1.7192E-06 0 465 + 1.73E-06 0 465 + 1.71161E-06 0 465 + 1.72026E-06 0 465 +
f9 1.6955E-06 0 465 + 1.65E-06 0 465 + 1.72134E-06 0 465 + 2.38546E-06 464 0 -
f10 1.6312E-06 0 465 + 1.68E-06 0 465 + 1.61983E-06 0 465 + 0.105931503 374 91 =
f11 1.717E-06 0 465 + 1.63E-06 0 465 + 1.69762E-06 0 465 + 1.6987E-06 0 465 +
f12 1.7203E-06 465 0 - 1.73E-06 0 465 + 1.70407E-06 0 465 + 2.53974E-05 87 378 +
f13 4.6856E-06 114 351 + 1.71E-06 0 465 + 1.72026E-06 0 465 + 0.213067448 360 105 =
f14 1.6479E-06 465 0 - 1.71E-06 0 465 + 1.71701E-06 0 465 + 0.000818862 165 300 +
f15 0.00017326 455 10 - 1.69E-06 0 465 + 1.6987E-06 0 465 + 5.71959E-05 140 325 +
f16 1.7094E-06 0 465 + 1.71E-06 0 465 + 1.64996E-06 0 465 + 0.000147072 189 276 +
f17 1.7084E-06 0 465 + 1.71E-06 0 465 + 1.70514E-06 0 465 + 0.213140129 329 136 =
f18 1.6955E-06 465 0 - 1.67E-06 0 465 + 1.6573E-06 0 465 + 0.347528924 429 36 =
f19 1.6689E-06 465 0 - 1.71E-06 0 465 + 1.67948E-06 0 465 + 0.490258801 255 210 =
f20 1.6014E-06 0 465 + 1.59E-06 0 465 + 1.6987E-06 0 465 + 0.006704096 234 231 -
f21 1.6955E-06 0 465 + 1.71E-06 0 465 + 1.69227E-06 0 465 + 0.000114466 140 325 +
f22 2.5424E-06 0 435 + 1.66E-06 0 465 + 1.67948E-06 0 465 + 0.733803304 387 78 =
f23 1.7279E-06 0 465 + 1.71E-06 0 465 + 1.66045E-06 0 465 + 1.88878E-06 30 435 +
f24 1.7094E-06 0 465 + 1.72E-06 0 465 + 1.67736E-06 0 465 + 1.70192E-06 0 465 +
f25 0.02532555 429 36 - 1.71E-06 0 465 + 1.6594E-06 0 465 + 0.021520573 189 276 +
f26 1.7008E-06 0 465 + 1.7E-06 0 465 + 1.70837E-06 0 465 + 3.34703E-05 59 406 +
f27 2.7675E-06 30 435 + 1.71E-06 0 465 + 1.651E-06 0 465 + 1.71918E-06 0 465 +
f28 0.0014519 399 66 - 1.71E-06 0 465 + 1.70514E-06 0 465 + 0.000185783 455 10 -
f29 1.6626E-06 0 465 + 1.66E-06 0 465 + 1.61366E-06 0 465 + 2.24716E-06 59 406 +
f30 0.06539147 312 153 = 1.71E-06 0 465 + 1.64786E-06 0 465 + 1.70084E-06 0 465 +
+/=/- 19/1/9 29/0/0 29/0/0 18/8/3

TABLE 12. Comparison of the statistical results given by different algorithms for the tension/compression spring design.

Algorithms Mean Std Best

SPSO2011 1.2740053422E-02 2.1015454933E-04 1.2668568083E-02
CLPSO 1.2681123221E-02 1.0067750257E-05 1.2666003761E-02
VS 2.0771906735E-03 6.0796797162E-05 2.0515878589E-03
DSA 1.2689608555E-02 2.6331407303E-05 1.2665243763E-02
GSA 1.4061168010E-02 5.5983636450E-04 1.2847699268E-02
BMA 1.2672629520E-02 1.8521468389E-05 1.2665237641E-02
OSA 1.7909558754E-02 3.9871654857E-04 1.5798493626E-02
BOA 1.5629662591E-02 1.7967403472E-03 1.3294647811E-02
ASO 1.2786005959E-02 1.6052859369E-04 1.2666145059E-02
CALFASO 1.2701335939E-02 5.0398521446E-05 1.2665461607E-02

24 VOLUME 4, 2016

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
 This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI
10.1109/ACCESS.2020.2979921, IEEE Access

Author et al.: Preparation of Papers for IEEE TRANSACTIONS and JOURNALS

TABLE 13. Determining the algorithm that statistically provides the best solution for each benchmark function by utilizing the two-sided Wilcoxon signed-rank test
(α = 0.05): 100 dimensions.

Fun. SPSO2011 vs CALFASO CLPSO vs CALFASO VS vs CALFASO DSA vs CALFASO GSA vs CALFASO
p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win
f1 0.50382169 374 91 = 0.236589 374 91 = 1.32761E-05 87 378 + 0.019514364 420 45 - 0.018145 420 45 -
f3 1.7322E-06 0 465 + 1.71E-06 0 465 + 1.671E-06 465 0 - 1.72895E-06 0 465 + 1.7E-06 0 465 +
f4 3.7067E-05 455 10 - 0.071663 234 231 = 0.000696758 455 10 - 2.15869E-05 462 3 - 1.7E-06 0 465 +
f5 1.9185E-06 30 435 + 1.71E-06 0 465 + 1.69334E-06 0 465 + 1.73222E-06 0 465 + 1.7E-06 0 465 +
f6 1.7322E-06 0 465 + 0.056171 140 325 = 1.58612E-06 0 465 + 7.67385E-06 450 15 - 1.67E-06 0 465 +
f7 1.7333E-06 0 465 + 1.71E-06 0 465 + 1.71377E-06 0 465 + 1.7344E-06 0 465 + 1.65E-06 0 465 +
f8 1.9197E-06 30 435 + 1.71E-06 0 465 + 1.66045E-06 0 465 + 1.73222E-06 0 465 + 1.7E-06 0 465 +
f9 0.97538391 345 120 = 1.4E-06 0 465 + 1.40411E-06 0 465 + 1.73331E-06 0 465 + 1.65E-06 0 465 +
f10 1.7333E-06 0 465 + 1.63E-06 0 465 + 1.73816E-05 87 378 + 1.73331E-06 0 465 + 1.89E-06 30 435 +
f11 0.00022206 59 406 + 1.7E-06 0 465 + 2.66327E-06 59 406 + 1.7344E-06 0 465 + 1.7E-06 0 465 +
f12 0.00048952 140 325 + 1.69E-06 0 465 + 1.69655E-06 0 465 + 1.92092E-06 30 435 + 1.71E-06 0 465 +
f13 1.7289E-06 0 465 + 2.01E-06 30 435 + 1.58511E-06 0 465 + 0.00089443 165 300 + 1.68E-06 0 465 +
f14 2.1201E-06 464 1 - 1.91E-06 30 435 + 0.004613221 114 351 + 5.74839E-06 30 435 + 0.002712 87 378 +
f15 1.7322E-06 0 465 + 1.89E-06 30 435 + 1.60546E-06 0 465 + 0.78126371 374 91 = 1.7E-06 0 465 +
f16 0.00642281 189 276 + 2.56E-06 59 406 + 8.20469E-06 87 378 + 1.7344E-06 0 465 + 1.68E-06 0 465 +
f17 0.13588709 275 190 = 0.404275 360 105 = 6.28576E-06 59 406 + 1.7344E-06 0 465 + 1.69E-06 0 465 +
f18 2.6002E-06 462 3 - 1.71E-06 0 465 + 4.62087E-06 59 406 + 1.7344E-06 0 465 + 1.69E-06 0 465 +
f19 1.73E-06 0 465 + 4.03E-05 114 351 + 1.68907E-06 0 465 + 0.184621877 312 153 = 1.71E-06 0 465 +
f20 1.7333E-06 0 465 + 3.68E-05 140 325 + 0.00052173 189 276 + 1.7344E-06 0 465 + 1.72E-06 0 465 +
f21 1.7333E-06 0 465 + 1.71E-06 0 465 + 1.70514E-06 0 465 + 1.7344E-06 0 465 + 1.69E-06 0 465 +
f22 1.7322E-06 0 465 + 1.68E-06 0 465 + 0.001814714 212 253 + 1.7344E-06 0 465 + 1.72E-06 0 465 +
f23 1.353E-05 59 406 + 1.68E-06 0 465 + 1.6531E-06 0 465 + 0.001035681 255 210 - 1.69E-06 0 465 +
f24 2.1253E-06 30 435 + 1.67E-06 0 465 + 1.68587E-06 0 465 + 1.7344E-06 0 465 + 1.68E-06 0 465 +
f25 0.00016021 437 28 - 0.796993 387 78 = 0.152241458 399 66 = 0.015658483 410 55 - 1.66E-06 0 465 +
f26 2.1214E-06 30 435 + 1.66E-06 0 465 + 1.66256E-06 0 465 + 1.7344E-06 0 465 + 1.64E-06 0 465 +
f27 1.7333E-06 0 465 + 0.036746 410 55 - 0.00056394 189 276 + 4.7292E-06 462 3 - 1.7E-06 0 465 +
f28 1.7322E-06 465 0 - 0.125018 387 78 = 8.35981E-06 455 10 - 0.009271025 410 55 - 1.68E-06 0 465 +
f29 1.7235E-06 0 465 + 9.47E-05 165 300 + 1.63224E-06 0 465 + 2.35198E-06 59 406 + 1.63E-06 0 465 +
f30 1.7322E-06 0 465 + 0.349091 374 91 = 1.71918E-06 0 465 + 8.18775E-05 450 15 - 1.72E-06 0 465 +
+/=/- 21/3/5 21/7/1 25/1/3 19/2/8 28/0/1
Fun BMA vs CALFASO OSA vs CALFASO BOA vs CALFASO ASO vs CALFASO
p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win p-Value R+ R- Win
f1 0.08188258 429 36 = 1.68E-06 0 465 + 1.67948E-06 0 465 + 0.338075795 410 55 =
f3 0.03318507 444 21 - 1.69E-06 0 465 + 1.61983E-06 0 465 + 1.61366E-06 0 465 +
f4 2.6025E-05 462 3 - 1.7E-06 0 465 + 1.69013E-06 0 465 + 4.10158E-06 30 435 +
f5 1.7148E-06 0 465 + 1.64E-06 0 465 + 1.64473E-06 0 465 + 1.72243E-06 0 465 +
f6 1.6752E-06 0 465 + 1.7E-06 0 465 + 1.68693E-06 0 465 + 1.70407E-06 0 465 +
f7 1.6891E-06 0 465 + 1.65E-06 0 465 + 1.64786E-06 0 465 + 0.105644529 345 120 =
f8 1.571E-06 0 465 + 1.66E-06 0 465 + 1.69227E-06 0 465 + 1.71809E-06 0 465 +
f9 1.6437E-06 0 465 + 1.68E-06 0 465 + 1.6594E-06 0 465 + 2.99539E-05 464 1 -
f10 1.6869E-06 0 465 + 1.7E-06 0 465 + 1.63744E-06 0 465 + 0.000276709 140 325 +
f11 2.8183E-06 59 406 + 1.65E-06 0 465 + 1.6987E-06 0 465 + 2.64378E-06 30 435 +
f12 0.00017395 450 15 - 1.68E-06 0 465 + 1.7246E-06 0 465 + 0.428234573 234 231 =
f13 1.8994E-06 30 435 + 1.66E-06 0 465 + 1.69013E-06 0 465 + 1.6531E-06 0 465 +
f14 1.6322E-06 465 0 - 1.65E-06 0 465 + 1.60751E-06 0 465 + 0.018028184 212 253 +
f15 0.00866761 234 231 - 1.68E-06 0 465 + 1.651E-06 0 465 + 0.236415003 374 91 =
f16 1.7148E-06 0 465 + 1.71E-06 0 465 + 1.67948E-06 0 465 + 0.000123618 165 300 +
f17 1.7127E-06 0 465 + 1.68E-06 0 465 + 1.69227E-06 0 465 + 0.205288147 387 78 =
f18 1.6944E-06 465 0 - 1.7E-06 0 465 + 1.71593E-06 0 465 + 0.000122597 140 325 +
f19 0.00023672 140 325 + 1.69E-06 0 465 + 1.6912E-06 0 465 + 0.004078592 189 276 +
f20 1.7019E-06 0 465 + 1.71E-06 0 465 + 1.68055E-06 0 465 + 0.00010494 87 378 +
f21 1.7257E-06 0 465 + 1.65E-06 0 465 + 1.69548E-06 0 465 + 1.70514E-06 0 465 +
f22 1.6944E-06 0 465 + 1.68E-06 0 465 + 1.67842E-06 0 465 + 0.089548148 294 171 =
f23 1.6322E-06 0 465 + 1.69E-06 0 465 + 1.68374E-06 0 465 + 1.68055E-06 0 465 +
f24 1.7116E-06 0 465 + 1.72E-06 0 465 + 1.71161E-06 0 465 + 1.71809E-06 0 465 +
f25 8.346E-06 459 6 - 1.67E-06 0 465 + 1.63536E-06 0 465 + 1.651E-06 0 465 +
f26 1.6636E-06 0 465 + 1.66E-06 0 465 + 1.71161E-06 0 465 + 1.87324E-05 87 378 +
f27 0.00095579 140 325 + 1.7E-06 0 465 + 1.68907E-06 0 465 + 1.67206E-06 0 465 +
f28 1.8478E-06 464 1 - 1.71E-06 0 465 + 1.70622E-06 0 465 + 0.001702015 212 253 +
f29 1.6848E-06 0 465 + 1.67E-06 0 465 + 1.671E-06 0 465 + 9.498E-05 165 300 +
f30 2.3062E-06 462 3 - 1.7E-06 0 465 + 1.70192E-06 0 465 + 0.991790439 399 66 =
+/=/- 19/1/9 29/0/0 29/0/0 21/7/1

TABLE 14. Average rankings achieved with the Friedman test on all benchmarks. The highest ranking is shown in bold.

Algorithms Ranked in different dimensions

30 50 100
SPSO2011 6.05 6.17 6.69
CLPSO 7.55 7.00 6.14
VS 5.26 5.55 6.24
DSA 7.24 7.13 6.93
GSA 4.07 4.17 4.24
BMA 6.34 5.69 5.83
OSA 1.48 1.41 1.52
BOA 1.69 1.66 1.72
ASO 7.17 7.45 6.93
CALFASO 8.14 8.75 8.76
p-value 7.68E-11 7.38E-11 4.23E-11

VOLUME 4, 2016 25

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.