Article

An Enhanced Differential Evolution Algorithm with

Bernstein Operator and Refracted Oppositional-Mutual
Learning Strategy
Fengbin Wu 1, Junxing Zhang 2, Shaobo Li 2,*, Dongchao Lv 3 and Menghan Li 3

1 State Key Laboratory of Public Big Data, College of Computer Science and Technology, Guizhou University,
Guiyang 550025, China
2 State Key Laboratory of Public Big Data, Guizhou University, Guiyang 550025, China

3 School of Mechanical Engineering, Guizhou University, Guiyang 550025, China

* Correspondence: lishaobo@gzu.edu.cn

Abstract: Numerical optimization has been a popular research topic within various engineering ap-
plications, where differential evolution (DE) is one of the most extensively applied methods. How-
ever, it is difficult to choose appropriate control parameters and to avoid falling into local optimum
and poor convergence when handling complex numerical optimization problems. To handle these
problems, an improved DE (BROMLDE) with the Bernstein operator and refracted oppositional-
mutual learning (ROML) is proposed, which can reduce parameter selection, converge faster, and
avoid trapping in local optimum. Firstly, a new ROML strategy integrates mutual learning (ML)
and refractive oppositional learning (ROL), achieving stochastic switching between ROL and ML
during the population initialization and generation jumping period to balance exploration and ex-
ploitation. Meanwhile, a dynamic adjustment factor is constructed to improve the ability of the al-
gorithm to jump out of the local optimum. Secondly, a Bernstein operator, which has no parameters
Citation: Wu, F.; Zhang, J.; Li, S.; setting and intrinsic parameters tuning phase, is introduced to improve convergence performance.
Lv, D.; Li, M. An Enhanced Finally, the performance of BROMLDE is evaluated by 10 bound-constrained benchmark functions
Differential Evolution Algorithm
from CEC 2019 and CEC 2020, respectively. Two engineering optimization problems are utilized
with Bernstein Operator and
simultaneously. The comparative experimental results show that BROMLDE has higher global op-
Refracted Oppositional-Mutual
timization capability and convergence speed on most functions and engineering problems.
Learning Strategy.
Entropy 2022, 24, 1205.
Keywords: refracted oppositional learning; mutual learning; refracted oppositional-mutual learn-
https://doi.org/10.3390/e24091205
ing; differential evolution; Bernstein operator; CEC 2019 and 2020
Academic Editor: Mikhail Sheremet

Received: 26 July 2022

Accepted: 26 August 2022
Published: 29 August 2022 1. Introduction
Publisher’s Note: MDPI stays neu- Recently, numerical optimization has become a trending research topic of interest for
tral with regard to jurisdictional many researchers and is broadly used to handle many engineering optimization prob-
claims in published maps and institu- lems, such as mobile robots path planning [1], vehicle problem [2], and task scheduling
tional affiliations. [3]. These optimization problems can be expressed as NP-Hard problems, which are dif-
ficult to derive high-quality solutions by traditional approaches owing to the reliance of
traditional methods on the choice of starting points and vulnerability to optimal local
problems [4,5]. Fortunately, the meta-heuristic algorithms (MAs) have the features of high
Copyright: © 2022 by the authors. Li- efficiency, low demands for the starting point, and robustness [6]. These overcome the
censee MDPI, Basel, Switzerland. limitations of traditional approaches to addressing NP problems. In the past decades, tre-
This article is an open access article mendous MAs have been suggested to handle numerical optimization tasks, such as dif-
distributed under the terms and con-
ferential evolution (DE) [7], PSO [8], SA [9], cockroach swarm optimization [10], and so
ditions of the Creative Commons At-
on. Among them, DE, a population-based MA, is extensively utilized in the parameter
tribution (CC BY) license (https://cre-
training of neural networks [11], problem prediction [12], and path planning of unmanned
ativecommons.org/licenses/by/4.0/).
aerial vehicles [13], etc., because of its features such as simple model, easy execution,

Entropy 2022, 24, 1205. https://doi.org/10.3390/e24091205 www.mdpi.com/journal/entropy

Entropy 2022, 24, 1205 2 of 27

stronger search capability, and robustness. Regrettably, DE is susceptible to the control

parameters and has drawbacks of easily falling into local optimum as well as low conver-
gence.
Control parameters setting may impact the convergence performance of MAs. Given
this, investigators have developed various strategies for DE algorithms. For instance, the
authors in [14] propose a quantization orthogonal crossover operator on the basis of an
orthogonal design, where the operator crossover probabilities are considered as structural
parameters. Unfortunately, its parameters are set artificially, and the stability of the algo-
rithm is difficult to be guaranteed. To further improve algorithm performance, parameter
adaptation approaches have been proposed. The authors in [15] introduce an automatic
adaptation parameter mechanism that performs depending on the deviation of the objec-
tive function between the optimal individuals and the total population in the preceding
generation, which helps improve the performance of the mutation stage. An adaptive pa-
rameter DE algorithm is reported in [16], which achieves parameter adaption using Q-
learning. Furthermore, parametric random selection methods have also been studied by
many researchers. To list a few, the authors in [17] present a self-adaptive parameters DE
algorithm, where parameters ( CR and F ) are randomly selected from a list of stored suc-
cess parameters or generated at random. Further, the authors in [18] introduce a zoning
strategy to obtain the optimal combination of control parameters, and an adaptive DE
algorithm having zoning evolution is proposed. Specifically, an easily controllable and
non-recursive Bernstein-search DE algorithm (BSDE) is reported in [19], which is not re-
quired to control the parameter setting operation and has a flexible random mutation and
crossover process. Therefore, BSDE is simpler than other parameter tuning algorithms
and aids in saving the time of algorithm search. Nevertheless, the algorithm may still be
locally optimal and low convergence. Thus, one motivation is raised for this paper.
Moreover, it is critical that the MAs should balance exploration and exploitation
[20,21]. For this, the opposition-based learning (OBL) strategies are the powerful search
framework [22–24]. The OBL mechanism is to explore improved candidate solutions by
considering both the original points and their opposite counterparts. It is appropriate for
population initialization of MAs and has performed significantly in improving the con-
vergence of the MAs [25]. Hence, many variants of the OBL strategy have been reported
to strengthen MAs in terms of trade-offs between exploration and exploitation. The au-
thors in [26] propose an opposition-based DE having a protective jumping rate, which
achieves stopping the opposite operator when the success rate of the opposite individual
falls to a constant threshold. In [27], OBL with the current optimum DE algorithm is pro-
posed, and its concept is that instead of using the center point to calculate the opposite
point, the best point of the current point is utilized. Meanwhile, OBL variants based on
expanded search space have also been extensively studied. To be specific, in [28], a re-
fracted oppositional learning (ROL) strategy is incorporated into the artificial bee colony
algorithm, promoting the diversity of the population and guiding it to explore the global
optimal solution. Based on the ROL strategy, the authors in [29] propose a cuckoo search
algorithm with refraction learning, improving the capability of cuckoo search to avoid
local optimal positions. Regrettably, their suitable scale factors are difficult to select. Fur-
thermore, a neighborhood opposition-based DE is developed in [30] by executing the
Gaussian perturbation operation around the opposite point, and its search neighborhood
is further expanded. In [31], a dynamic OBL mechanism with asymmetric search space is
proposed, which facilitates the exploitation and exploration capabilities. However, the
newly introduced weights that need to be adjusted may bring an additional burden for
applications. To overcome this shortcoming, an enhanced basic DE algorithm is devel-
oped in [32] by integrating the OBL strategy and mutual learning (ML) strategy, called the
oppositional-mutual learning DE (OMLDE) algorithm. Nevertheless, it may still suffer
from the difficulty of choosing the suitable control parameters for DE and low conver-
gence accuracy. Thus, another motivation is derived here.
Entropy 2022, 24, 1205 3 of 27

Based on the no free lunch theorem, no single algorithm can be suitable for all opti-
mization problems [33]. Therefore, it is valuable for modifying existing algorithms, devel-
oping new algorithms, and mixing different algorithms to obtain better results in practical
applications. Inspired by the above discussions, with the help of Bernstein-search, ROL
and ML strategies, an enhanced DE algorithm with both Bernstein operator and refracted
oppositional-mutual learning strategy with a dynamic adjustment factor mechanism
(called BROMLDE) is proposed to achieve fast convergence, jumping out of local opti-
mum as well as reduced parameter selection in this paper. Highlights of this paper are as
follows:
(1) The ROL strategy with a dynamic adjustment factor changed with function evalua-
tion quantity is presented, which facilitates jumping out of the local extremum space.
(2) Integrating ML into ROL strategy, a novel refracted oppositional-mutual learning
(ROML) strategy is proposed for better trade-off algorithm exploration and exploita-
tion.
(3) BROMLDE can be easily operated in parallel, achieving a rapid search. Moreover,
BROMLDE is a partially elitist selective method since it uses both the fittest points
and global minimizer solution in its system equations.
(4) Compared with BSDE [19], BROMLDE integrates the ROML into the initialization
phase of the population and the generation phase of jumping.
(5) Different from the OMLDE algorithm [32], BROMLDE does not require the adjust-
ment of intrinsic control parameters.
(6) Several numerical experiments are investigated on the CEC 2019 and CEC 2020
benchmarks to validate the function optimization performance. Additionally, two
constrained engineering problems are used to verify the feasibility of the proposed
BROMLDE.
The arrangement of the rest of this paper is as below: Section 2, the differential evo-
lution is presented. Section 3, the developed BROMLDE is stated. Numerical experiments
and results analysis are described in Section 4. Lastly, conclusions and future work are
provided in Section 5.

2. Differential Evolution Algorithm

2.1. The Structure of Typical DE
A typical DE [7] includes the mutation operator, crossover phase, and selection
phase. The population Pg is constructed as Pg  [ X 1, g , X 2, g ,, X N p , g ] for any generation ( g
), Pg is derived from Equation (1), where X i , g is the i th individual vector. Each X i , g
has D dimension, where i  1, 2,, N p , thus X i , g  [ x1i , g , x2i , g ,, xDi , g ]T .

P( i , j ), g ~ U  low j , g , up j , g  | j  1, 2,..., D , (1)

in which N p is the population Pg size, and D denotes the population dimension.

2.1.1. Mutation Operator

A mutation operator can generate a mutant vector Vi , g  [v1i , g , v2i , g ,, vDi , g ]T . For in-
stance, a classical mutation strategy “DE/rand/1” is given as follows:

Vi , g  F  X d 1, g  X d 2, g   X d 3, g , i  d 1  d 2  d 3, (2)

where three individuals X d 1, g , X d 2, g , and X d 3, g are randomly obtained in Pg . The muta-

tion scale factor F is usually in [0,1] .
Entropy 2022, 24, 1205 4 of 27

2.1.2. Crossover Phase

The trail vector U i , g  [u1i , g , u2i , g ,, uDi , g ]T is constructed according to X i , g and Vi , g in
the process of crossover operator. The vector is updated by Equation (3):

v ji , g , if rjCR‖j  n j ,
u ji , g   (3)
 x ji , g , otherwise,

where rj [0,1] , the crossover rate CR is usually in [0,1] , and the random number n j is
chosen in [1: D].

2.1.3. Selection Phase

For this phase, the individuals X i , g 1 are generated by using Equation (4):

U i , g , if f U i , g   f  X i ,G  ,
X i , g 1   (4)
 X i , g , otherwise,

where f () is a fitness function. The population Pg 1  [ X 1, g 1 , X 2, g 1 ,..., X N p , g 1 ] can be ob-

tained.

3. The Proposed BROMLDE Algorithm

ROL strategy combining the refraction principle [28,34] from physics with an OBL
strategy is a strong method to strengthen MAs [28,29,34,35]. In this paper, the ROL strat-
egy is applied to augment the performance of the BROMLDE algorithm. In addition, the
ROL strategy and ML strategy are combined to achieve improved exploitation capacity.

3.1. Bernstein Polynomials

The Bernstein polynomials can be utilized to consistently approximate a continuous
function on a closed range, where polynomials of 2nd degree [19,36] are defined using
the Equations (5) and (6):

 m m p
B p , m (t )    t p 1  t  , p  0,1,..., m, (5)
 p
 m m!
where   . The 2nd degree Bernstein polynomials are defined as Equation (6).
 p  p !( m  p )!
For p  0 and p  m, B p , m  0.

 B0,2  t   1  t 2 ,

 B1,2  t   2t 1  t  , (6)

 B2,2  t   t .
2

Figure 1a shows the 2nd degree Bernstein polynomials when 0  t  1.

Entropy 2022, 24, 1205 5 of 27

Figure 1. (a) The 2nd degree Bernstein polynomials; (b) the definition of ROL strategy.

3.2. ROL Strategy

The principle of ROL is to calculate the fitness values for the current solution and
their ROL solution and to select a superior solution by comparing the fitness values and
further iterating. The ideology of ROL is shown in Figure 1b.
In Figure 1b, the x -axis is the dividing line, the normal is the y -axis, the point O is
the midpoint of the search range [ a , b ] , and the angles of incidence and refraction are 1
and  2 , respectively, as well as l and l  indicating the length of incidence and refrac-
tion light, respectively. x indicates a point in the region [a, b] , and x stands for the re-
verse position of the point x. The geometric relationship of the lines in Figure 1b is ex-
pressed below:

sin1    a  b  / 2  x  l , (7)

sin 2   x   ( a  b ) / 2  l  . (8)

The refraction rate  is defined by using Equation (9):

sin1 l   a  b  / 2  x 


  , (9)
sin 2 l  x   a  b  / 2 


let h  l l , then Equation (9) can be reshaped as Equation (10), then the point x can be
derived by applying Equation (10):
ab ab x
x    , (10)
2 2h h

if h  1 and   1 , the Equation (10) can be changed to Equation (11) [37]:

x   a  b  x, (11)
In general, Equation (10) could be modified to handle D dimensional space problems,
 is usually taken as 1, then it gets the following formula:

a j  bj a j  bj xi , j
xi, j    , j  1, 2, , D, (12)
2 2h h
in which xi , j is the point of the j th dimension of the i th individual. The xi, j is the op-
posite position of xi , j . The a j and b j are the lower and upper bounds of the j th dimen-
sion on the search space, respectively.
Entropy 2022, 24, 1205 6 of 27

Obviously, the solution obtained by using Equation (11) is fixed, and the changing
refracted solution can be obtained by adjusting h in Equation (12), which further avoids
the locally extreme value space.

3.3. ML Strategy
Generally, in each generation, the individual having the best function value is con-
sidered to be the optimal current generation individual. Nevertheless, greater knowledge
in some dimensions may be provided by the individuals having worse fitness values.
Hence, to facilitate the exchange of knowledge, individuals should improve their
knowledge with the help of their interaction with each other. Given this, the ML strategy
is motivated [32,38].
Let x be a point in [a, b] , the ML individual can be obtained using Equation (13):

, g  xi , g  i , g  xr , g  xi , g  | r  i, i  1, 2,..., N p ,
xiML (13)

where xr , g denotes randomly chosen individual, and i , g are a random number in [0,1] .

3.4. ROML with Adjustment Factor Mechanism Strategy

Let X  [ Llgower , Lupper
g ] be a position in a D dimensional space in each generation g ,
where the bound vectors L upper
g 
 L upper
1, g
upper
,L
2, g , , L upper
D,g  and Llower
g   L1,lower
g , L2, g , , LD , g  are up-
lower lower

dated as:

Lupper  
j , g  max  x1, j , x2, j , , xN p , j  ,
  (14)

Llower
j, g  min   x 1, j , x2, j ,, xN p , j ,
 (15)

where j  1,, D. Define Ci , g  [ x1i, g , x2i , g ,, xDi , g ]T as a ROML individual of the present
generation g , it can be given by applying Equation (16):

 Llower
i, g  Lupper
i, g Llower
i, g  Lupper
i, g X i, g
   , if rand (0,1)  0.5,
Ci , g  2 2 h i ( g ) hi (g) (16)
 X  X  X ,
 i,g i, g  r,g i,g  otherwise,

in which i , g is a random value of [0,1] , X r , g (r  i ) indicates a randomly chosen individ-

ual. A ROML population PgC can be generated, i.e., PgC  C1, g , , C N p , g . Moreover, to keep
ROML effective, the boundaries should be checked using Equation (17):

C( i , j ), g  rand  Llower
j , g , L j , g  , if
upper
j , g  C ( i , j ), g  L j , g .
C ( i , j ), g  Llower upper
(17)

After the ROML step, N p most suitable individuals are chosen from P0 , PgC  ac-
cording to their fitness values. Furthermore, it is worth noting that the moderator adjust-
ment factor hi ( FES ) is an essential parameter affecting the learning performance of
ROML. To achieve a wide range of refracted inverse solutions generated in the beginning
stage of the algorithm and a small range of refracted opposite solutions generated in the
later stage, a tuning factor that can be changed with the amount of function evaluation is
designed by the trial-and-error method based on the literature [39] as follows:
1 15
 
hi ( FES )  1   FES MaxFES  3  , (18)
 
where FES are current function evaluations and maximal function evaluations are MaxFES
.
Entropy 2022, 24, 1205 7 of 27

3.4. Proposed BROMLDE

In this section, the proposed BROMLDE with Bernstein operator, ROL strategy hav-
ing adjustment factor mechanism, and ML strategy is presented in detail. First, the ROML
population initialization procedure is presented. Then, the mutation and crossover with
the Bernstein polynomials process are described. Finally, the ROML new population with
generation jumping is stated, as well as the overall procedure of the proposed BROMLDE
also be given.

3.4.1. ROML Initialization

During the starting generation ( g  0 ), the original population Pg  P0  [ X 1,0 , , X Np , 0 ]
is generated using Equation (1). The ROML strategy is utilized to generate a new initial
population P0C  [C1,0 , C2,0 ,, CNp,0 ] , where N p is the population P0C size, Ci,0 ,(i  1,..., N p )
are obtained from Equation (16). Then, a new population is constituted via Equation (19):

P0   P0  P0C  . (19)

The function values for ascending sorting of P0 are computed by Equation (20):


fitnessP0  sort   P0  ,  (20)

where  indicates the objective function. Then, N p fittest individuals are picked from
P0 by using Equation (21). Moreover, to facilitate later understanding, we set P , g  P ,0 .

 fitnessP ,0 , P ,0    fitnessP0 , P0  |   1 : N p  . (21)

Both the best solution, Pbest , and the global minimization function value, Psol , of
the problem are calculated using Equation (22):

 Psol , Pbest    min  fitnessP0  , P0  . (22)

3.4.2. Mutation and Crossover with Bernstein Polynomials

BROMLDE, which updates the starting mutation matrix M ( i , j ), g  0, i  [1: N p ], j  [1: D]
at each iteration by utilizing Equation (23), controls the mutation process using the up-
dated M g .

M (i , J ), g  1. (23)

In Equation (23), J  u  1: D  | u  permute 1: D  where the function permute() can arbi-
trarily change the sequence of the elements of ().  is given using Equation (24):

 switch 0
 2
 case 1   1    ,

 case 2   2    1    , (24)
 case 3   2,

end

where  ~ U (0,1) and 0  [3  13 ] , 1 ~ U [0 1] , 0 U 1: 3 , the  is computed employing
2 nd degree Bernstein polynomials. The step size Fg of the evolution is obtained by using
Equation (25):
Entropy 2022, 24, 1205 8 of 27


 3
 

 (1,1:D ), g   (1,1:D ), g   Q(1,1:N p ), g , if 2  3 ,
Fg  
3

(25)
 (3N ,1), g  Q(1, D ), g , otherwise,
 p
where ( 2:3) ,  g and  g are random values that will be updated with each call, where
( 2:3), g , g ~ U (0,1),  g ~ N (0,1) , and matrix Q(,), g  1 .
In BROMLDE, the trial vector Tg is obtained by making use of Equation (26):

 3
 3

Tg  Fg  M g   w*   E g  1   w*   Pbest  Pg +Pg | w(1:
*

N p ,1) ~ U (0,1), (26)

where Eg  w  PK1 , g  (1  w)  PK2 , g | w(1: N p ,1:D ), g ~ U (0,1) ,  indicates Hadamart multiplication

operator, K 1 and K 2 are specified in Equation (27):

K1  permute 1: N p  , K 2  permute 1: N p  | K1  1 : N p  , K1  K 2 . (27)

If T(i , j ), g  low j , g || T(i, j ), g  up j , g , T(i, j ), g values are updated using the Equation (28):

low j , g   3  up j , g  low j , g  , if T( i , j ), g  low j , g ,

T( i , j ), g   (28)
up j , g    low j , g  up j , g  , if T( i , j ), g  up j , g ,
3

where  ~ U (0,1) . The function values of Tg are obtained by applying Equation (29):

fitnessTg   Tg  . (29)

Based on the selection process of Equation (30), an updated population can be ob-
tained.
if fitnessT , g  fitnessP , g ,  P , g , fitnessP , g   T , g , fitnessT , g  |   1: N . (30)

3.4.3. ROML New Population with Generation Jumping

Now, based on the updated population Pg in Equation (30), a new ROML popula-
tion PgC  [C1, g , C2, g ,, CNp , g ] can be obtained by Equations (16)–(18), if the jumping rate J r
is bigger than the selection probability, where N p is the population PgC size. The objec-
tive function values of the new population PgC are computed by Equation (31):

fitnessPgC    PgC  . (31)

Then, the new population Pg , including Pg and PgC , and its objective function val-
ues for ascending sorting are denoted as follows:

 Pg   Pg  PgC 
 (32)
 fitnessPg  sort  fitnessPg  fitnessPg .
 C

Afterward, N p most suitable individuals are selected from Pg using Equation (33):

 fitnessP , g , P , g    fitnessPg , Pg  |   1 : N p  . (33)

Based on Equations (31)–(33), the individuals P , g ,   [1: N p ] obtaining a better objec-

tive function value will make up the new generation population.
In the current evolutionary step, both the optimal solution, Pbest , and correspond-
ing objective function value, Psol , are provided by the updated population Pg in Equa-
tion (33), and both of them are updated by employing Equation (34):
Entropy 2022, 24, 1205 9 of 27

 Psol , Pbest    min  fitnessP  , P  .

g g (34)

The pseudo-code of BROMLDE is provided in Algorithm 1.

Algorithm 1: The BROMLDE Algorithm Procedure

Input: Objective function:  , Search-space limits: ( low, up ), Population size: N p ,
Dimension of problem: D , Maximal function evaluations: MaxFES , Jumping
rate: J r
Output: Psol : Global minimum Pbest : Global minimizer
1 Set the current function evaluations FES  0
2 Set the population bounds randomly generate an initial population P0
3 for i 1 to N p do //ROML population initialization (generation g = 0)
4 i ,0  rand (0,1)
FES 15
5 hi ( FES )  (1  3 )
MaxFES
 Llower
i ,0  Lupper Llower  Lupper X

i ,0
 i ,0 i,0
 i ,0 , if rand (0,1)  0.5,
6 Ci ,0  2 2hi ( g ) hi ( g )
 X   ( X  X ), otherwise,
 i ,0 i , 0 r , 0 i,0

7 Check the bounds in the current generation by Equation (17)

8 end
9 Get population Pg by selecting N p fittest points from {P0  P0C }
10 FES  FES  2 N p
11 Get Psol and Pbest by Equations (20)–(22)
12 while FES  MaxFES do //Main loop(generation g > 0)
13 M (i , j ), g  0 //Generate mutation matrix ( M )
14 for i  1 to N p do
15 u  permute(1: D )
16 Generate  , where  ~ U (0,1)
17 Generate 0 ,0  [3  13 ], 1 ~ U [0 1],0  U {1: 3}
18 switch 0 do
19 case 1 do   (1   ) 2 ;
20 case 2 do   2    (1   );
21 case 3 do    2 ;
22 end
23 J  u  1:  D   | u  permute 1: D  ; M(i, J ), g  1
24 end
25 Calculate the evolutionary step size Fg by Equation (25)
26 Generate the trial vector Tg by Equations (26) and (27) and control the
boundaries of Tg by Equation (28)
27 Update population Pg by Equations (29) and (30)
28 FES  FES  N p
29 if then
rand  J r //ROML population with generation jumping
30 Update the bounds by calculating the smallest and biggest values of all
dimensions in the population Pg
31 for i 1 to N p do
32 i , g  rand (0,1)
FES 15
33 hi ( FES )  (1  3 )
MaxFES
Entropy 2022, 24, 1205 10 of 27

 Llower upper
i , g  Li , g Llower
i , g  Li , g
upper
X i, g
   , if rand (0,1)  0.5,
34 Ci , g  2 2hi ( g ) hi ( g )
 X   ( X  X ), otherwise,
 i, g i, g r, g i, g

35 Check the bounds in the current generation by Equation (17)

36 end
37 Select N p fittest individuals from {Pg  PgC } and update the population Pg
by Equation (33)
38 FES  FES  N p
39 end
40 Update Psol and Pbest by Equation (34)
41 end

3.5. Computational Complexity

The computational complexity of BROMLDE mainly depends on three parts: ROML
population initialization, mutation, crossover, and selection, as well as ROML new popu-
lation with generation jumping. The complexity of these worst-case scenarios is as below:
(1) For the ROML population initialization, the process requires generating the starting
population and its corresponding RMOL population and then selecting the N p best
individuals from the two populations as the new initial population of the algorithm.
Therefore, the time complexity is O( Np  D)  O  Np  log2 (2Np)  .
(2) In the mutation, crossover, and selection of the BROMLDE algorithm, the algorithm
mainly includes the initialization and update of starting mutation matrix M , the ob-
taining of step size Fg , and trial vector Tg . The time complexity is
O( Np  D) + O( Np) + O( D).
(3) In ROML new population with generation jumping, it consists mainly of the genera-
tion of the ROML population with Np size, and the selection of Np most suitable
individuals from the ROML population and initial population. Further, the time com-
plexity is O( Np  D)  O  Np  log2 (2Np)  .
Thus, the whole-time complexity of the developed BROMLDE can be estimated as
O( Np  D) + O( Np) + O( D)  O  Np  log2 (2Np)  .

4. Numerical Experiments and Results Analysis

4.1. Experiment Setup
To investigate the performance of the developed BROMLDE, numerical experiments
on the CEC 2019 benchmark functions [40] (see Table 1) and CEC 2020 test suites [41](see
Table 2) are conducted by comparison with various well-known optimization methods.
Those methods include BSDE [19], OMLDE [32], weighted differential evolution (WDE)
[42], adaptive DE with optional external archive (JADE) [43], success-history adaptation
DE (SHADE) [44], PSO [8], CMAES [45], and simulated annealing (SA) [9]. Moreover,
BROMLDE is also compared to the IEEE CEC 2020 winning DE algorithm variants
IMODE [46] and J2020 [47] to evaluate its performance. To make the experimental com-
parison fair, the comparison algorithm is run under identical test conditions. The whole
numerical experiments are performed on PlatEMO [48] of MATLAB 2021b on a computer
with CPU AMD Ryzen 5 3550H @2.10GHz and 16G RAM, Win10 64-bit operating system.
The population size N p is 100, and the jumping rate J r is 0.05, which is consistent with
OMLDE. Then, the maximal function evaluations ( MaxFES ) are set to 10,000 as the termina-
tion condition, and 30 independent runs are performed. Moreover, the parameter settings
of other counterparts refer to their settings.
Entropy 2022, 24, 1205 11 of 27

Moreover, the Wilcoxon rank-sum test with a significant level of   0.05 is em-
ployed to judge the difference between BROMLDE and its competitors in this paper. More
specifically, the results of taking the minimum fitness function value for each of the 30
independent runs are obtained. Then, the probability p-value corresponding to the
BROMLDE algorithm and each of its competitors is calculated separately by using
MATLAB. Finally, the determination of whether there are significant differences between
algorithms is based on the p-value and significance level  . The symbols applied to the
Wilcoxon rank-sum test are described as “+”, ”−”, and “=”, which indicate that BROMLDE
has significantly superior, inferior, and no significant difference between BROMLDE and
the compared algorithm, respectively. Furthermore, the basic statistical evaluations in-
cluding the global minimum average (AVG) and global minimum standard deviation
(STD) are utilized for the obtained minimum fitness value results.

Table 1. CEC 2019 benchmark functions [40].

No. Functions D Search Range Best

F1 Storn’s Chebyshev Polynomial Fitting Problem 9 [–8192, 8192] 1
F2 Inverse Hilbert Matrix Problem 16 [−16,384, 16,384] 1
F3 Lennard–Jones Minimum Energy Cluster 18 [−4, 4] 1
F4 Rastrigin’s Function 10 [−100, 100] 1
F5 Griewangk’s Function 10 [−100, 100] 1
F6 Weierstrass Function 10 [−100, 100] 1
F7 Modified Schwefel’s Function 10 [−100, 100] 1
F8 Expanded Schaffer’s F6 Function 10 [−100, 100] 1
F9 Happy Cat Function 10 [−100, 100] 1
F10 Ackley Function 10 [−100, 100] 1

Table 2. CEC 2020 test suites [41].

No. Functions Best

Unimodal Function F1 Shifted and Rotated Bent Cigar Function 100
F2 Shifted and Rotated Schwefel’s Function 1100
Multimodal Shifted and
F3 Shifted and Rotated Lunacek bi-Rastrigin Function 700
Rotated Functions
F4 Expanded Rosenbrock’s Plus Griewangk’s Function 1900
F5 Hybrid Function 1 (N = 3) 1700
Hybrid Functions F6 Hybrid Function 2 (N = 4) 1600
F7 Hybrid Function 3 (N = 5) 2100
F8 Composition Function 1 (N = 3) 2200
Composition Functions F9 Composition Function 2 (N = 4) 2400
F10 Composition Function 3 (N = 5) 2500
Search Range: [-100,100] ( D is the population dimension)
D

4.2. Numerical Function Optimization Problems

4.2.1. Experimental Results for CEC 2019
This subsection focuses on comparing the optimization results of BROMLDE and
other methods including BSDE, OMLDE, WDE, PSO, SHADE, JADE, and CMAES to solve
the CEC 2019 benchmark functions (see Table 1). The AVG and STD of the results obtained
from the tests performed using F1–F10 are listed in Table 3. The minimum AVG in Table
3 is highlighted in bold. Based on the results of the minimum fitness value, the Wilcoxon
rank-sum test findings of BROMLDE and its competitors in Table 4 are symbolized (+, −,
=). +, =, and − indicate that BROMLDE performs better, equal, and worse than the com-
pared methods, respectively.
Entropy 2022, 24, 1205 12 of 27

As presented in Table 1, CEC 2019 benchmark functions have different dimensions,

the dimensions of the functions F1, F2, and F3 are 9, 16, and 18, respectively. In addition,
functions F4–F10 have the same dimension 10. According to the experimental setting rules
in Section 4.1, the AVG and STD of the minimum fitness values on F1–F10 of the CEC 2019
benchmark functions are recorded in Table 3, we can see that BROMLDE obtains the min-
imum AVG for 6 functions out of the total 10 functions compared to the other algorithms,
which are functions F1, F3, F6, F7, F8, and F10.
According to the Wilcoxon rank-sum test results in Table 4 (last line), we can see that
BROMLDE achieves more than 6 significantly better results (“+”) compared to BSDE, OM-
LDE, WDE, SHADE, and CMAES; BROMLDE also yields more than 5 superior results
compared to PSO and JADE. In other words, the mean percentage of the goodness of
BROMLDE for the 10 functions is 72.86%  7i 1+i (10  7)  100%  . The general results show that
the ROML strategy can effectively enhance the optimization ability of DE.
Figure 2 presents the convergence diagrams of BROMLDE and other methods for the
10 tested functions (F1–F10) on CEC 2019, where the vertical axis is the logarithm of the
minimum value of the functions and the horizontal axis is functional evaluation numbers.
It can be seen that although PSO converges fastest, the local optimum situation occurs.
BROMLDE has a fast descent rate on most of the tested functions. Moreover, we can also
conclude that for a limited number of evaluations, smaller fitness values can be obtained
for BROMLDE on the functions F1, F3, F6, F7, F8, and F10. The good performance of
BROMLDE is due to the initialization of ROML at the start and the exploring ability of
ROML.

Table 3. Minimum fitness value on F1–F10 of CEC 2019 benchmark functions.

No. Metric BSDE OMLDE WDE PSO SHADE JADE CMAES BROMLDE
AVG 2.3365 × 1010 1.3686 × 109 6.0428 × 1010 2.2371 × 1010 8.8385 × 109 3.1163 × 1010 8.6816 × 1010 1.1022 × 109
F1
STD 1.3624 × 1010 2.1840 × 109 3.4700 × 1010 1.7275 × 1010 4.8209 × 109 1.5759 × 1010 1.6158 × 1011 1.6019 × 109
AVG 8.6463 × 10 1.3452 × 102 1.0931 × 103 2.7232 × 10 2.6338 × 10 1.9322 × 10 7.5171 × 104 2.5057 × 10
F2
STD 3.8929 × 10 1.2929 × 102 4.4152 × 102 5.3829 × 10 4.5771 2.7313 1.2393 × 104 1.0313 × 10
AVG 1.2702 × 10 1.2703 × 10 1.2703 × 10 1.2703 × 10 1.2702 × 10 1.2702 × 10 1.2705 × 10 1.2702 × 10
F3
STD 9.0900 × 10−6 2.0974 × 10−4 1.3621 × 10−4 7.5574 × 10−4 6.5941 × 10−6 1.9165 × 10−5 3.1064 × 10−3 4.5257 × 10−6
AVG 2.7441 × 102 4.0491 × 103 2.5733 × 103 1.2030 × 103 8.4969 × 10 8.4365 × 10 1.9674 × 103 2.1898 × 102
F4
STD 7.4444 × 10 2.1316 × 103 7.7972 × 102 6.6790 × 102 1.3287 × 10 2.5848 × 10 1.0480 × 104 1.6873 × 102
AVG 1.7300 2.5679 2.5137 1.6473 1.6712 1.5414 1.0218 1.5682
F5
STD 1.3887 × 10−1 3.1217 × 10−1 1.8449 × 10−1 4.3138 × 10−1 1.0437 × 10−1 1.1315 × 10−1 1.1879 × 10−1 8.3493 × 10−2
AVG 8.6195 1.1608 × 10 9.4992 8.5524 1.0043 × 10 9.0537 1.3562 × 10 8.5392
F6
STD 6.7997 × 10−1 6.7551 × 10−1 8.8662 × 10−1 1.1533 7.6363 × 10−1 6.5637 × 10−1 6.6821 × 10−1 4.8619 × 10−1
AVG 2.9658 × 102 9.4097 × 102 5.4830 × 102 3.2512 × 102 5.8042 × 102 4.1793 × 102 1.0757 × 103 2.8929 × 102
F7
STD 1.0557 × 102 1.8477 × 102 1.2762 × 102 2.8856 × 102 1.4751 × 102 9.2663 × 10 2.1772 × 102 9.4926 × 10
AVG 5.4627 6.5042 5.9984 5.4877 5.9982 5.6098 6.4208 5.3501
F8
STD 3.6571 × 10−1 2.7377 × 10−1 2.1249 × 10−1 6.8809 × 10−1 3.0162 × 10−1 4.7670 × 10−1 1.6190 2.4341 × 10−1
AVG 5.0168 3.3530 × 102 5.5962 × 102 3.4069 × 10 3.3978 3.0946 2.5553 4.5696
F9
STD 1.8876 1.8130 × 102 2.1148 × 102 4.0954 × 10 2.6188 × 10−1 3.1284 × 10−1 9.5374 × 10−2 6.2230
AVG 2.0021 × 10 2.0499 × 10 2.0311 × 10 2.0041 × 10 2.0314 × 10 2.0059 × 10 2.1022 × 10 1.9994 × 10
F10
STD 7.7741 × 10−1 1.8716 × 10−1 6.6710 × 10−2 5.4341 × 10−2 1.8459 × 10−1 5.6194 × 10−1 8.8160 × 10−2 7.1465 × 10−1

Table 4. Wilcoxon rank-sum test of BROMLDE and its competitors on F1–F10 of CEC 2019 bench-
mark functions.

BSDE OMLDE WDE PSO SHADE JADE CMAES

F1 + = + + + + +

F2 + + + + + − +
Entropy 2022, 24, 1205 13 of 27

F3 + + + = + + +

F4 + + + + − − +

F5 + + + = + = −

F6 = + + = + + +

F7 = + + = + + +

F8 = + + = + + +

F9 + + + + = − −

F10 = + + + + = +

Statistics
Number 6/0/4 9/0/1 10/0/0 5/0/5 8/1/1 5/3/2 8/2/0
(+/−/=)
Entropy 2022, 24, 1205 14 of 27

Figure 2. Convergence graph of BROMLDE and its competitors on CEC 2019.

4.2.2. Experimental Results for CEC 2020

The optimization results of BROMLDE and other algorithms, including BSDE [19],
OMLDE [32], PSO, SA, IMODE [46], and J2020 [47], to solve the CEC 2020 benchmark
functions in dimensions 5 and 10 are compared in this subsection.
As shown in Table 2, those benchmark functions can be generally divided into the
unimodal function (F1), multimodal shifted and rotated functions (F2–F4), hybrid func-
tions (F5–F7), and composition functions (F8–F10). Based on the experimental setting rules
in Section 4.1, the AVG, STD and Wilcoxon rank-sum test results of the test functions de-
rived from the proposed BROMLDE and other test approaches are calculated by using
F1–F10 when the problem dimension D is equal to 5 and 10 are displayed in Tables 5–8.
The smallest average of each function in the table is marked in bold font. Additionally,
according to the statistical results, the Wilcoxon rank-sum test is employed to determine
the differences between BROMLDE and its competitors (+, =, and − denote that BROMLDE
performs better, equal, and worse than its competitors, respectively).
For the 5-dimensional test problems, based on Table 5, one can see that BROMLDE
exhibits outstanding performance compared to other tested methods. Among the 10 CEC
2020 test functions, BROMLDE yields five minimum fitness value results, namely two
multimodal shifted and rotated functions F3 and F4, two hybrid functions F5 and F6, and
one composition function F9. This shows that BROMLDE is more dominant in solving
multimodal and hybrid function problems than the compared algorithm. Furthermore,
from Table 6 (last line), BROMLDE is superior to BSDE, OMLDE, WDE, PSO, SA, J2020,
Entropy 2022, 24, 1205 15 of 27

and IMODE by 7, 10, 10, 6, 9, 5, and 4 cases, respectively, among the 10 functions. At this
time, the mean good rate of BROMLDE reaches 72.86%  (i71+i ) (10  7)  100% . That is to say,
those results significantly exceed the inferior functions.
BROMLDE still shows excellent performance among all these methods when the di-
mension of the problem is increased to 10. The results are provided in Tables 7 and 8. In
detail, the developed BROMLDE is still the winner compared to OMLDE, WDE, PSO, SA,
and IMODE on more than 5 functions based on the Wilcoxon rank-sum test findings in
Table 8 (last line). The percentage of the goodness of BROMLDE on 10 functions is 77.14%.
For the AVG and STD of the fitness values in Table 7, we can see that BROMLDE also
maintains the highest ranking with more than 4 best results. Meanwhile, BROMLDE still
has great potential in solving hybrid function problems. Based on the analysis of the above
findings, we can summarize that BROMLDE performs excellently compared to other
tested algorithms.

Table 5. Results of BROMLDE and other methods for solving 5D functions.

No. Metric BSDE OMLDE WDE PSO SA J2020 IMODE BROMLDE

AVG 4.8740 × 104 2.0986 × 107 8.5297 × 106 4.4365 × 103 4.2100 × 103 3.4065 × 105 8.9505 × 102 9.0706 × 103
F1
STD 5.0069 × 104 2.2874 × 107 4.9708 × 106 3.6216 × 103 4.1008 × 103 1.2306 × 106 4.9379 × 102 1.4826 × 104
AVG 1.2414 × 103 1.5199 × 103 1.3496 × 103 1.4760 × 103 1.5509 × 103 1.3121 × 103 1.2256 × 103 1.2381 × 103
F2
STD 5.6217 × 10 9.2984 × 10 9.9400 × 10 1.7947 × 102 1.9126 × 102 1.2762 × 102 5.5404 × 10 6.2432 × 10
AVG 7.0965 × 102 7.2101 × 102 7.2148 × 102 7.1313 × 102 7.3505 × 102 7.1375 × 102 7.0892 × 102 7.0783 × 102
F3
STD 2.0255 5.7317 4.0302 4.8783 1.7421 × 10 4.2279 1.1038 1.1270
AVG 1.9007 × 103 1.9047 × 103 1.9022 × 103 1.9005 × 103 1.9053 × 103 1.9016 × 103 1.9006 × 103 1.9004 × 103
F4
STD 3.3877 × 10−1 2.6435 5.6901 × 10−1 3.6936 × 10−1 3.7258 7.2313 × 10−1 1.7476 × 10−1 1.1759 × 10−1
AVG 1.7167 × 103 3.0457 × 103 1.9032 × 103 5.5299 × 103 2.3202 × 10 3.3174 × 103 1.7070 × 103
3 1.7070 × 103
F5
STD 7.5758 1.0090 × 103 1.2768 × 102 5.2416 × 103 1.1974 × 103 7.0944 × 103 2.8809 1.1347 × 10
AVG 1.6010 × 103 1.6136 × 103 1.6035 × 103 1.6251 × 103 1.7276 × 103 1.6009 × 103 1.6010 × 103 1.6008 × 103
F6
STD 2.6168 × 10−1 8.6732 2.0798 3.6268 × 10 9.1687 × 10 6.0805 × 10−1 2.2067 × 10−1 1.9990 × 10−1
AVG 2.1005 × 103 2.1044 × 103 2.1015 × 103 2.1120 × 103 2.1234 × 103 2.1001 × 103 2.1001 × 103 2.1003 × 103
F7
STD 1.9307 × 10−1 2.9249 5.1746 × 10−1 1.5631 × 10 3.1755 × 10 1.9415 × 10−1 5.7708 × 10−2 1.5899 × 10−1
AVG 2.2159 × 103 2.2379 × 103 2.2268 × 103 2.2530 × 103 2.5333 × 103 2.2186 × 103 2.2046 × 103 2.2066 × 103
F8
STD 9.9944 1.0055 × 10 7.1466 4.7534 × 10 3.6440 × 102 1.8780 × 10 3.0763 6.0099
AVG 2.5034 × 103 2.5307 × 103 2.5299 × 103 2.5654 × 103 2.7298 × 103 2.5064 × 103 2.5033 × 103 2.4993 × 103
F9
STD 1.9814 × 10 1.1469 × 10 1.2249 × 10 1.0929 × 102 9.8491 × 10 3.7058 × 10 1.8782 2.3732 × 10
AVG 2.8209 × 103 2.8511 × 103 2.8473 × 103 2.8470 × 103 2.8626 × 103 2.7933 × 103 2.8327 × 103 2.8460 × 103
F10
STD 5.4998 × 10 6.6783 9.4384 1.0499 × 10 7.4130 × 10 7.0970 × 10 5.8544 × 10 4.3048

Table 6. Wilcoxon rank-sum test of BROMLDE and its competitors on F1–F10 of CEC 2020 bench-
mark functions ( 5D ).

BSDE OMLDE WDE PSO SA J2020 IMODE

F1 + + + = = = −
F2 = + + + + + =
F3 + + + + + + +
F4 + + + = + + +
F5 + + + + + + +
F6 + + + + + = +
F7 + + + + + − −
F8 + + + + + + =
F9 = + + = + = =
F10 = + + = + − −
Statistics Number
7/0/3 10/0/0 10/0/0 6/0/4 9/0/1 5/2/3 4/3/3
(+/−/=)
Entropy 2022, 24, 1205 16 of 27

Table 7. Results of BROMLDE and other methods for solving 10D functions.

No. Metric BSDE OMLDE WDE PSO SA J2020 IMODE BROMLDE

AVG 1.4476 × 108 2.4555 × 109 1.5449 × 109 4.0704 × 108 4.6326 × 103 3.6412 × 107 1.7090 × 107 4.4096 × 107
F1
STD 8.3021 × 107 1.3660 × 109 4.9592 × 108 5.1296 × 108 3.7971 × 103 4.3766 × 107 6.3048 × 106 7.2445 × 107
AVG 1.9834 × 103 2.7348 × 103 2.3048 × 103 2.1206 × 103 2.1731 × 103 1.6655 × 103 1.9956 × 103 1.8547 × 103
F2
STD 1.3991 × 102 1.4098 × 102 1.1851 × 102 3.6835 × 102 2.8766 × 102 2.7621 × 102 1.3242 × 102 1.5977 × 102
AVG 7.4407 × 102 7.9016 × 102 8.2616 × 102 7.5025 × 102 7.9052 × 102 7.5559 × 102 7.4235 × 102 7.3110 × 102
F3
STD 5.7745 1.6326 × 10 1.8826 × 10 1.6186 × 10 4.1993 × 10 1.3715 × 10 4.9878 6.5706
AVG 1.9460 × 103 6.6667 × 103 2.1915 × 103 2.0936 × 103 1.9091 × 103 1.9059 × 103 1.9041 × 103 1.9299 × 103
F4
STD 2.6716 × 10 9.4365 × 103 2.5856 × 102 5.5618 × 102 5.2152 1.9810 6.3848 × 10−1 4.7804 × 10
AVG 9.7848 × 104 1.5909 × 105 9.7802 × 104 1.9558 × 105 9.7668 × 105 4.4576 × 104 2.1550 × 104 5.2912 × 104
F5
STD 6.6057 × 104 9.6210 × 104 7.5923 × 104 3.3832 × 105 9.0480 × 105 1.3231 × 105 1.5227 × 104 8.1556 × 104
AVG 1.7118 × 103 2.0008 × 103 1.7653 × 103 1.9071 × 103 1.9416 × 103 1.6834 × 103 1.6963 × 103 1.6636 × 103
F6
STD 4.5849 × 10 1.0720 × 102 5.6569 × 10 1.3882 × 102 1.3931 × 102 7.2439 × 10 5.4220 × 10 4.6814 × 10
AVG 8.9530 × 103 2.0206 × 104 1.5814 × 104 9.5733 × 103 2.0077 × 106 1.2221 × 104 4.3612 × 103 4.3279 × 103
F7
STD 5.6729 × 103 2.1511 × 104 8.8941 × 103 8.1418 × 103 2.5879 × 106 3.4438 × 104 1.3724 × 103 1.7359 × 103
AVG 2.3216 × 103 2.5316 × 103 2.4407 × 103 2.3360 × 103 3.1952 × 103 2.3166 × 103 2.3118 × 103 2.3100 × 103
F8
STD 1.9100 × 10 1.0034 × 102 7.0259 × 10 2.0004 × 10 7.9140 × 102 8.2685 5.9606 9.0332
AVG 2.6409 × 103 2.7814 × 103 2.7274 × 103 2.7447 × 103 2.8628 × 103 2.7239 × 103 2.7013 × 103 2.6473 × 103
F9
STD 5.6065 × 10 6.7746 × 10 4.3935 × 10 1.1248 × 102 1.0451 × 102 6.2291 × 10 5.6735 × 10 6.7168 × 10
AVG 2.9502 × 103 3.0807 × 103 3.0229 × 103 2.9518 × 103 3.0048 × 103 2.9424 × 103 2.9285 × 103 2.9448 × 103
F10
STD 8.0811 8.1623 × 10 2.8419 × 10 2.2054 × 10 3.8234 × 10 1.8236 × 10 9.5728 1.0122 × 10

Table 8. Wilcoxon rank-sum test of BROMLDE and its competitors on F1–F10 of CEC 2020 bench-
mark functions ( 10D ).

BSDE OMLDE WDE PSO SA J2020 IMODE

F1 + + + + − = −
F2 + + + + + − +
F3 + + + + + + +
F4 + + + + = − −
F5 + + + = + − =
F6 + + + + + = +
F7 + + + + + + =
F8 + + + + + + +
F9 = + + + + + +
F10 + + + = + = −
Statistics Number
9/0/1 10/0/0 10/0/0 8/0/2 8/1/1 4/3/3 5/3/2
(+/−/=)

The convergence plots of BROMLDE and other compared approaches on the 5 and
10-dimensional CEC 2020 functions (F1–F10) are presented in Figures 3 and 4, respec-
tively. In those plots, the vertical axis is the logarithm of the minimum value of the func-
tions and the horizontal axis is the functional evaluation numbers. From those figures, we
can see that although PSO and SA converge faster on some functions, the local optimum
situation may occur. Compared with other algorithms, BROMLDE has a faster descent
speed and better optimization capability in most functions. The reason is that the combi-
nation of the Bernstein search and the ROML strategy allows BROMLDE to reach a better
trade-off between global exploration and local exploitation capabilities. In the early stage,
the ROML strategy can provide strong search abilities and helps to localize the exact
search in the late stage. Moreover, Bernstein search may reduce the difficulty of parameter
setting and improve the convergence accuracy. It reveals that our proposed BROMLDE
can reach better convergence properties and global optimization ability.
Entropy 2022, 24, 1205 17 of 27
Entropy 2022, 24, 1205 18 of 27

Figure 3. Convergence graphs of F1–F10 in 5D .

Entropy 2022, 24, 1205 19 of 27

Figure 4. Convergence graphs of F1–F10 in 10D .

4.2.3. Comparison with IMODE

In this subsection, on the basis of Section 4.1, the 20-dimensional CEC 2020 test func-
tions (F1–F10) are used to compare the performance of BROMLDE and IMODE, the AVG
and STD results are presented in Table 9, and the Wilcoxon rank-sum test findings are
also embedded in it. One can see that IMODE has a greater advantage in solving unimodal
function (F1) and multimodal shifted and rotated functions except for F3. BROMLDE is
superior to IMODE in solving hybrid functions F5–F7. In addition, BROMLDE outper-
forms IMODE in solving the composition functions F8 and F10 and is worse in solving F9
of composition functions. The Wilcoxon rank-sum test results (+, =, and − denote that
BROMLDE performs better, equal, and worse than IMODE, respectively) display that the
average good rate of BROMLDE is 25% ( statistics number
4
:
 100% ) for the unimodal function and

the multimodal functions, is 66.67% ( statistics number

3
:
 100% ) for hybrid functions, and is 33.33%

( statistics number
3
:
 100% ) for composition functions. Those results also reflect the no free lunch

theorem that no single algorithm can be applied to all optimization problems [33]. Based
on the above analysis, one can conclude that BROMLDE is more appropriate for solving
hybrid function problems and it performs worse in solving unimodal functions and the
multimodal shifted and rotated functions as well as composition functions.

Table 9. Results of BROMLDE and IMODE on F1–F10 of CEC 2020 benchmark functions ( 20D ).

No. Metric IMODE BROMLDE

Unimodal Function/Multimodal Shifted and Rotated Functions
AVG 7.7793 × 108 8.0666 × 108
F1 =
STD 2.0501 × 108 4.7072 × 108
AVG 3.5623 × 103 3.9156 × 103
F2 −
STD 2.4748 × 102 2.7165 × 102
AVG 8.3820 × 102 8.2117 × 102
F3 +
STD 1.1526 × 10 1.8004 × 10
AVG 1.9284 × 103 2.1141 × 103
F4 −
STD 6.3149 5.7483 × 102
Hybrid Functions
AVG 7.9308 × 105 6.3443 × 105
F5 +
STD 2.7024 × 105 3.6768 × 105
AVG 2.0119 × 103 1.9989 × 103
F6 =
STD 8.9159 × 10 9.4389 × 10
Entropy 2022, 24, 1205 20 of 27

AVG 4.6835 × 105 2.4124 × 105

F7 +
STD 2.5776 × 105 1.4449 × 105
Composition Functions
AVG 2.7476 × 103 2.5188 × 103
F8 +
STD 1.5594 × 102 1.2362 × 102
AVG 2.9273 × 103 2.9362 × 103
F9 −
STD 1.2475 × 10 1.4749 × 10
AVG 3.0580 × 103 3.0562 × 103
F10 =
STD 2.4327 × 10 2.6982 × 10
Total Statistics (+/−/=) 4/3/3

4.3. Real-World Engineering Optimization Problems

To further verify the feasibility of our proposed BROMLDE in practical engineering
applications, for the solution of the car side impact (CSI) design problem and the speed
reducer (SR) design problem, BROMLDE, some DE variants (BSDE, OMLDE, and WDE)
and the superior algorithms (PSO and SNS) for solving these problems are used.

4.3.1. CSI Design Problem

The goal of the CSI design problem is to obtain the minimum weight of the door
satisfying 10 constraints on 11 influence variables [6]. Those variables are listed in Table
10. The authors in [49] simplify the analytical formulation of this optimization problem.
Figure 5 [6] shows a model for the CSI design problem. Then, the objective function of this
design problem is Equation (35):
min f ( x)  1.98  4.90x1  6.67x2  6.98x3  4.01x4  1.78x5  2.73x7 (35)

subject to:
1 ( x)  1.16  0.3717 x2 x4  0.00931x2 x10  0.484 x3 x9  0.01343 x6 x10  1  0,
 ( x)  46.36  9.9 x  12.9 x x  0.1107 x x  32  0,
 2 2 1 2 3 10

 3 ( x)  33.86  2.95 x3  0.1792 x3  5.057 x1 x2  11.0 x2 x8  0.0215 x5 x10  9.98 x7 x8


  22.0 x8 x9  32  0,
 ( x)  28.98  3.818 x  4.2 x x  0.0207 x x  6.63 x x  7.7 x x  0.32 x x  32  0,
 4 3 1 2 5 10 6 9 7 8 9 10

 5 ( x) = 0.261  0.0159 x1x2  0.188 x1x8  0.019 x2 x7  0.0144 x3 x5  0.0008757 x5 x10

  0.08045 x6 x9  0.00139 x8 x11  0.00001575 x10 x11  0.32  0,

 ( x) = 0.214  0.00817 x  0.131x x  0.0704 x x  0.03099 x x  0.018 x x  0.0208 x x (36)
 6 5 1 8 1 9 2 6 2 7 3 8

  0.121x3 x9  0.00364 x5 x6  0.0007715 x5 x10  0.0005354 x6 x10  0.00121x8 x11


  0.00184 x9 x10  0.02 x22  0.32  0,
 ( x) = 0.74  0.61x  0.163 x x  0.001232 x x  0.166 x x  0.227 x 2  0.32  0,
 7 2 3 8 3 10 7 9 2

 8 ( x) = 4.72  0.5 x4  0.19 x2 x3  0.0122 x4 x10  0.009325 x6 x10  0.000191x112  4  0,


 9 ( x) = 10.58  0.674 x1 x2  1.95 x2 x8  0.02054 x3 x10  0.0198 x4 x10  0.028 x6 x10  9.9  0,
 ( x ) = 16.45  0.489 x x  0.843 x x  0.0432 x x  0.0556 x x  0.000786 x 2  15.7  0,
 10 3 7 5 6 9 10 9 11 11

in which 0.5  xi  1.5, i  1,...,7, x8 , x9  0.192,0.345 and 30  x10 , x11  30.
Entropy 2022, 24, 1205 21 of 27

Figure 5. A model of CSI design problem [6].

Table 10. Influence parameters of the weight of the door.

No. Variables Description of Variables

1 x1 Thicknesses of B-pillar Inner
2 x2 B-pillar Reinforcement
3 x3 Floor Side Inner
4 x4 Cross Members
5 x5 Door Beam
6 x6 Door Beltline Reinforcement
7 x7 Roof Rail
8 x8 Materials of B-pillar Inner
9 x9 Floor Side Inner
10 x10 Barrier Height
11 x11 Hitting Position

Table 11 concludes the best and workable experimental results of BROMLDE and its
competitors after 30 independent runs (where N p  100, MaxFES  15000 ). From Table 11,
BROMLDE can get the optimal objective function value, i.e., 22.2372. Furthermore, our
proposed BROMLDE has the smallest AVG and STD, and these are 2.2463×10 and
1.4463×10-1, respectively, compared to other algorithms tested here. Moreover, based on
the Wilcoxon rank-sum test results (+, =, and − denote that BROMLDE performs better,
equal, and worse than the compared algorithm, respectively), we can obtain that BROM-
LDE is better than BSDE, OMLDE, WDE, PSO, and SNS, respectively. It can be concluded
that the effectiveness and reliability of the proposed BROMLDE are superior to other
tested algorithms.

Table 11. Comparison results of the BROMLDE and its competitors for CSI design problem.

BSDE OMLDE WDE PSO SNS BROMLDE

x1 0.5026 0.5943 0.5213 0.5000 0.5400 0.5042
x2 1.0778 0.9961 1.0271 1.0302 0.9027 0.9831
x3 0.5219 0.5475 0.5115 0.5000 0.5400 0.5178
x4 1.2211 1.2961 1.2723 1.2807 1.3599 1.3132
x5 0.5648 0.5011 0.5068 0.5724 0.5400 0.5121
x6 1.1337 1.3552 1.4657 1.5000 0.5400 1.4363
x7 0.5311 0.5754 0.5645 0.5532 0.5400 0.5266
x8 0.1920 0.1920 0.1920 0.1920 0.3450 0.1920
x9 0.1920 0.1920 0.1920 0.1920 0.3450 0.1920
Entropy 2022, 24, 1205 22 of 27

x10 3.3461 −4.9938 −3.5587 1.2682 5.9180 −11.1769

x11 7.0345 4.9287 −3.3178 7.0414 20.8016 2.7355
1 ( x) −0.3603 −0.4153 −0.4093 −0.3635 −0.3933 −0.4813
2 ( x) −3.1044 −3.4401 −2.9162 −2.4132 −0.5110 −2.4075
 3 ( x) −1.7696 −1.7616 −1.6480 −1.6206 −1.6504 −1.1776
4 ( x) −2.3997 −2.9000 −2.5406 −2.0899 −2.4854 −2.8803
5 ( x) −0.0710 −0.0771 −0.0733 −0.0661 −0.0771 −0.0737
6 ( x) −0.1024 −0.0997 −0.0963 −0.0921 −0.1105 −0.0939
7 ( x) −0.0049 −0.0012 −0.0033 −1.6653 × 10−16 −0.0030 −4.1458 × 10−4
8 ( x) −0.0025 −0.0112 −0.0073 −0.0108 −0.0383 −0.0025
9 ( x) −0.0274 −0.2094 −0.1592 −0.0187 −0.2601 −0.2999
10 ( x) −0.0116 −0.0897 −0.0201 −0.2127 −0.2893 −0.1312
Optimal f ( x) 22.6264 23.0179 22.5002 22.4562 22.3046 22.2372
AVG 2.3078 × 10 2.4141 × 10 2.3283 × 10 2.4435 × 10 2.3069 × 10 2.2463 × 10
STD 2.3420 × 10−1 4.8138 × 10−1 3.9897 × 10−1 1.0472 4.4591 × 10−1 1.4463 × 10−1
+/−/= + + + + +

4.3.2. SR Design Problem

SR design problem (see Figure 6) aims to design the speed reducer subject to 11 con-
straints. One has 7 variables (see Table 12). The mathematical expression of this problem
is given in Equation (37) [6]:

min f ( x)  0.7854  x1  x22   3.3333  x32  14.9334  x3  43.0934   1.508  x1   x62  x72 
(37)
7.4777   x62  x72   0.78054   x4  x62  x5  x72  ,

subject to:
Entropy 2022, 24, 1205 23 of 27

27
1 ( x)   1  0,
x1  x22  x3
397.5
 2 ( x)   1  0,
x1  x22  x32
1.93  x43
 3 ( x)   1  0,
x2  x3  x64
1.93  x53
 4 ( x)   1  0,
x2  x3  x74
2
 745.0  x4  6 3
 5 ( x)     16.9 10 (110 x6 )  1  0,
x 
 2 3  x
2
 745.0  x5  6 3 (38)
 6 ( x)     157.5 10 (85x7 )  1  0,
x
 2 3  x
x x
 7 ( x)  2 3  1  0,
40
5x
8 ( x)  2  1  0,
x1
x
9 ( x)  1  1  0,
12 x2
1.5 x6  1.9
10 ( x)   1  0,
x4
1.1x7  1.9
11 ( x)   1  0,
x5

in which the bounds are as follows:

2.6  x1  3.6
0.7  x  0.8
 2

17  x3  28

7.3  x4  8.3 (39)
7.3  x  8.3
5

2.9  x6  3.9

5.0  x7  5.5

Figure 6. Schematic diagram of the SR [6].

Entropy 2022, 24, 1205 24 of 27

Table 12. Influence variables of the SR problem.

No. Variables Descriptions

1 b(  x1 ) Face Width
2 m( x2 ) Module of Teeth
3 z (= x3 ) The Number of Teeth in the Pinion
4 l1 ( x4 ) Length of the First Shaft Between Bearings
5 l2 ( x5 ) Length of the Second Shaft Between Bearings
6 d1 ( x6 ) The Diameter of First Shafts
7 d2 (= x7 ) The Diameter of Second Shafts

Table 13 records the optimal and feasible experimental results of BROMLDE and the
tested algorithms after 30 independent runs (where N p  100, MaxFES  15000 ). According to
Table 13, BROMLDE can obtain the optimal fitness value, i.e., 5.4421×103. Meanwhile, the
proposed BROMLDE has the minimum AVG (5.4660 × 103) compared with other compet-
itors. Through the Wilcoxon rank-sum test results (+, =, and − denote that BROMLDE per-
forms better, equal, and worse than the compared algorithm, respectively), we can ob-
serve that BROMLDE is still better than most algorithms. This further shows that our pro-
posed algorithm is workable in solving practical problems.

Table 13. Comparison results of the BROMLDE and its competitors for SR design problem.

BSDE OMLDE WDE PSO SNS BROMLDE

x1 3.5031 3.5183 3.5027 3.5050 3.5122 3.5022
x2 0.7003 0.7000 0.7000 0.7000 0.7010 0.7001
x3 28 28 28 28 28 28
x4 7.4203 8.0487 7.3417 7.8632 7.3010 7.3106
x5 7.7615 7.7408 7.7770 7.7717 7.7685 7.7838
x6 3.3619 3.3534 3.3479 3.3501 3.3599 3.3464
x7 5.2912 5.2889 5.3032 5.2858 5.2878 5.2864
1 ( x) −21.1076 −21.2716 −21.0613 −21.0881 −21.3246 −21.0638
2 ( x) −949.5132 −954.1054 −948.2157 −948.9666 −955.5880 −948.2858
 3 ( x) −4.2014 −2.8237 −4.2928 −3.1479 −4.4976 −4.3613
4 ( x) −30.9423 −31.1344 −31.0309 −30.6659 −30.7999 −30.5336
5 ( x) −15.2784 −6.5270 −1.6458 −3.4119 −13.4128 −0.1271
6 ( x) −2.5637 −1.4703 −8.3399 −3.2401 × 10−11 −0.9238 −0.2888
7 ( x) −20.3907 −20.4000 −20.3991 −20.4000 −20.3720 −20.3972
8 ( x) −0.0020 −0.0262 −0.0036 −0.0071 −0.0102 −0.0024
9 ( x) −6.9980 −6.9738 −6.9964 −6.9929 −6.9898 −6.9976
10 ( x) −0.4774 −1.1185 −0.4198 −0.9381 −0.3611 −0.3911
11 ( x) −0.0412 −0.0230 −0.0435 −0.0573 −0.0520 −0.0687
Optimal f ( x ) 5.4532 × 103 5.4675 × 103 5.4531 × 103 5.4491 × 103 5.4672 × 103 5.4421 × 103
AVG 5.4666 × 103 5.5240 × 103 5.4708 × 103 5.5495 × 103 5.4829 × 103 5.4660 × 103
STD 8.1468 3.4116 × 10 9.5649 8.1021 × 10 5.6624 1.7858 × 10
+/−/= = + = + +

5. Conclusions and Future Work

In this paper, a DE algorithm based on the Bernstein operator and refracted opposi-
tional-mutual learning strategy is proposed to enhance the optimization effect of the al-
gorithm. More specifically, a random switching scheme allows the selection of ROL and
ML for all individuals in the ROML initialization phase and the ROML generation
Entropy 2022, 24, 1205 25 of 27

jumping phase, which helps to balance the exploration and exploitation. A dynamic ad-
justment factor varying with the number of evaluations in the ROML strategy is proposed,
contributing to the tuning of the search space and jumping out of the local optimum.
Moreover, a Bernstein operator is introduced to control the mutation and crossover
phases, improving the convergence accuracy of the algorithm, and making it more effi-
cient. Experiments are performed on CEC 2019 and CEC 2020 benchmark functions, and
the experimental results show that the proposed BROMLDE outperforms the compared
algorithm. Meanwhile, the Wilcoxon rank-sum test and convergence analysis reveal that
the BROMLDE is considerably better than other tested algorithms. Particularly, BROM-
LDE is superior to IMODE in solving hybrid function problems from CEC 2020 (20D).
Additionally, BROMLDE and the tested algorithms (BSDE, OMLDE, WDE, PSO, and
SNS) are used on a practical engineering problem, and the result further verifies the ap-
plicability of the algorithm in solving real-life engineering issues. Therefore, it is worth
recommending ROML strategies to other algorithms to enhance their performance. How-
ever, when BROMLDE is compared with IMODE for the CEC 2020 test functions, BROM-
LDE is only superior in solving hybrid function problems and performs inferiorly in other
problems. Given this, we will explore new learning strategies to further improve the per-
formance of the algorithm in future research work.

Author Contributions: Conceptualization, F.W. and J.Z.; methodology, F.W. and J.Z.; software,
F.W. and D.L.; validation, M.L. and D.L.; formal analysis, J.Z.; investigation, F.W. and M.L.; re-
sources, S.L.; writing—original draft preparation, F.W.; writing—review and editing, F.W.; super-
vision, J.Z.; project administration, S.L.; funding acquisition, S.L., J.Z. and F.W. All authors have
read and agreed to the published version of the manuscript.
Funding: This work is supported in part by the National Key Technologies R&D Program of China
(2018AAA0101803 and 32020YFB1713300), in part by the Natural Science Foundation of China
(51635003 and 61863005), in part by the Guizhou Province Postgraduate Innovation Fund
(YJSKYJJ(2021)030), in part by Guizhou Provincial Science and Technology Projects(ZK [2022] 142),
in part by the Foundation of Key Laboratory of Advanced Manufacturing Technology, Ministry of
Education, Guizhou University (GZUAMT2021KF [11]), in part by the Science and Technology In-
cubation Planning Project of Guizhou University ([2020]75), and in part by the Key Laboratory of
Ministry of Education Project (QKHKY [2020]245).
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Song, B.; Wang, Z.; Zou, L. On Global Smooth Path Planning for Mobile Robots Using a Novel Multimodal Delayed PSO
Algorithm. Cognit. Comput. 2017, 9, 5–17.
2. Hu, W.B.; Liang, H.L.; Peng, C.; Du, B.; Hu, Q. A Hybrid Chaos-Particle Swarm Optimization Algorithm for the Vehicle Routing
Problem with Time Window. Entropy 2013, 15, 1247–1270.
3. Abd Elaziz, M.; Xiong, S.; Jayasena, K.P.N.; Li, L. Task Scheduling in Cloud Computing Based on Hybrid Moth Search
Algorithm and Differential Evolution. Knowl. Based Syst. 2019, 169, 39–52.
4. Chen, M.N.; Zhou, Y.Q.; Luo, Q.F. An Improved Arithmetic Optimization Algorithm for Numerical Optimization Problems.
Mathematics 2022, 10, 2152.
5. Deng, W.; Xu, J.J.; Song, Y.J.; Zhao, H.M. Differential Evolution Algorithm with Wavelet Basis Function and Optimal Mutation
Strategy for Complex Optimization Problem. Appl. Soft Comput. 2021, 100, 106724.
6. Bayzidi, H.; Talatahari, S.; Saraee, M.; Lamarche, C.P. Social Network Search for Solving Engineering Optimization Problems.
Comput. Intell. Neurosci. 2021, 2021, 8548639.
7. Storn, R.; Price, K. Differential Evolution—A Simple and Efficient Heuristic for Global Optimization over Continuous Spaces.
J. Glob. Optim. 1997, 11, 341–359.
8. Eberhart, R.; Kennedy, J. A New Optimizer Using Particle Swarm Theory. In Proceedings of the MHS’95 Sixth International
Symposium on Micro Machine and Human Science, Nagoya, Japan, 4–6 October 1995; pp. 39–43.
9. Bertsimas, D.; Tsitsiklis, J. Simulated Annealing. Stat. Sci. 1993, 8, 10–15.
10. Kwiecien, J.; Pasieka, M. Cockroach Swarm Optimization Algorithm for Travel Planning. Entropy 2017, 19, 213.
11. Bas, E. The Training of Multiplicative Neuron Model Based Artificial Neural Networks with Differential Evolution Algorithm
for Forecasting. J. Artif. Intell. Soft Comput. Res. 2016, 6, 5–11.
Entropy 2022, 24, 1205 26 of 27

12. Peng, L.; Liu, S.; Liu, R.; Wang, L. Effective Long Short-Term Memory with Differential Evolution Algorithm for Electricity Price
Prediction. Energy 2018, 162, 1301–1314.
13. Tong, B.D.; Chen, L.; Duan, H.B. A Path Planning Method for UAVs Based on Multi-Objective Pigeon-Inspired Optimisation
and Differential Evolution. Int. J. Bio-Inspired Comput. 2021, 17, 105–112.
14. Wang, Y.; Cai, Z.; Zhang, Q. Enhancing the Search Ability of Differential Evolution through Orthogonal Crossover. Inf. Sci. 2012,
185, 153–177.
15. Lieu, Q.X.; Do, D.T.T.; Lee, J. An Adaptive Hybrid Evolutionary Firefly Algorithm for Shape and Size Optimization of Truss
Structures with Frequency Constraints. Comput. Struct. 2018, 195, 99–112.
16. Huynh, T.N.; Do, D.T.T.; Lee, J. Q-Learning-Based Parameter Control in Differential Evolution for Structural Optimization. Appl.
Soft Comput. 2021, 107, 107464.
17. Pan, Q.K.; Suganthan, P.N.; Wang, L.; Gao, L.; Mallipeddi, R. A Differential Evolution Algorithm with Self-Adapting Strategy
and Control Parameters. Comput. Oper. Res. 2011, 38, 394–408.
18. Fan, Q.Q.; Yan, X.F. Self-Adaptive Differential Evolution Algorithm with Zoning Evolution of Control Parameters and Adaptive
Mutation Strategies. IEEE Trans. Cybern. 2016, 46, 219–232.
19. Civicioglu, P.; Besdok, E. Bernstain-Search Differential Evolution Algorithm for Numerical Function Optimization. Expert Syst.
Appl. 2019, 138, 112831.
20. Morales-Castañeda, B.; Zaldívar, D.; Cuevas, E.; Fausto, F.; Rodríguez, A. A Better Balance in Metaheuristic Algorithms: Does
It Exist?. Swarm Evol. Comput. 2020, 54, 100671.
21. Xiao, Y.; Sun, X.; Zhang, Y.; Guo, Y.; Wang, Y.; Li, J. An Improved Slime Mould Algorithm Based on Tent Chaotic Mapping and
Nonlinear Inertia Weight. Int. J. Innov. Comput. Inf. Control 2021, 17, 2151–2176.
22. Dinkar, S.K.; Deep, K. Opposition-Based Antlion Optimizer Using Cauchy Distribution and Its Application to Data Clustering
Problem. Neural Comput. Appl. 2020, 32, 6967–6995.
23. Yu, X.; Xu, W.; Li, C. Opposition-Based Learning Grey Wolf Optimizer for Global Optimization. Knowl. Based Syst. 2021, 226,
107139.
24. Shakya, H.K.; Singh, K.; More, Y.S.; Biswas, B. Opposition-Based Genetic Algorithm for Community Detection in Social
Networks. Proc. Natl. Acad. Sci. India Sect. A Phys. Sci. 2022, 92, 251–263.
25. Li, J.H.; Gao, Y.L.; Wang, K.G.; Sun, Y. A Dual Opposition-Based Learning for Differential Evolution with Protective Mechanism
for Engineering Optimization Problems. Appl. Soft Comput. 2021, 113, 107942.
26. Esmailzadeh, A.; Rahnamayan, S. Opposition-based differential evolution with protective generation jumping. In Proceedings
of the 2011 IEEE Symposium on Differential Evolution (SDE), Paris, France, 11–15 April 2011; pp. 1–8.
27. Xu, Q.; Wang, N.; Fei, R. Influence of Dimensionality and Population Size on Opposition-Based Differential Evolution Using
the Current Optimum. Inf. Technol. J. 2013, 12, 105.
28. Shao, P.; Yang, L.; Tan, L.; Li, G.Q.; Peng, H. Enhancing Artificial Bee Colony Algorithm Using Refraction Principle. Soft Comput.
2020, 24, 15291–15306.
29. Abed-alguni, B.H.; Alawad, N.A.; Barhoush, M.; Hammad, R. Exploratory Cuckoo Search for Solving Single-Objective
Optimization Problems. Soft Comput. 2021, 25, 10167–10180.
30. Zhao, X.C.; Feng, S.; Hao, J.L.; Zuo, X.Q.; Zhang, Y. Neighborhood Opposition-Based Differential Evolution with Gaussian
Perturbation. Soft Comput. 2021, 25, 27–46.
31. Xu, Y.; Yang, Z.; Li, X.; Kang, H.; Yang, X. Dynamic Opposite Learning Enhanced Teaching–Learning-Based Optimization.
Knowl. Based Syst. 2020, 188, 104966.
32. Xu, Y.L.; Yang, X.F.; Yang, Z.L.; Li, X.P.; Wang, P.; Ding, R.Z.; Liu, W.K. An Enhanced Differential Evolution Algorithm with a
New Oppositional-Mutual Learning Strategy. Neurocomputing 2021, 435, 162–175.
33. Wolpert, D.H.; Macready, W.G. No Free Lunch Theorems for Optimization. IEEE Trans. Evol. Comput. 1997, 1, 67–82.
34. Shao, P.; Wu, Z.J.; Zhou, X.Y.; Tran, D.C. FIR Digital Filter Design Using Improved Particle Swarm Optimization Based on
Refraction Principle. Soft Comput. 2017, 21, 2631–2642.
35. Long, W.; Wu, T.; Jiao, J.; Tang, M.; Xu, M. Refraction-Learning-Based Whale Optimization Algorithm for High-Dimensional
Problems and Parameter Estimation of PV Model. Eng. Appl. Artif. Intell. 2020, 89, 103457.
36. Azhari, F.; Heidarpour, A.; Zhao, X.-L. On the Use of Bernstain-Bézier Functions for Modelling the Post-Fire Stress-Strain
Relationship of Ultra-High Strength Steel (Grade 1200). Eng. Struct. 2018, 175, 605–616.
37. Rahnamayan, S.; Tizhoosh, H.R.; Salama, M.M.A. Opposition-Based Differential Evolution. IEEE Trans. Evol. Comput. 2008, 12,
64–79.
38. Rao, R.V.; Savsani, V.J.; Vakharia, D.P. Teaching-Learning-Based Optimization: A Novel Method for Constrained Mechanical
Design Optimization Problems. Comput. Des. 2011, 43, 303–315.
39. Cheng, G.; Guohui, Z.; Bo, H.; Jin, L. HHO Algorithm Combining Mutualism and Lens Imaging Learning. Comput. Eng. Appl.
2022, 58, 76–86.
40. Price, K.V.; Awad, N.H.; Ali, M.Z.; Suganthan, P.N. Problem Definitions and Evaluation Criteria for the 100-Digit Challenge
Special Session and Competition on Single Objective Numerical Optimization. In Technical Report; Nanyang Technological
University Singapore: Singapore, 2018.
41. Yue, C.T.; Price, K.V.; Suganthan, P.N.; Liang, J.J.; Ali, M.Z.; Qu, B.Y.; Awad, N.H.; Biswas, P.P. Problem Definitions and
Evaluation Criteria for the CEC 2020 Special Session and Competition on Single Objective Bound Constrained Numerical
Entropy 2022, 24, 1205 27 of 27

Optimization. Comput. Intell. Lab. Zhengzhou Univ. Zhengzhou China Tech. Rep. Nov. 2019. Available: https://github.com/P-N-
Suganthan/2020-Bound-Constrained-Opt-Benchmark (accessed on 25 July 2022).
42. Civicioglu, P.; Besdok, E.; Gunen, M.A.; Atasever, U.H. Weighted Differential Evolution Algorithm for Numerical Function
Optimization: A Comparative Study with Cuckoo Search, Artificial Bee Colony, Adaptive Differential Evolution, and
Backtracking Search Optimization Algorithms. Neural Comput. Appl. 2020, 32, 3923–3937.
43. Zhang, J.Q.; Sanderson, A.C. JADE: Adaptive Differential Evolution with Optional External Archive. IEEE Trans. Evol. Comput.
2009, 13, 945–958.
44. Tanabe, R.; Fukunaga, A. Success-History Based Parameter Adaptation for Differential Evolution. In Proceedings of the 2013
IEEE Congress on Evolutionary Computation, Cancun, Mexico, 20–23 June 2013; pp. 71–78.
45. Hansen, N.; Ostermeier, A. Completely Derandomized Self-Adaptation in Evolution Strategies. Evol. Comput. 2001, 9, 159–195.
46. Sallam, K.M.; Elsayed, S.M.; Chakrabortty, R.K.; Ryan, M.J. Improved Multi-Operator Differential Evolution Algorithm for
Solving Unconstrained Problems. In Proceedings of the 2020 IEEE Congress on Evolutionary Computation (CEC), Glasgow,
UK, 19–24 July 2020; pp. 1–8.
47. Brest, J.; Maučec, M.S.; Bošković, B. Differential Evolution Algorithm for Single Objective Bound-Constrained Optimization:
Algorithm J2020. In Proceedings of the 2020 IEEE Congress on Evolutionary Computation (CEC), Glasgow, UK, 19–24 July 2020;
pp. 1–8.
48. Tian, Y.; Cheng, R.; Zhang, X.Y.; Jin, Y.C. PlatEMO: A MATLAB Platform for Evolutionary Multi-Objective Optimization. IEEE
Comput. Intell. Mag. 2017, 12, 73–87.
49. Youn, B.D.; Choi, K.K. A New Response Surface Methodology for Reliability-Based Design Optimization. Comput. Struct. 2004,
82, 241–256.