A Dynamic Adaptive Particle Swarm Optimization and

Advances in Nonlinear Dynamics and Vibrations on Mechanical Systems - Research Article
Advances in Mechanical Engineering

2019, Vol. 11(3) 1–27
Ó The Author(s) 2019
A dynamic adaptive particle swarm DOI: 10.1177/1687814018824930
journals.sagepub.com/home/ade
optimization and genetic algorithm for
different constrained engineering
design optimization problems
Hao Zhu1,2 , Yumei Hu3 and Weidong Zhu4
Abstract
A dynamic adaptive particle swarm optimization and genetic algorithm is presented to solve constrained engineering
optimization problems. A dynamic adaptive inertia factor is introduced in the basic particle swarm optimization algo-
rithm to balance the convergence rate and global optima search ability by adaptively adjusting searching velocity during
search process. Genetic algorithm–related operators including a selection operator with time-varying selection probabil-
ity, crossover operator, and n-point random mutation operator are incorporated in the particle swarm optimization
algorithm to further exploit optimal solutions generated by the particle swarm optimization algorithm. These operators
are used to diversify the swarm and prevent premature convergence. Tests on nine constrained mechanical engineering
design optimization problems with different kinds of objective functions, constraints, and design variables in nature
demonstrate the superiority of the dynamic adaptive particle swarm optimization and genetic algorithm against several
other meta-heuristic algorithms in terms of solution quality, robustness, and convergence rate in most cases.
Keywords
Constrained engineering design optimization problems, continuous and discrete design variables, meta-heuristic, dynamic
adaptive, particle swarm optimization, genetic algorithm
Date received: 18 August 2018; accepted: 12 December 2018
Handling Editor: Yunn-Lin Hwang
Introduction especially when the objective functions and constraints

are discontinuous and not smooth.1 Numerous
A great number of optimization algorithms have been
proposed to solve different engineering design optimi- 1
Department of Mechanics and Engineering Science, Sichuan University,
zation problems which are usually nonlinearly con-
Chengdu, China
strained ones. The optimization algorithms can be 2
Key Laboratory of Deep Underground Science and Engineering
roughly divided into two categories: a stochastic algo- (Ministry of Education), School of Architecture and Environment,
rithm and deterministic one. The traditional determi- Sichuan University, Chengdu, China
3
nistic optimization methods, such as the steepest State Key Laboratory of Mechanical Transmissions, Chongqing
University, Chongqing, China
descend method, quasi-Newton method, and interior- 4
Department of Mechanical Engineering, University of Maryland,
reflective Newton method, are usually gradient-based Baltimore County, Baltimore, MD, USA
algorithms and differentiable conditions of objective
functions are required to meet. These methods are inef- Corresponding author:
Hao Zhu, Department of Mechanics and Engineering Science, Sichuan
ficient and inaccurate for complex optimization prob-
University, Chengdu 610065, China.
lems with strong nonlinearity and high dimensions Email: haozhu@scu.edu.cn
Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 Advances in Mechanical Engineering
stochastic optimization algorithms, such as the particle constrained PSO algorithm with the stagnation detec-
swarm optimization (PSO) algorithm,2 genetic algo- tion and dispersion mechanism to tackle real word non-
rithm (GA),3–5 firefly algorithm,6 ant colony optimiza- linear and constrained engineering optimization
tion,7 artificial bee colony (ABC),8 mine blast problems. Yang and colleagues27,28 proposed an accel-
algorithm (MBA),9 simulated annealing (SA) algo- erated particle swarm optimization (APSO) algorithm
rithm,10 biogeography-based optimization (BBO) algo- based on the basic PSO algorithm, in which the velocity
rithm11, have been proposed to overcome these vector is removed and particle best positions are
drawbacks. These stochastic optimization algorithms replaced by randomness. This algorithm greatly
are usually meta-heuristic and inspired by physical and improves calculation efficiency and implementation
natural phenomena. convenience. However, this algorithm is easily trapped
Among all these stochastic optimization algorithms, in premature convergence particularly for the problems
the PSO algorithm is widely applied to solve different with high nonlinearity due to the deficiency of diver-
engineering optimization problems as it is efficient in sity.1 This disadvantage was improved by Guedria1 by
computation, easy for implementation, and reliable in incorporating memories of individual particles into
searching for global optima.12–16 The PSO algorithm APSO forming a new algorithm called improved adap-
first proposed by Kennedy and Eberhart2 is based on tive particle swarm optimization (IAPSO).
social sharing of information between individuals in a To improve the swarm diversity and increase conver-
group and is originated from mimicking the flocking gence rate, many hybrid optimization algorithms with
behavior of a swarm of fish and imitating the schooling some operators or other algorithms incorporated into
behavior of birds. The PSO algorithm is made up of a PSO have been proposed.29–34 Novitasari et al.29 pro-
population of particles which are randomly moving posed a hybrid algorithm that combines the SA with
within the parameter space. The position of each indi- PSO algorithm to deal with constrained optimization
vidual particle in the parameter space denotes a candi- problems. He and Wang30 proposed a similar hybrid
date solution of the design optimization problem. By algorithm to optimize a support vector machine. Wang
changing searching velocities and positions of particles, and Yin31 introduced a ranking selection scheme into
the optimal solution is found. The ability of searching the basic PSO to automatically control search perfor-
optima of the PSO algorithm mainly relies on mutual mance of the swarm, which results in a new algorithm
interaction (social learning) and influence of individual called ranking selection–based particle swarm optimiza-
particles (cognitive learning). Particles move toward the tion (RSPSO). The crossover operators or mutation
currently global best position of the swarm in each operators used in GAs were largely adopted by
iteration. A particle can escape from a local optimum researchers and combined with PSO to generate new
with the help of neighboring particles. But if most of its algorithms, such as the modified particle swarm optimi-
neighboring particles are limited to a local extreme zation (MPSO),32 quantum-behaved PSO using muta-
point, it is attracted to the trap of the local optimum, tion operator with Gaussian distribution (G-QPSO),33
and as a result, premature convergence of the algorithm straightforward particle swarm optimization (SPSO)
and the stagnation phenomenon17 occur. To overcome with a logistic chaotic mutation operator,34 self-
these drawbacks of the basic PSO algorithms, different organizing hierarchical particle swarm optimizer with
improvements have been proposed. A descending time-varying acceleration coefficients (HPSO-TVAC),22
dynamic inertia factor or accelerating factor is widely and so on. These operators increase swarm diversity
adopted to balance the convergence rate and space and prevent premature convergence and stagnation of
searching ability of the PSO algorithm during search the PSO algorithms. The hybrid optimization algo-
process.16,18,19 Eberhart and Shi20 applied a random rithms talked above have been used to solve different
inertia weight factor to deal with dynamic systems. specific engineering optimization problems.
Clerc21 presented a constriction factor K to control the In this work, a dynamic adaptive particle swarm
convergence velocity. Apart from using time-varying optimization and genetic algorithm (DAPSO-GA) pre-
inertia weights (TVIW), time-varying accelerating coef- viously proposed by us in Zhu et al.35 is used to solve
ficients (TVAC) were also proposed and used to con- constrained engineering design optimization problems
trol the convergence rate and solution quality.22,23 A with different kinds of design variables. A dynamic
co-evolutionary particle swarm optimization (CPSO) adaptive inertia factor is used in the PSO algorithm to
was presented by He and Wang24 to solve constrained adjust its convergence rate and control the balance of
engineering optimization problems. They used a global and local optima exploration. GA-related opera-
multiple-swarm technique to evolve decision solutions tors including a selection operator with time-varying
and adapt penalty factors. Later, Krohling and selection probability, crossover operator, and n-point
Coelho25 improved the CPSO by dynamically adjusting random mutation operator are incorporated into the
the accelerated coefficients which satisfy Gaussian PSO to further exploit the optimal solutions generated
probability distribution. Worasucheep26 presented a by the PSO-related algorithm. These operators are used
Zhu et al. 3
to diversify the swarm and prevent premature conver- The procedure of the basic PSO algorithm begins
gence. The remainder of this work is organized as bel- with population initialization of particles with random
lows. The DAPSO-GA for both continuous and positions and velocities. The positions and velocities of
discrete optimization problems with constraints is spe- each particle are then updated by equations (1) and (2).
cifically introduced in section ‘‘Introduction of the After that, the corresponding fitness of each particle is
DAPSO-GA.’’ In section ‘‘Constrained engineering evaluated and ranked, and Pi (l) and Pg (l) are updated.
optimization problems,’’ four benchmark constrained The above procedure is repeated until an ending criter-
engineering optimization problems with continuous ion is met. The ending criterion is usually the maximum
design variables and five ones with discrete or mixed number of iterations or a sufficiently low error bound.
design variables are used to evaluate performance of
the DAPSO-GA on real word engineering optimization
PSO-related algorithm in the DAPSO-GA. A dynamic adap-
problems. Conclusions are drawn in section
tive inertia factor vi (l) is introduced into the basic
‘‘Conclusion.’’
PSO to adaptively adjust its searching velocity during
iterations
Introduction of the DAPSO-GA
vi ðl + 1Þ = vi ðlÞ vi ðlÞ + c1 r1 ðPi ðlÞ xi ðlÞÞ
The DAPSO-GA is a hybrid algorithm that combines ð4Þ
+ c 2 r2 P g ðl Þ x i ðl Þ
the GA and PSO algorithm. Specifically, the GA-
related operators including selection, crossover, and where
n-point random mutation operators are incorporated
into the PSO algorithm with craft. These GA-related bi ðlÞp
vi ðlÞ = vmin +ðvmax vmin Þ sin 2 ½vmin , vmax
operators are used to diversify the swarm and further 2
explore the possible optima based on the feasible solu- ð5Þ
tion provided by the PSO algorithm.
in which
PSO-related algorithm f i ðl Þ f g ðl Þ
bi ðlÞ = 2 ½0, 1, i = 1, 2, . . . , N ð6Þ
Basic PSO algorithm. The basic PSO algorithm is made f w ðl Þ f g ðl Þ
up of a population of particles that are randomly with fi (l) being the fitness value of the ith particle in the
spread within the parameter space. The position of lth iteration, and fg (l) and fw (l) being the best and
each individual particle in the parameter space denotes worst fitness values of the swarm in the lth iteration,
a candidate solution of the design optimization prob- respectively; and they satisfy fg (l) ł fi (l) ł fw (l) and
lem. Each particle has a velocity and moves in the para- thus bi (l) 2 ½0, 1. Particles with the best fitness value
meter space. The position and velocity of the particle i and worst fitness value are called the best particle and
are adjusted in each iteration worst particle in the swarm, respectively. From equa-
tions (5) and (6), the inertia factor is adaptively adjusted
vi ðl + 1Þ = v vi ðlÞ + c1 r1 ðPi ðlÞ xi ðlÞÞ in the range ½vmin , vmax during iteration. The better fit-
ð1Þ
+ c 2 r2 P g ðl Þ x i ðl Þ ness value a particle has, the smaller the inertia factor
is. Large inertia factor represents a large searching velo-
xi ðl + 1Þ = xi ðlÞ + vi ðlÞ ð2Þ city and thus, more solution spaces will be explored. In
where xi (l) and vi (l) are the position and velocity of contrast, small inertia factor can help the PSO algo-
the particle at time step l, respectively; Pi (l) is the his- rithm further exploit the solution space around the best
torical best position of the particle i so far and Pg (l) is particle. Hence, this dynamic adjustment of the inertia
factor can adaptively balance the convergence rate and
the global best position of the whole swarm up to a time
global optima search ability of the PSO algorithm.
step l; r1 and r2 are random numbers within a range
Each particle position xi is limited in the range
from 0 to 1; v is an inertia factor; and c1 and c2 are two
½xmin , xmax . If xi locates outside this range, it will be
accelerating factors used to scale influence of the best
replaced by
positions of the particle i and global best position of the
swarm, respectively. To ensure convergence of the PSO
xkmax , if xki .xkmax
algorithm, the two accelerating factors are constrained xki = , k = 1, 2, . . . , D ð7Þ
xkmin , if xki \xkmin
by13,16
in which D is the particle dimension and xki is the posi-
0\ðc1 + c2 Þ\4 tion of the ith particle in the kth dimension. Each parti-
ð3Þ
ðc1 + c2 Þ=2 1\v\1 cle velocity vi (l) is limited in ½vmin , vmax , in which
Figure 1. Flowchart of the crossover operator of the GA-related algorithm in the DAPSO-GA.

vmin = (xmax xmin )=2 and vmax = (xmax xmin )=2. fi ðlÞ fg ðlÞ
0\ \h
ð9Þ
If the particle velocity violates this limit, it will be f g ðl Þ
replaced by
where fi (l) is the current fitness value of the ith particle

vkmax , if vki .vkmax at the lth iteration, fg (l) is the best fitness value of the
vki = , k = 1, 2, . . . , D ð8Þ swarm that corresponds to its global best position, and
vkmin , if vki \vkmin
h = hmax l(hmax hmin )=ltot is the time-varying
in which vki is the velocity of the ith particle in the kth selection probability which descends from hmax to hmin
dimension. during iteration process.
GA-related algorithm Crossover and mutation operator. When the GA-selection

criterion is met, the following two GA-related operators
In the DAPSO-GA, GA-related operators, that is, the
are used to update the particle position: randomly gen-
selection operator with time-varying selection probabil-
erate a number a 2 ½0, 1, and then a crossover opera-
ity, crossover operator, and n-point random mutation
tor is applied if a ł Pc , where Pc is predefined crossover
operator are introduced to further exploit the optimal
probability; otherwise, an n-point random mutation
solutions generated by the adaptive PSO algorithm.
operator is applied and 1 Pc is corresponding muta-
GA uses a population which consists of individuals or
tion probability.
chromosomes and each individual stands for a poten-
tial solution. In the GA-related algorithm, each particle
in the swarm is regarded as an individual or chromo- Crossover operator. A random crossover operator is
some and the swarm constitutes a population. Each adopted here to generate a new individual (particle).
individual is represented by applying decimal coding The flowchart of the crossover operator is illustrated in
(the real value). Figure 1. First, two particles should be selected as par-
ents (pa and ma) for breeding. Suppose the ith particle
is already selected as pa according to the GA-selection
Adaptive dynamic selection operator. A particle that meets criterion, and then another jth particle is randomly
the GA-selection criterion below is selected to update selected as ma from the swarm, where j 2 ½1, M and
its position via the following crossover and mutation j 6¼ i. Second, two cutting points (CPs), that is, e1 and
operators in iteration e2 , are randomly generated and then two sub-vectors
Zhu et al. 5
Figure 2. Procedures of the n-point mutation operator.
(s1 and s2 ) are picked out from the position vectors of number of generations (iterations) S, maximum and
the parents, where 0\e1 ł e2 ł D. Components of the minimum inertia factors vmax and vmin , respectively,
selected two sub-vectors are inter-changed and then a accelerating factors c1 and c2 , maximum and mini-
new particle (offspring) is generated. mum selection probability hmax and hmin , respec-
tively, crossover probability pc , upper and lower
Mutation operator. An n-point random mutation limits of the position of each particle xui and xli ,
operator is used, where n is the mutation dimension respectively. In this work, vmax = 0:7, vmin = 0:4,
(i.e. the number of components or genes of the selected hmax = 0:7, hmin = 0:15, c1 = c2 = 2, and pc = 0:5
particle or chromosome for mutation) which is a ran- are used.
dom integer in ½1, D. It means that there are in total n Step 2: Initialize the swarm: randomly generate a
points (genes) in the selected particle (chromosome) to swarm with a size of M and the initial position of
be changed via mutation. Procedure of the n-point ran- each particle is given by
dom mutation operator is shown in Figure 2. First, the
mutation dimension n of the selected particle is identi- xi ð0Þ = xli + rand xui xli , i = 1, 2, . . . , M ð11Þ
fied by n = round(rand 3 D), in which rand is a random
number in ½0, 1 and round is an operator to round off Step 3: Evaluate the fitness value of each initially
the product of rand and D. Second, n different integers generated particle and rank their positions. The ini-
(i.e. m1 , m2 , . . . , mn ), which are limited in the range tial best particle position Pi (0) and initial global best
½1, D, are randomly generated. These integers repre- and worst positions Pg (0) and Pw (0) of the swarm,
sent the mutation positions in the position vector of the respectively, are then identified.
selected particle. Next, values of the components of the Step 4: Update the current position xi (l) and velo-
selected particle position vector are randomly changed city vi (l) of the ith particle according to equations
via the following equation (2) and (4)–(6).
Step 5: Evaluate the current fitness value of each par-
p k
xki = xkmin + sin rand 3 3 xmax xkmin , k = m1 , ticle, and update the best particle position Pi (l) and
2 global best and worst positions of the swarm Pg (l)
m2 , . . . , mn and Pw (l), respectively.
ð10Þ Step 6: Generate new particles (offspring) according
to the GA-related algorithm to diversify the swarm.
If the GA-selection criterion in equation (9) is met,
the crossover operator and n-point random muta-
Implementation procedure of the DAPSO-GA
tion operator are applied to update the position of a
algorithm selected particle to generate a new particle xi (l) as
Flowchart of the DAPSO-GA is shown in Figure 3 and presented in section ‘‘Crossover and mutation
it is briefly described as follows: operator.’’
Step 7: Evaluate the fitness value of the new particle
Step 1: Set initial values of the optimization para- fi (l) and compare it with the best and worst fitness
meters including the population size M, maximum values of the swarm fg (l) and fw (l), respectively. If
Figure 3. Flowchart of the DAPSO-GA.
fi (l)\fg (l), replace the best particle by it; otherwise, optimization problems with discrete variables. For the
replace the worst particle by it if fi (l)\fw (l). Update discrete optimization problems, the DAPSO-GA can
the best particle position Pi (l) and global best and be modified using the rounding off approach. In this
worst positions of the swarm Pg (l) and Pw (l) if approach, either the continuous or discrete variables
necessary. are treated as continuous variables during optimization
Step 8: Repeat the above steps 4–7 until the termina- processes. Only at the end of the optimization proce-
tion criterion, which is a predefined number of itera- dure, the discrete variables will be rounded off to evalu-
tion, is met and then output the optimal results. ate the fitness value of each particle as shown below
fi ðxi , lÞ = fi ðround ðxi Þ, lÞ ð12Þ

Strategies of the DAPSO-GA for discrete optimization
Values of the discrete variables are in fact not chan-
problems ged as seen in equation (12) and keep unchanged until
The DAPSO-GA talked above is suitable for a continu- at the end of each generation of iteration. For conve-
ous optimization problem, but cannot handle the nient description, the DAPSO-GA using the rounding
Zhu et al. 7
Figure 4. Schematic of the tension/compression spring.
off approach is called a discrete DAPSO-GA and is Constrained engineering optimization

used to solve the discrete optimization problems later. problems
In this section, nine famous constrained benchmark
Constraints handling mechanical engineering optimization problems which
have different objective functions, design variables and
For constrained optimization problems, a feasible solu- constraints in nature are adopted to test the perfor-
tion should satisfy all boundary constraints in the form mance of the proposed DAPSO-GA in terms of solu-
of the equalities and/or inequalities. Two strategies are tion quality and stability as well as convergence rate.
used in this work to handle the constraints on design These 10 constrained engineering optimization prob-
variables and problem-specific constraints. In the lems are divided into continuous and discrete optimiza-
DAPSO-GA, each particle position will be reset to the tion problems according to the categories of their
maximum or minimum boundary value once the limits variables, and the rounding off strategy talked in sec-
on design variables are violated. Global optima usually tion ‘‘Strategies of the DAPSO-GA for discrete optimi-
occur on or near the boundary of the solution (design) zation problems’’ is used in the DAPSO-GA to deal
space for the majority of design optimization prob- with the discrete optimization problems. Statistical
lems.9 Hence, this strategy can increase the probability results and best solutions of all algorithms for these
for finding global optimal solutions. Penalty function engineering optimization problems are obtained over
strategies such as the penalty factor method1,35–38 and 30 independent runs.
the concept of parameter free penalty function39,40 are
widely used to solve different constrained optimization
problems. The penalty factor method is adopted in this Constrained engineering optimization problems with
work to handle the problem-specific constraints. The continuous variables
constrained optimization problem using the penalty
function strategy can be described as bellow Tension/compression spring design problem. Figure 4 shows
! a schematic of a tension/compression spring.41 The
X
N X
L
design aim of the tension/compression problem (i.e. the
Minimize fp ðxÞ = f ðxÞ + b mj gj ðxÞ + jhm ðxÞj
j=1 m=1 objective function f (x)) is to minimize its weight with
ð13Þ constraints on minimum deformation, shear stress,
surge frequency, and maximum outside diameter.
where f (x) and fp (x) are original and penalized objective These constraints constitute four nonlinear inequality
functions, respectively; N and L are the total number of equations as detailed in Appendix 1 (section ‘‘Tension/
inequality constraints and equality constraints, respec- compression spring design problem’’). The design prob-
tively; gj (x) and hm (x) are the jth inequality constraint lem has three design variables including the wire dia-
and mth equality constraint, respectively, and meter d, mean coil diameter D, and number of active
coils P, which are denoted by x1 , x2 , and x3 in the

mj = 1, if the constraint gj ðxÞ is violated objective function and constraint functions, respec-
ð14Þ tively. The DAPSO-GA proposed is used to solve this
mj = 0, else
optimization problem. The swarm size and maximum
b is the penalty factor which is a large positive con- iteration number are 10 and 200, respectively. Figure 5
stant that satisfies b f (x) and b = 1020 is adopted in presents the convergence history of GA, standard PSO,
later application in this work. By introducing the pen- and the proposed DAPSO-GA for the tension/compres-
alty term, the constrained optimization problem sion spring problem. It is seen that the standard PSO
becomes an unconstrained one as seen in equation (13). and DAPSO-GA convergence faster than GA, while the
previously proposed algorithms. Statistical optimization

results of all algorithms are listed in Table 2. As seen
from Table 2, the DAPSO-GA provides the best solution
with least NFEs 2000. HEAA has the best robustness in
terms of providing optimal solutions with standard
deviation (SD) value of only 1.4E–9 for the tension/com-
pression spring design problem. The PSO algorithm pro-
vides the worst solution (0.012857) with the largest SD
value and GA1 requires the highest NFEs (900,000).
Figure 6 shows the inertia weight versus number of
iterations of the DAPSO-GA on the tension/compression
spring design problem. From Figure 6, the inertia weight-
ing factor varies between 0.7 and 0.4. A large inertia
Figure 5. Convergence history of GA, standard PSO, and the weighting factor is used when the fitness value of a parti-
proposed DAPSO-GA for the tension/compression spring cle is far away from the global best fitness value; other-
design problem. wise, a small one is used. The dynamic inertia weighting
factor adaptively adjusts the search velocity so that the
DAPSO-GA has better global optimum searching abil- exploitation and exploration are well balanced.
ity. The DAPSO-GA is also compared with several other
meta-heuristic optimization algorithms including the
APSO,27 IAPSO,1 MBA,9 ABC2,40 GA1,41 GA2,42 water Symmetric three-bar truss design problem. Figure 7 presents
cycle algorithm (WCA),43 differential evolution (DE),44 the schematic diagram of a symmetric three-bar truss
differential evolution with level comparison (DELC),45 structure. The symmetric three-bar truss structure is
Nelder-Mead and Particle Swarm Optimization (NM- made up of steel and is subjected to two constant load-
PSO),46 hybrid evolutionary algorithm and adaptive con- ings P1 = P2 = P. The optimization design problem of
straint handling technique (HEAA),47 differential evolu- the three-bar truss structure, which was described by
tion with dynamic stochastic selection (DEDS),36 Ray and Liew,49 is to minimize the volume subject to
quantum-behaved particle swarm optimization stress constraints as detailed in Appendix 1 (section
(QPSO),48 G-QPSO,48 society and civilization (SC),49 lea- ‘‘Symmetric three-bar truss design problem’’). The
gue championship algorithm (LCA),50 cultural algo- design variables are cross-sectional areas of the three
rithms with evolutionary programming (CAEP),9 unified bars: x1 , x2 , and x3 . The DAPSO-GA with a swarm
particle swarm optimization (UPSO),51(m + l) ES,52 size of 20 and maximum number of iterations of 5000 is
and PSO-DE.53 The optimal solutions obtained by the used to solve this optimization problem. The optimal
proposed DAPSO-GA and above optimization algo- solution obtained by this algorithm is compared with
rithms are listed in Table 1. It is apparently seen that the those obtained by other optimization algorithms such
proposed DAPSO-GA finds the best solution with the as Hernendez,54 dynamic stochastic selection for multi-
objective function value 0.009872 that was not found by member differential evolution (DSS-MDE),36 SC,49
Table 1. Comparison of optimal solutions obtained from different optimization algorithms for tension/compression spring design
problem.
DV x1 x2 x3 g1 (x) g2 (x) g3 (x) g4 (x) f (x)
IAPSO 0.051685 0.356629 11.294175 21.97E–10 24.64E–10 24.05361 21.091686 0.01266523

APSO 0.052588 0.378343 10.138862 21.549E–4 28.328E–4 24.089171 21.069069 0.0127
MBA 0.051656 0.35594 11.344665 20.0009 20.1344 24.052248 20.728268 0.012665
GA1 0.051480 0.351661 11.632201 22.08E–3 21.10E–4 24.026318 24.026318 0.0127047834
WCA 0.051680 0.356522 11.30041 21.65E–13 27.9E–14 24.053399 20.727864 0.012665
DELC 0.051689 0.356717 11.288965 23.4E–9 2.44E–9 24.053785 20.727728 0.012665
NM-PSO 0.051620 0.355498 11.333272 1.01E–3 9.94E–4 24.061859 20.728588 0.012630
HEAA 0.051689 0.356729 11.288293 3.96E–10 23.59E–10 24.053808 20.72772 0.012665
DEDS 0.051689 0.356717 11.288965 1.45E–9 21.19E–9 24.053785 20.727728 0.012665
G-QPSO 0.051515 0.352529 11.538862 24.83E–5 23.57E–5 24.0455 20.73064 0.012665
ABC2 0.051689 0.356720 11.288832 22.53E–13 25.76E–13 24.05378 20.7277 0.012665
DAPSO-GA 0.050 0.3744328 8.54657332 21.2581E–8 21.4491E–7 24.860733 20.717045 0.0098724562
IAPSO: improved adaptive particle swarm optimization; APSO: accelerated particle swarm optimization; DV: design variable; G-QPSO: quantum-
behaved PSO using mutation operator with Gaussian distribution; MBA: mine blast algorithm; DAPSO-GA: dynamic adaptive particle swarm
optimization and genetic algorithm.
Note: The boldfaced data in each table mean the best one among all the results provided by different algorithms.
Zhu et al. 9
Table 2. Comparison of statistical results obtained from different optimization algorithms for tension/compression spring design
problem.
Algorithms Worst Mean Best SD NFEs
IAPSO 0.01782864 0.013676527 0.01266523 1.57E–3 2000

APSO 0.014937 0.013297 0.012700 6.85E–4 120,000
MBA 0.012900 0.012713 0.012665 6.3E–5 7650
LCA 0.01266667 0.01266541 0.01266523 3.88E–7 15,000
WCA 0.012952 0.012746 0.012665 8.06E–5 11,750
SC 0.016717 0.012922 0.012669 5.9E–4 25,167
PSO-DE 0.012665 0.012665 0.012665 1.2E–8 24,950
HEAA 0.012665 0.012665 0.012665 1.4E–9 24,000
DEDS 0.012738 0.012669 0.012665 1.3E–5 24,000
DELC 0.012665 0.012665 0.012665 1.3E–7 20,000
DE 0.012790 0.012703 0.012670 2.7E–5 204,800
PSO 0.071802 0.019555 0.012857 1.1662E–2 2000
QPSO 0.018127 0.013854 0.012669 1.341E–3 2000
G-QPSO 0.017759 0.013524 0.012669 1.268E–3 2000
NM-PSO 0.012633 0.012631 0.012630 8.47E–7 80,000
HPSO 0.012719 0.012707 0.012665 1.58E–5 81,000
CPSO 0.012924 0.012730 0.012674 5.2E–4 240,000
CAEP 0.015116 0.013568 0.012721 8.42E–4 50,020
GA1 0.012822 0.012769 0.012704 3.94E–5 900,000
GA2 0.012973 0.012742 0.012681 5.9E–5 80,000
UPSO – 0.02294 0.01312 7.20E–03 100,000
(m + l) ES – 0.013165 0.012689 3.9E–04 30,000
ABC2 0.012710407 0.01266897 0.01266523 9.43E–06 –
DAPSO-GA 0.015354687 0.0107 0.0098724562 1.591E–3 2000
SD: standard deviation; IAPSO: improved adaptive particle swarm optimization; APSO: accelerated particle swarm optimization; MBA: mine blast
algorithm; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm; PSO: particle swarm optimization; CPSO: co-
evolutionary particle swarm optimization; NFE: number of function evaluation; G-QPSO: quantum-behaved PSO using mutation operator with
Gaussian distribution; HPSO: hybrid particle optimization algorithm; PSO-DE: Particle swarm optimization with differential optimization.
Note: The boldfaced data mean optimal results provided by the DAPSO-GA algorithm.
Figure 6. Inertia weight versus number of iterations.

Figure 7. Schematic diagram of the three-bar truss.
swarm with an intelligent information sharing (SIIS),55
and PSO-TVAC22 as seen in Table 3.Table 4 presents Tables 3 and 4, almost all optimization algorithms pro-
the comparison of statistical results obtained from these vide similar optimal solutions. The proposed DAPSO-
optimization algorithms for the three-bar truss design GA provides the best solution with the minimum SD
problem in terms of the worst, mean, and best solutions value. DAPSO-GA and PSO-TVAC convergence to the
as well as the SD values and NFEs. As seen from best solution with similar NFEs which is less than those
Table 3. Comparison of optimal solutions obtained from different optimization algorithms for the three-bar truss design problem.
DV Hernendez DSS-MDE SC SIIS PSO-TVAC DAPSO-GA
x1 0.788 0.7886751359 0.788621037 0.795 0.7887058767 0.7886769887

x2 0.408 0.4082482868 0.408401334 0.395 0.4081613457 0.4082430493
g1 (x) 1.637E–3 22.104E–11 28.275E–9 23.376E–3 24.448E–13 22.3114E–09
g2 (x) 21.4636 21.4641 21.4639 21.4809 21.4642 21.4641075
g3 (x) 20.5348 20.5359 20.5361 20.5225 20.5358 20.535892
f (x) 263.9 263.8958434 263.8958466 264.3 263.895844071 263.895843684
DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm; DV: design variable; PSO-TVAC: Particle swarm optimization with
time-varying accelerating coefficients.
Table 4. Comparison of statistical results obtained from different optimization algorithms for the three-bar truss design problem.
Algorithms M S Worst Mean Best SD NFEs
DSS-MDE 10 300 263.8958498 263.8958436 263.8958434 9.72E–7 15,000

SC 20 1000 263.96975 263.9033 263.8958466 1.26E–2 17,610
PSO-TVAC 20 300 263.948096212 263.903085482 263.895848599 1.27E–02 6000
DAPSO-GA 20 300 263.947633138 263.902926027 263.895843684 8.30E–03 7131
SD: standard deviation; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm; NFE: number of function evaluation.
Figure 8. Convergence history of GA, standard PSO, and the Figure 9. Schematic diagram of the welded beam.
proposed DAPSO-GA for the three-bar truss design problem.
Welded beam design. The welded beam design problem
of other optimization algorithms. DSS-MDE provides is a famous constrained optimization problem which is
the best solution with the largest NFEs (15,000). Thus, widely used as a benchmark problem to evaluate per-
the superiority of the proposed DAPSO-GA for the formance of newly proposed optimization algorithms.9
three-bar truss structure design problem in solution Figure 9 shows the schematic diagram of a welded
quality and convergence rate is justified. Figure 8 beam structure which consists of a beam and weld. The
shows the convergence history of GA, standard PSO, optimization target is the minimum fabrication cost of
and the proposed DAPSO-GA for the three-bar truss the beam subject to constraints on bending and shear
structure design problem. It is seen that the DAPSO- stress (s and t) on the bar, bucking load (Pb ), and its
GA converges faster to the near optimal solution at end deflections (d). The design variables for this design
early iterations and then gradually improves the solu- problem are the weld thickness h, weld length l, beam
tion accuracy due to the technique of the proposed width t, and beam thickness b, which are respectively
algorithm in adaptively balancing the exploration and denoted by x1 , x2 , x3 , and x4 in the objective function
exploitation during searching process. and constraint equations as presented in Appendix 1
Zhu et al. 11
Table 5. Comparison of optimal solutions obtained from different optimization algorithms for the welded beam design optimization
problem.
DV GA3 IAPSO APSO MBA WCA NM-PSO CAEP
x1 0.2489 0.2057296 0.202701 0.205729 0.205728 0.205830 0.205700

x2 0.1730 3.47048866 3.574272 3.470493 3.470522 3.468338 3.470500
x3 8.1789 9.03662391 9.040209 9.036626 9.036620 9.036624 9.036600
x4 0.2533 0.20572964 0.2059215 0.205729 0.205729 0.20573 0.205700
g1 (x) 25758.604 21.05E–10 2117.46706 20.001614 20.034128 20.02525 1.988676
g2 (x) 2255.5769 26.91E–10 251.712981 20.016911 23.49E–05 20.053122 4.481548
g3 (x) 20.004400 27.66E–15 20.003221 22.10E–7 21.19E–06 0.000100 0
g4 (x) 22.982866 23.4329838 23.421741 23.432982 23.432980 23.433169 23.433213
g5 (x) 20.123900 20.0807296 20.077701 20.080729 20.080728 20.080830 20.080700
g6 (x) 20.234160 20.23554032 20.235571 20.235540 20.235540 20.235540 20.235538
g7 (x) 244.65.271 25.80E–10 218.367012 20.001464 20.013503 20.031555 2.603347
f (x) 2.433116 1.7248523 1.736193 1.724853 1.724856 1.724717 1.724852
DV CPSO GA1 GA2 HPSO-GA ABC2 DAPSO-GA
x1 0.202369 0.2088 0.205986 0.2057296 0.2057245 0.205728318

x2 3.544214 3.4205 3.471328 3.25312 3.25325369 2.994714573
x3 9.048210 8.9975 9.020224 9.0366239 9.03664438 9.036612639
x4 0.205723 0.2100 0.206480 0.2057296 0.20572999 0.205730191
g1 (x) 213.655547 20.337812 20.103049 – 20.17975428 20.0732
g2 (x) 278.814077 2353.9026 20.231747 – 20.18697948 20.0057
g3 (x) 23.35E–03 20.0012 25.0E–04 – 20.00000549 21.873E–6
g4 (x) 23.424572 23.411865 23.430044 – 23.45240767 23.4755
g5 (x) 20.077369 20.0838 20.080986 – 20.08072450 20.0807
g6 (x) 20.235595 20.235649 20.235514 – 20.22831066 20.2355
g7 (x) 24.472858 2363.2324 258.64688 – 20.03957707 20.0434
f (x) 1.728024 1.7483094 1.78226 1.6952471 1.69526388 1.6600473
IAPSO: improved adaptive particle swarm optimization; APSO: accelerated particle swarm optimization; MBA: mine blast algorithm; CPSO:
co-evolutionary particle swarm optimization; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm; DV: design variables.
(section ‘‘Welded beam design’’). The DAPSO-GA with algorithms. The proposed algorithm can stably find the
a swarm size of 50 and maximum number of iterations best solution with almost the fewest NFEs (13,356)
of 5000 is used to solve this optimization problem. which is only larger than that of the IAPSO (12,500).
The optimization algorithms previously used to In terms of SD, the proposed algorithm has better
solve this design optimization problem include GA3,56 robustness in detecting the best solution than other
GA4,35 APSO, IAPSO, MBA, LCA, WCA, DE, SC, reported optimization algorithms apart from the
NM-PSO, PSO-DE, HPSO,29 CPSO,24 CAEP, GA1, IAPSO, MBA, LCA, hybrid particle swarm optimiza-
hybrid PSO-GA (HPSO),39 ABC2,40 and GA2. Table 5 tion and genetic algorithm (HPSO-GA), ABC2, and
presents the comparison of optimal solutions provided PSO-DE. Figure 10 shows the convergence history of
by the previously reported algorithms and proposed GA, standard PSO, and the proposed DAPSO-GA for
DAPSO-GA. From Table 5, a new optimal solution, the welded beam design problem. It is seen that the
which is better than those provided by previously pro- standard PSO and DAPSO-GA convergence faster
posed algorithms, is found by the proposed DAPSO- than GA, while the DAPSO-GA has better global opti-
GA with the objective function value of 1.6600473. mum searching ability.
Note that the optimal solution provided by CAEP is
infeasible as the constraints g1 (x) and g2 (x) are vio- Belleville disc spring design problem. As shown in Figure 11,
lated. Table 6 presents the comparison of statistical Belleville disc spring is made up of several conical discs
results provided by all previously reported algorithms with uniform rectangular cross-sections. The design
and proposed DAPSO-GA for the welded beam design objective of the Belleville disc spring is to minimize its
optimization problem in terms of the worst, mean, and total weight subject to geometric constraints concerns the
best solutions as well as the SD and NFEs. As seen outer and inner diameter, slope and height to maximum
from Table 6, DAPSO-GA provides better solutions height, and kinematic and strength constraints concerns
than the newly proposed optimization algorithm WCA, the compression deformation and stress and height to
MBA, and IAPSO as well as other optimization deformation. There are four design variables for this
Table 6. Comparison of statistical results obtained from different optimization algorithms for the welded beam design optimization
problem.
Algorithm Worst Mean Best SD NFEs
GA3 2.64583 2.39203 (median) 2.38119 – 320,080

GA4 2.64583 2.39289 (median) 2.38119 – 40,080
APSO 1.993999 1.877851 1.736193 7.6118E–02 50,000
IAPSO 1.7248624 1.7248528 1.7248523 2.02E–06 12,500
MBA 1.724853 1.724853 1.724853 6.94E–19 47,340
LCA 1.7248523 1.7248523 1.7248523 7.11E–15 15,000
WCA 1.744697 1.726427 1.724856 4.29E–03 46,450
DE 1.824105 1.768158 1.733461 2.21E–02 204,800
SC 6.399678 3.002588 2.385434 9.60E–01 33,095
NM-PSO 1.733393 1.726373 1.724717 3.50E–03 80,000
PSO-DE 1.724852 1.724852 1.724852 6.70E–16 66,600
HPSO 1.814295 1.749040 1.724852 4.01E–02 81,000
CPSO 1.782143 1.748831 1.728024 1.29E–02 240,000
CAEP 3.179709 1.971809 1.724852 4.43E–01 50,020
GA1 1.785835 1.771973 1.748309 1.12E–02 900,000
GA2 1.993408 1.792654 1.728226 7.74E–02 80,000
HPSO-GA 1.6952741 1.6952741 1.6952741 2.192E–09 –
ABC2 1.6953706 1.6953084 1.69526388 2.84E–05 –
DAPSO-GA 1.66876995 1.66043211083 1.66004730498 1.608237E–03 13,356
SD: standard deviation; APSO: accelerated particle swarm optimization; IAPSO: improved adaptive particle swarm optimization; MBA: mine blast
algorithm; CPSO: co-evolutionary particle swarm optimization; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm.
Figure 11. Schematic diagram of the Belleville disc spring.
reported algorithms and proposed DAPSO-GA. Note

that the optimal solutions provided by the Gene AS1
and Siddall are infeasible as the first and second con-
straints are violated by them, respectively. Hence, they
Figure 10. Convergence history of GA, standard PSO, and the are not used for comparison. From Table 7, the pro-
proposed DAPSO-GA for the welded beam design problem. posed algorithm and MBA provide better solutions
against other optimization algorithms with the objective
design problem including the spring external and internal function value of 1.9796747. Table 8 presents the com-
diameters (De and Di ), spring thickness (t), and spring parison of statistical results provided by the previously
height (h), which are denoted by x1 , x2 , x3 , and x4 , respec- reported algorithms and proposed DAPSO-GA for the
tively. The DAPSO-GA with a swarm size of 50 and Belleville disc spring design optimization problem in
maximum number of iterations of 1000 is used to solve terms of the worst, mean, and best solutions as well as
this optimization problem. the SD values and NFEs. As seen from Table 8, the
The optimization algorithms previously used to solve proposed DAPSO-GA, ABC, TLBO, and MBA almost
this design optimization algorithm include MBA, ABC, provide the same best solutions, but the proposed algo-
teaching-learning-based optimization (TLBO),57 treat- rithm requires the fewest NFEs 9000 and ABC and
ing constrains as objectives (TCO),58 Siddall,59 Gene TLBO requires the most NFEs 150,000. In terms of
AS1,60 and Gene AS2.60 Table 7 presents the compari- SD, MBA has better robustness in detecting the best
son of optimal solutions provided by the previously solution than other optimization algorithms. Figure 12
Zhu et al. 13
Table 7. Comparison of optimal solutions obtained from different optimization algorithms for the Belleville disc spring design
optimization problem.
DV Coello Gene AS1 Gene AS2 Siddall TLBO MBA DAPSO-GA
x1 0.208 0.205 0.210 0.204 0.204143 0.204143 0.20414335

x2 0.2 0.201 0.204 0.200 0.20 0.20 0.2
x3 8.751 9.534 9.268 10.03 10.03047 10.0304732 10.03047329
x4 11.067 11.627 11.499 12.01 12.01 12.01 12.01
g1 (x) 2145.4109 210.3396 2127.2624 134.0816 1.77E–06 4.58E–04 2.9296E–06
g2 (x) 39.75018 2.8062 194.222554 212.5328 7.46E–08 3.04E–07 26.7998E–08
g3 (x) 0 0.0010 0.0040 0 5.80E–11 9.24E–10 0.7797037
g4 (x) 1.592 1.5940 1.5860 1.596 1.595857 1.595856 1.59585664
g5 (x) 0.943 0.3830 0.5110 0 2.35E–09 0 0
g6 (x) 2.316 2.0930 2.2310 1.98 1.979527 1.979526 1.97952679
g7 (x) 0.21364 0.20397 0.20856 0.19899 0.198966 0.198965 0.1989657
f (x) 2.121964 2.01807 2.16256 1.978715 1.979675 1.9796747 1.979674757
MBA: mine blast algorithm; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm.
Table 8. Comparison of statistical results obtained from different optimization algorithms for the Belleville disc spring design
ABC 2.104297 1.995475 1.979675 0.07 150,000

TLBO 1.979757 1.979687 1.979675 0.45 150,000
MBA 2.005431 1.984698 1.9796747 7.78e–03 10,600
DAPSO-GA 2.558209 2.132861 1.9796747 0.2358 9000
SD: standard deviation; ABC: artificial bee colony; MBA: mine blast algorithm; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic
algorithm.
DAPSO-GA convergence faster than GA, while the

DAPSO-GA has better global optimum searching
ability.
Constrained engineering optimization problems

with discrete variables
Speed reducer design problem. Figure 13 shows a sche-
matic diagram of a speed reducer. The design optimiza-
tion scheme of the speed reducer is to minimize its
weight subject to strength constraints concerning gear
teeth bending stress and surface stress, stresses in and
transverse deflections of shafts.1 The design variables
of this design problem include the face width (b), teeth
module (m), number of teeth in the pinion (z), length of
the first and second shafts between their bearings (l1
and l2 ), diameter of the first shaft (d1 ), and diameter of
Figure 12. Convergence history of GA, standard PSO, and the
proposed DAPSO-GA for the Belleville disc spring design the second shaft (d2 ). These design variables are
problem. denoted by x1 , x2 , x3 , x4 , x5 , x6 , and x7 , respectively, in
the objective function and constraint equations as pre-
shows the convergence history of GA, standard PSO, sented in Appendix 2 (section ‘‘Speed reducer design
and the proposed DAPSO-GA for the Belleville disc problem’’). The design variable x3 (i.e. number of teeth
design problem. It is seen that the standard PSO and in the pinion) is a discrete (integer) design variable and
Figure 13. Schematic diagram of speed reducer.
the reported algorithms including DEDS, DELC,

HEAA, WCA, LCA, and IAPSO provide similar best
solutions (x = (3:5, 0:7, 17, 7:3, 7:715319, 3:350214,
5:286654)) with the objective function value of
2994.4711. Table 10 presents the comparison of statisti-
cal results provided by the previously reported algo-
rithms and proposed DAPSO-GA for the speed
reducer design optimization problem in terms of the
worst, mean, and best solutions as well as the SD val-
ues and NFEs. As seen from Table 10, DAPSO-GA
and IAPSO can stably find the same best solutions with
fewer number of iterations and medium SD value at
the same time compared with other algorithms.
Although DELC, differential evolution with dynamic
stochastic selection (DEDS), and LCA can locate the
best solution with the lowest level of SD (10212) among
Figure 14. Convergence history of the proposed DAPSO-GA all algorithms, they need much more NFEs (30,000,
for the speed reducer design problem. 30,000, and 24,000, respectively) than DAPSO-GA
(7320) and IAPSO (6000).
the remainder variables are continuous. The discrete
DAPSO-GA with a swarm size of 30 and maximum Gear train design problem. Figure 15 shows a schematic
number of iterations of 200 is used to solve this optimi- diagram of a gear train which consists of four gears.
zation problem. Figure 14 shows the convergence his- The scheme of the gear train design optimization prob-
tory of the proposed DAPSO-GA for the speed reducer lem is to minimize the error between the obtained gear
design problem. The objective function value reduces ratio and the required gear ratio of 1/6.3962 subject to
fast to the near minimum at early iterations (less than constraints only on the allowable ranges of design vari-
50 iterations), which presents the high convergence rate ables (side constraints), which are the number of teeth
of this algorithm for this constrained discrete design of the four gears. It is a discrete optimization problem
problem. as all design variables are integers. Numbers of teeth of
This design optimization algorithm was previously gears A, B, D, and F (i.e. design variables) in Figure 15
solved by researchers using different optimization algo- are respectively denoted by x1 , x2 , x3 , and x4 in the objec-
rithms such as DEDS, DELC,45 HEAA, MDE,61 PSO- tive function as presented in Appendix 2 (section ‘‘Gear
DE,54 WCA, MBA, LCA, APSO, IAPSO, TLBO, (m train design problem’’). The discrete DAPSO-GA with a
+ l)-ES, SC, and ABC. Table 9 presents the compari- swarm size of 30 and maximum number of iterations of
son of optimal solutions provided by the previously 100 is used to solve this optimization problem.
reported algorithms and proposed DAPSO-GA. As This design problem was solved before by many
seen from Table 9, the proposed algorithm and most of researchers using different optimization algorithms
Zhu et al. 15
Table 9. Comparison of optimal solutions obtained from different optimization algorithms for the speed reducer design
DV DEDS DELC HEAA MDE PSO-DE WCA
x1 3.5 3.5 3.500022 3.50001 3.5 3.5

x2 0.7 0.7 0.7 0.7 0.7 0.7
x3 17 17 17.000012 17 17 17
x4 7.3 7.3 7.300427 7.300156 7.3 7.3
x5 7.715319 7.715319 7.715377 7.800027 7.8 7.715319
x6 3.350214 3.350214 3.350230 3.350221 3.350214 3.350214
x7 5.286654 5.286654 5.286663 5.286685 5.2866832 5.286654
g1 ðxÞ 20.0739153 20.0739153 20.0739218 20.0739179 20.0739153 20.0739153
g2 ðxÞ 20.197999 20.197999 20.198005 20.198001 20.197999 20.197999
g3 ðxÞ 20.499172 20.499172 20.499094 20.499144 20.499172 20.499172
g4 ðxÞ 20.904644 20.904644 20.904642 20.901471 20.901472 20.904644
g5 ðxÞ 5.9647E-07 5.9647E-07 21.3025E-05 25.4109E-06 5.9647E-07 5.9647E-07
g6 ðxÞ 2.6369E-07 2.6369E-07 24.8334E-06 2.2119E-08 1.6887E-08 2.6369E-07
g7 ðxÞ 20.7025 20.7025 20.7025 20.7025 20.7025 20.7025
g8 ðxÞ 0 0 26.2857E-06 2.8571E-06 0 0
g9 ðxÞ 20.583333 20.583333 20.583331 20.583332 20.583333 20.583333
g10 ðxÞ 20.0513259 20.0513259 20.0513781 20.0513447 20.051326 20.0513259
g11 ðxÞ 5.1845E-08 5.1845E-08 26.1825E-06 20.0108558 20.0108524 5.1845E-08
f ðxÞ 2994.4711 2994.4711 2994.49911 2996.35669 2996.34817 2994.47107
DV MBA LCA APSO IAPSO DAPSO-GA
x1 3.5 3.5 3.501313 3.5 3.5

x2 0.7 0.7 0.7 0.7 0.7
x3 17 17 18 17 17
x4 7.300033 7.3 8.127814 7.3 7.3
x5 7.715772 7.8 8.042121 7.71532 7.71531911
x6 3.350218 3.350215 3.352446 3.3502147 3.35021467
x7 5.286654 5.286683 5.287076 5.286654 5.28665447
g1 ðxÞ 20.0739153 20.073915 20.125692 20.073915 20.073915
g2 ðxÞ 20.197999 20.197999 20.284903 20.197999 20.198
g3 ðxÞ 20.499167 20.499172 20.34888 20.499172 20.499172
g4 ðxÞ 20.904627 20.901472 20.898038 20.904644 20.904644
g5 ðxÞ 22.9302E-06 4.0079E-13 21.3515E-03 4.0079E-13 21.4093E-10
g6 ðxÞ 3.5053E-07 24.785E-14 22.6199E-04 2.6680E-11 23.9881E-11
g7 ðxÞ 20.7025 20.7025 20.685 20.7025 20.7025
g8 ðxÞ 0 0 23.75E-04 22.744E-11 22.744E-11
g9 ðxÞ 20.583333 20.583333 20.583177 20.583333 20.583333
g10 ðxÞ 20.0513294 20.051326 20.147536 20.051326 20.0513258
g11 ðxÞ 25.8659E-06 20.010852 20.0405785 1.5542E-09 25.04E-12
f ðxÞ 2994.48245 2994.4711 3187.63049 2994.4711 2994.47107
MBA: mine blast algorithm; APSO: accelerated particle swarm optimization; IAPSO: improved adaptive particle swarm optimization; DAPSO-GA:
dynamic adaptive particle swarm optimization and genetic algorithm; MDE: modified differential evolution.
such as Gene AS1, Gene AS2, SC, ABC, MBA, aug- Statistical results provided by the previously reported
mented Lagrangian (AL) method,62 branch and bound algorithms and proposed DAPSO-GA for this design
(BB) method,63 APSO, IAPSO, and UPSO.Table 11 optimization problem are compared in terms of the
presents the comparison of optimal solutions provided worst, mean, and best solutions as well as the SD val-
by the previously reported algorithms and proposed ues and NFEs, as shown in Table 12. It is demon-
DAPSO-GA. According to the research of H Barbosa strated that the proposed DAPSO-GA, MBA, and
(September 1996, personal communication, San IAPSO are superior to other algorithms in terms of
Francisco, CA) who computes all possible gear teeth both SD and NFEs. The mean, best, and worst solu-
combinations (494 or about 5.76 million), it can be vali- tions provided by these three algorithms are at a same
dated that the optimal solutions provided by Gene level, and they stably convergence to the best solution
AS1, ABC, and the proposed DAPSO-GA are globally with similar computing efforts and SD values. Figure
best solutions. Whereas SC, MBA, APSO, and IAPSO 16 shows the convergence history of the proposed
find a different best solution as shown in Table 11. DAPSO-GA for the gear train design problem.
Table 10. Comparison of statistical results obtained from different optimization algorithms for the speed reducer design
SC 300.964736 3001.758264 2994.744241 4.0 54,456

PSO-DE 2996.348204 2996.348174 2996.348167 6.4E–06 54,350
DELC 2994.471066 2994.471066 2994.471066 1.9E–12 30,000
DEDS 2994.471066 2994.471066 2994.471066 3.6E–12 30,000
HEAA 2994.752311 2994.613368 2994.499107 7.0E–02 40,000
MDE – 2996.367220 2996.356689 8.2E–03 24,000
(m + l)-ES – 2996.348000 2996.348000 0 30,000
ABC – 2997.05800 2997.05800 0 30,000
TLBO – 2996.34817 2996.34817 0 10,000
WCA 2994.505578 2994.474392 2994.471066 7.4E–03 15,150
LCA 2994.47106614683 2994.47106614682 2994.47106614682 2.66E–12 24,000
MBA 2999.652444 2996.769019 2994.482453 1.56 6300
APSO 4443.017639 3822.640624 3187.630486 366.146 30,000
IAPSO 2994.47106615489 2994.47106614777 2994.47106614598 2.65E–09 6000
DAPSO-GA 2994.4713663 2994.4710726 2994.47106616 1.61415E–05 7320
SD: standard deviation; ABC: artificial bee colony; MBA: mine blast algorithm; APSO: accelerated particle swarm optimization; IAPSO: improved
adaptive particle swarm optimization; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm.
Figure 15. Schematic diagram of gear train.
Table 11. Comparison of optimal solutions obtained from different optimization algorithms for the gear train design optimization
problem.
DV Gene AS1 Gene AS2 SC ABC MBA AL BB APSO IAPSO DAPSO-GA
x1 49 33 43 49 43 33 45 43 43 49
x2 16 14 16 16 16 15 22 16 16 16
x3 19 17 19 19 19 13 18 19 19 19
x4 43 50 49 43 49 41 60 49 49 43
f (x) 2.7 E–12 1.4E–09 2.7E–12 2.7E–12 2.7E–12 2.1E–08 5.7E–06 2.7E–12 2.7E–12 2.7E–12
ABC: artificial bee colony; MBA: mine blast algorithm; AL: augmented Lagrangian; BB: branch and bound; APSO: accelerated particle swarm
optimization; IAPSO: improved adaptive particle swarm optimization; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic
algorithm.
Multiple disc clutch brake design problem. Figure 17 shows a minimum problem which aims to minimize its total
a schematic diagram of a multiple disc clutch brake. mass subject to geometrical constraints and constraints
The design problem of the multiple disc clutch brake is concerning shear stress, temperature, relative speed of
Zhu et al. 17
Table 12. Comparison of statistical results obtained from different optimization algorithms for the gear train design optimization
problem.
UPSO – 3.80562E-08 2.700857E–12 1.09E–07 100,000

MBA 2.062904E–08 2.471635E–09 2.700857E–12 3.94E–09 1120
SC 2.3576E–09 1.9841E–09 2.7009E–12 3.5546E–09 5000
APSO 7.072678E–06 4.781676E–07 2.700857E–12 1.44E–06 8000
IAPSO 1.82738E–08 5.492477E–09 2.700857E–12 6.36E–09 800
DAPSO-GA 2.7264505E–8 5.7898764E–09 2.70085714E–12 8.0549E–09 1438
SD: standard deviation; MBA: mine blast algorithm; APSO: accelerated particle swarm optimization; IAPSO: improved adaptive particle swarm
optimization; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm.
Figure 17. Schematic diagram of the multiple disc clutch brake.

Figure 16. Convergence history of the proposed DAPSO-GA
for the gear train design problem.
the slip–stick, and stopping time.64 The design variables This design optimization problem was previously
for this design problem are inner and outer radius (ri studied by many researchers using different optimiza-
and r0 ), disc thickness (A), actuating force (F), and tion algorithms such as non-dominated sorting genetic
number of contact surfaces (Z), which are denoted by algorithm (NSGA-II),65 TLBO, WCA, ABC, APSO,
x1 , x2 , x3 , x4 , and x5 , respectively. The objective variable and IAPSO. Table 13 presents the comparison of opti-
x4 only contains in the constraint equations (a side con- mal solutions provided by the earlier reported algo-
straint). All design variables are discrete and should be rithms and proposed DAPSO-GA. It is shown that the
selected from x1 = 60, 61, . . . , 80; x2 = 90, 91, . . . , DAPSO-GA, IAPSO, WCA, and TLBO have the same
110; x3 = 1, 1:5, . . . , 3; x4 = 600, 610, . . . , 1000; objective function value of 0.31365661, although the
x5 = 2, 3, . . . , 9. The discrete DAPSO-GA with a values of the variable x4 in the optimal solutions pro-
swarm size of 40 and maximum number of iterations of vided by these four algorithms are different. This is
100 is used to solve this optimization problem. All because x4 only needs to satisfy the constraint condi-
design variables are regarded as continuous variables tions and is independent of the objective function.
and rounded off until at the end of the iterations. Statistical results provided by the previously reported
Besides, novel techniques are applied on the discrete algorithms and DAPSO-GA for this design optimiza-
variables x3 and x4 in this algorithm: x3 is regarded as a tion problem are compared as shown in Table 14. The
continuous variable limited to the range ½2, 6 and statistical results demonstrate the superiority of the
divided by two after being rounded to an integer; x4 is proposed DAPSO-GA against all proposed optimiza-
regarded as a continuous variable limited to the range tion algorithms in both NFEs and SD value. APSO
½60, 100 and multiplied by 10 after being rounded to performs the worst among all algorithms in terms of
an integer. solution quality (mean and best solutions), SD value
Table 13. Comparison of optimal solutions obtained from different optimization algorithms for the multiple disc clutch brake design
DV NSGA-II TLBO WCA APSO IAPSO DAPSO-GA
x1 70 70 70 76 70 70
x2 90 90 90 96 90 90
x3 1.5 1.0 1.0 1.0 1.0 1.0
x4 1000 810 910 840 900 1000
x5 3 3 3 3 3 3
g1 (x) 0 0 0 0 0 0
g2 (x) 22.00 24.00 24.00 24.00 24.00 24.00
g3 (x) 0.90052816 0.91942781 0.90948063 0.92227317 0.91047534 0.90052816
g4 (x) 9790.5816 9830.3711 9809.4293 9824.2113 9811.5234 9790.5816
g5 (x) 7894.6966 7894.6966 7894.6966 7738.378 7894.6966 7894.6966
g6 (x) 60,625.0 37,706.25 49,768.75 48,848.372 48,562.5 60,625.0
g7 (x) 11,647.293 14,297.987 12,768.578 12,873.649 12,906.636 11,647.293
g8 (x) 3352.7067 702.0132 2231.4215 2126.3515 2093.3635 3352.7067
f (x) 0.4704 0.313656 0.313656 0.337181 0.31365661 0.31365661
APSO: accelerated particle swarm optimization; IAPSO: improved adaptive particle swarm optimization; DAPSO-GA: dynamic adaptive particle
swarm optimization and genetic algorithm.
Table 14. Comparison of statistical results obtained from different optimization algorithms for the multiple disc clutch brake design
ABC 0.352864 0.324751 0.313657 – .900

TLBO 0.392071 0.327166 0.313657 – .900
WCA 0.313656 0.313656 0.313656 1.69E–16 500
APSO 0.716313 0.506829 0.337181 9.767E–02 2000
IAPSO 0.313656 0.313656 0.313656 1.13E–16 400
DAPSO-GA 0.313656 0.313656 0.313656 1.129E–16 216
SD: standard deviation; ABC: artificial bee colony; APSO: accelerated particle swarm optimization; IAPSO: improved adaptive particle swarm
optimization; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm.
and NFEs, and IAPSO and WCA ranks the second

and third in terms of both SD value and NFEs,
respectively.Figure 18 shows the convergence history of
the proposed DAPSO-GA for the multiple disc clutch
brake design problem. It quickly convergences to the
best solution with less than 10 iterations due to the well
balance between exploration and exploitation in
searching process.
Pressure vessel design problem. Figure 19 presents a sche-

matic diagram of a pressure vessel. Two hemispherical
heads are capped at the two ends of the cylindrical ves-
sel. The pressure vessel design problem is first presented
by Kannan and Kramer62 and the design objective is to
minimize its total fabricating cost including materials,
forming, and welding costs. The design variables
include the shell thickness Ts , head thickness Th , inner Figure 18. Convergence history of the proposed DAPSO-GA
radius R, and cylindrical section length of the vessel L, for the multiple disc clutch brake design problem.
Zhu et al. 19
vessel design optimization problem in terms of the

worst, mean, and best solutions as well as the SD val-
ues and NFEs. It must be pointed out that the optimal
results provided by NM-PSO, WCA MBA, HPSO-GA,
and ABC are infeasible as the values of x1 and x2 are
not integer multiples of 0.0625. Hence, only the remain-
der of the earlier algorithms listed in Tables 15 and 16
are compared with the proposed algorithm. From
Tables 15 and 16, the proposed algorithm and IAPSO
Figure 19. Schematic diagram of the pressure vessel. provide better solutions compared with other algo-
rithms. Both of these two methods find the best solu-
tion with similar computation efforts (NFEs) which are
in which Ts and Th are discrete variables and are integer fewer than those of other optimization algorithms, but
multiples of 0.0625 while R and L are continuous vari- IAPSO is more robust as its SD value is smaller. Figure
ables. These four design variables are respectively 20 shows the convergence history of the proposed
denoted by x1 , x2 , x3 , and x4 in the objective function DAPSO-GA for the pressure vessel design problem.
and constraint equations as presented in Appendix 2
(section ‘‘Pressure vessel design problem’’). The pro- Rolling element bearing design problem. The schematic dia-
posed discrete DAPSO-GA with a swarm size of 25 gram of a rolling element bearing is shown in Figure
and maximum number of iterations of 500 is used to 21. The aim of the rolling element bearing design opti-
solve this optimization problem. The discrete variables mization is to maximize its dynamic loading bearing
x1 and x2 are always kept as continuous variables lim- capacity subject to the geometric and kinematic con-
ited in the range ½0:5, 99:5. Until at the end of the straints as well as the limit on the number of balls.67
optimization process, x1 and x2 are not rounded to be The design variables of this design optimization prob-
integers and multiplied by 0.0625. lem have five geometric parameters including the pitch
The pressure vessel design problem was previously diameter (Dm ), ball diameter (Db ), number of balls (Z),
studied by many researchers using different optimiza- inner and outer raceway curvature coefficients (fi and
tion algorithms including GA1, GA2, Cultural f0 ), and five other parameters only contain in the con-
Differential Evolution (CDE),66 PSO, CPSO, APSO, straint equations (KD min , KD max , e, e, and j). All
IAPSO, MBA, NM-PSO, G-QPSO, HPSO, WCA, design variables are continuous variables apart from
HPSO-GA, ABC2, and LCA. The optimal solution the number of balls (Z). These 10 design variables are
obtained from the proposed algorithm is compared respectively denoted by x1 , x2 , x3 , x4 , x5 , x6 , x7 , x8 , x9 ,
with those provided by the earlier reported algorithms and x10 , respectively. The proposed discrete DAPSO-
as listed in Table 15. Table 16 presents the comparison GA and PSO-TVAC with a swarm size of 50 and maxi-
of statistical results provided by the previously reported mum number of iterations of 200 are used to solve this
algorithms and proposed DAPSO-GA for the pressure optimization problem.
Table 15. Comparison of optimal solutions obtained from different optimization algorithms for the pressure vessel design
DV x1 x2 x3 x4 g1 (x) g2 (x) g3 (x) g4 (x) f (x)
GA1 0.8125 0.4375 42.0974 176.6540 22.01E–03 23.58E–02 224.7593 263.3460 6059.9463
GA2 0.8125 0.4375 42.0974 176.6540 20.2E–05 23.589E–02 227.8861 263.3460 6059.9463
CDE 0.8125 0.4375 42.0974 176.6376 26.67E–07 23.58E–02 23.71051 263.3623 6059.734
APSO 0.8125 0.4375 42.0974 176.6374 29.54E–07 23.59E–02 263.3626 20.9111 6059.7242
IAPSO 0.8125 0.4375 42.0974 176.6366 24.09E–13 23.58E–02 21.39E–07 263.3634 6059.7143
CPSO 0.8125 0.4375 42.0913 176.7465 21.37E–06 23.59E–04 2118.7687 263.2535 6061.0777
MBA 0.7802 0.3856 40.4292 198.4694 0 0 286.3645 241.5035 5889.3216
NM-PSO 0.8036 0.3972 41.6392 182.412 3.65E–05 3.79E–05 21.5914 257.5879 5930.3137
G-QPSO 0.8125 0.4375 42.0984 176.6372 28.79E–07 23.58E–02 20.2179 263.3628 6059.7208
WCA 0.7781 0.3846 40.3196 200.0000 22.95E–11 27.15E–11 21.35E–6 240.00 5885.3327
HPSO-GA 0.7782 0.3846 40.3196 200.0000 0 0 24.656E–10 240 5885.3328
ABC2 0.7782 0.3847 0.3211 199.9802 21.40E–06 22.84E–06 21.1418 240.0197 5885.4033
DAPSO-GA 0.8125 0.4375 42.0984 176.6366 24.09E–13 23.58E–02 21.39E–07 263.3634 6059.7143
APSO: accelerated particle swarm optimization; IAPSO: improved adaptive particle swarm optimization; CPSO: co-evolutionary particle swarm
optimization; MBA: mine blast algorithm; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm.
Table 16. Comparison of statistical results obtained from different optimization algorithms for the pressure vessel design
PSO 14,076.324 8756.6803 6693.7212 1492.5670 8000

APSO 7544.49272 6470.71568 6059.7242 326.9688 200,000
IAPSO 6090.5314 6068.7539 6059.7143 14.0057 7500
MBA 6392.5062 6200.64765 5889.3216 160.34 70,650
LCA 6090.6114 6070.5884 6059.8553 11.37534 24,000
WCA 6590.2129 6198.6172 5885.3327 213.0490 27,500
CDE 6371.0455 6085.2303 6059.7340 43.0130 204,800
GA1 6308.4970 6293.8432 6288.7445 7.4133 900,000
GA2 6469.3220 6177.2533 6059.9463 130.9297 80,000
QPSO 8017.2816 6440.3786 6059.7209 479.2671 8000
G-QPSO 7544.4925 6440.3786 6059.7208 448.4711 8000
NM-PSO 5960.0557 5946.7901 5930.3137 9.1610 80,000
HPSO 6288.6770 6099.9323 6059.7143 86.2000 81,000
CPSO 6363.8041 6147.1332 6061.0777 86.4500 240,000
HPSO-GA 5885.4864 5885.3821 5885.3328 0.049
ABC2 5895.1268 5887.5570 5885.4032 2.7453 –
DAPSO-GA 7319.0007 6267.1671 6059.7143 380.9406 9000
SD: standard deviation; PSO: particle swarm optimization; APSO: accelerated particle swarm optimization; IAPSO: improved adaptive particle swarm
optimization; MBA: mine blast algorithm; CPSO: co-evolutionary particle swarm optimization; DAPSO-GA: dynamic adaptive particle swarm
optimization and genetic algorithm.
This design optimization problem was previously

solved by many researchers using different optimiza-
tion algorithms such as GA5,67 ABC, TLBO, and
MBA. Optimal solutions given by these reported
algorithms and PSO-TVAC are compared with those
provided by the proposed DAPSO-GA in terms of
the values of design variables, objective function
value, and constraint accuracy, as detailed in Table
17. It must be emphasized that there are some errors
for the optimal solutions of GA5, TLBO, and MBA
given by Sadollah et al.9 in terms of the objective
function value, number of constraints, and constraint
accuracy, which are revised in this work as shown in
Table 17. Note that the optimal solutions provided by
GA5 and TLBO are infeasible as the fourth con-
straint g4 (x) is violated. Hence, their optimal solu-
tions are not used for later comparison. Table 18
presents the comparison of statistical results provided Figure 20. Convergence history of the proposed DAPSO-GA
by the previously reported algorithms and proposed for the pressure vessel design problem.
DAPSO-GA for the rolling element bearing design
optimization problem in terms of the worst, mean,
and best solutions as well as the SD values and proposed DAPSO-GA for the rolling element bearing
NFEs. As seen from Table 18, the proposed algo- design problem. Note that the proposed DAPSO-GA
rithm finds the best solution (81,859.80912) with the convergences fast to the best solution with less than
fewest NFEs (3650). ABC stably provides the similar 50 iterations thanks to the global optima searching
best solution (81,859.7416) with the smallest SD value technique.
but much more NFEs than the proposed algorithm.
PSO-TVAC convergences to the similar best solution
Conclusion
(81,859.7415974) with similar NFEs (3750) but much
larger SD value compared with the proposed algo- In this work, a DAPSO-GA is presented to solve con-
rithm. Figure 22 shows the convergence history of the strained engineering design optimization problems with
Zhu et al. 21
Figure 21. Schematic diagram of rolling element bearing.
Table 17. Comparison of optimal solutions obtained from different optimization algorithms for the rolling element bearing design
DV GA5 TLBO MBA PSO-TVAC DAPSO-GA
x1 125.7171 125.7191 125.7153 125.7191 125.7191

x2 21.423 21.4259 21.4233 21.4256 21.4256
x3 11 11 11 11 11
x4 0.515 0.515 0.515 0.5150 0.515
x5 0.515 0.515 0.515 0.5150 0.515
x6 0.4159 0.424266 0.488805 0.4169 0.4000
x7 0.651 0.633948 0.627829 0.7000 0.7000
x8 0.300043 0.3 0.300149 0.3000 0.3000
x9 0.0223 0.068858 0.097305 0.1000 0.0474
x10 0.751 0.799498 0.64095 0.6001 0.6000
g1 (x) 8.22E–04 21.2235E–07 5.6382E–04 3.0198E–014 21.2235E–07
g2 (x) 13.732999 13.15257 8.63025 14.8511804817 14.8512
g3 (x) 2.724000 1.52518 1.10143 3.64614410764 6.1488
g4 (x) 1.107 2.559363 22.04045 23.42559024083 23.4256
g5 (x) 0.717100 4.7191 0.7153 0.719055614672 0.7191
g6 (x) 4.857899 16.49544 23.61095 24.2809443853 11.1309
g7 (x) 0.0021288 22.999E–05 5.179283E–04 1.4797E–012 22.999E–05
g8 (x) 0 0 0 0 0
g9 (x) 0 0 0 0 0
f (x) 81,841.5108 81,859.74 81,843.68625 81,859.74159741332 81,859.80912
MBA: mine blast algorithm; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic algorithm.
different kinds of objective functions, design variables, meet the GA-selection criterion with time-varying selec-
and constraints in nature. The presented algorithm uses tion probability are adaptively selected to update their
a dynamic adaptive inertia weighting factor, which positions via a crossover and n-point mutation operator
adaptively adjusts the search velocity in optimum in each iteration process. Global best and worst posi-
searching process, to balance the exploitation (local tions of the PSO are updated according to the refined
search) and exploration (global search). In the pro- particle position generated by GA. With the three GA-
posed algorithm, GA-related operators are incorpo- related operators, the particle swarm is greatly diversi-
rated into PSO and used to refine the optimal solution fied and as a result, premature convergence is effectively
provided by the PSO. Few particles in the swarm that prevented. The promising prospect of the proposed
Table 18. Comparison of statistical results obtained from different optimization algorithms for the rolling element bearing design
GA5 – – 81,843.3 – 225,000

ABC 78,897.810 81,496.00 81,859.7416 0.69 10,000
TLBO 80,807.8551 81,438.987 81,859.7400 0.66 10,000
MBA – – 81,843.68625 211.52 15,100
PSO-TVAC 41,130.5723727 76,442.495342 81,859.7415974 11,888.8903 3750
DAPSO-GA 79,834.7818 81,066.4318 81,859.80912 742.9211 3650
SD: standard deviation; ABC: artificial bee colony; MBA: mine blast algorithm; DAPSO-GA: dynamic adaptive particle swarm optimization and genetic
algorithm.
support from the National Natural Science Foundation of

China (no. 51805339), Fundamental Research Funds for
Central Universities and the State Key Development Program
for Basic Research of China (no. 2014CB049401).
ORCID iDs
Hao Zhu https://orcid.org/0000-0003-3910-2947
Weidong Zhu https://orcid.org/0000-0003-2707-2533
References
1. Guedria NB. Improved accelerated PSO algorithm for
mechanical engineering optimization problems. Appl Soft
Comput 2016; 40: 455–467.
2. Kennedy J and Eberhart R. Particle swarm optimization.
In: IEEE international conference on neural networks
(ICNN), Perth, WA, Australia, vol. 4, 27 November–1
December 1995, pp.1942–1948. New York: IEEE.
Figure 22. Convergence history of the proposed DAPSO-GA 3. Holland JH. Outline for a logical theory of adaptive sys-
for the rolling element bearing design problem. tems. J ACM 1962; 9: 297–314.
4. Bagley JD. The behavior of adaptive systems which employ
genetic and correlation algorithms. Ann Arbor, MI: Dis-
DAPSO-GA for engineering constrained optimization sertation Abstracts International, University of Michi-
problems is evaluated by solving nine different bench- gan, 1967.
mark mechanical engineering design optimization prob- 5. Goldberg DE. Genetic algorithms in search, optimization
lems with continuous, discrete, or mixed design and machine learning. Boston, MA: Addison-Wesley
Longman Publishing Co., Inc., 1989.
variables. For most of the considered mechanical engi-
6. Yang XS. Firefly algorithms for multimodal optimiza-
neering design optimization problems, statistical results tion. In: Watanabe O and Zeugmann T (eds) Stochastic
show that the proposed DAPSO-GA convergences to algorithms: foundations and applications, vol. 5792. Ber-
the best or similar solution with the smallest SD values lin: Springer, 2009, pp.169–178.
and lowest computation efforts (NFEs) against other 7. Dorigo M, Birattari M and Stutzle T. Ant colony optimi-
meta-heuristic algorithms. zation artificial ants as a computational intelligence tech-
nique. IEEE Comput Intell M 2006; 1: 28–39.
Declaration of conflicting interests 8. Karaboga D and Basturk B. Artificial bee colony (ABC)
The author(s) declared no potential conflicts of interest with optimization algorithm for solving constrained optimiza-
respect to the research, authorship, and/or publication of this tion. In: Melin P, Castillo O, Aguilar LT, et al. (eds)
article. Foundations of fuzzy logic and soft computing. IFSA 2007
(Lecture Notes in Computer Science, vol. 4529). Berlin:
Funding Springer, 2007, pp.789–798.
9. Sadollah A, Bahreininejad A, Eskandar H, et al. Mine
The author(s) disclosed receipt of the following financial sup- blast algorithm: a new population based algorithm for
port for the research, authorship, and/or publication of this solving constrained engineering optimization problems.
article: The authors gratefully acknowledge to the financial Appl Soft Comput 2013; 13: 2592–2612.
Zhu et al. 23
10. Kirkpatrick S, Gelatt CD and Vecchi MP. Optimization 26. Worasucheep C. Solving constrained engineering optimi-
by simulated annealing. Science 1983; 220: 671–680. zation problems by the constrained PSO-DD. In: 2008
11. Garg H. An efficient biogeography based optimization 5th international conference on Electrical Engineering/
algorithm for solving reliability optimization problems. Electronics, Computer, Telecommunications and Informa-
Swarm Evol Comput 2015; 24: 1–10. tion Technology, Krabi, Thailand, 14–17 May 2008, pp.5–
12. Gouttefarde M and Gosselin CM. Analysis of the 8. New York: IEEE.
wrench-closure workspace of planar parallel cable-driven 27. Yang XS. Engineering optimization: an introduction with
mechanisms. IEEE T Robot 2006; 22: 434–445. metaheuristic applications. Hoboken, NJ: John Wiley &
13. Perez RE and Behdinan K. Particle swarm optimization Sons, Inc., 2010.
in structural design. In: Chan FTS and Tiwari MK (eds) 28. Yang XS, Deb S and Fong S. Accelerated particle swarm
Swarm intelligence: focus on ant and particle swarm opti- optimization and support vector machine for business
mization. Vienna: Intech Education and Publishing, 2007, optimization and applications. In: Fong S (ed.) Net-
pp.373–394. worked digital technologies. NDT 2011. Communications
14. Hassan R, Cohanim BE and de Weck OL. A comparison in computer and information science, vol. 136. Berlin:
of particle swarm optimization and the genetic algorithm. Springer, 2011, pp.53–66.
In: 46th AIAA/ASME/ASCE/AHS/ASC structures, struc- 29. Novitasari D, Cholissodin I and Mahmudy WF. Hybri-
tural dynamics, and materials conference, Austin, TX, 18– dizing PSO with SA for optimizing SVR applied to soft-
21 April 2005, AIAA paper no. 2005–1897. Reston, VA: ware effort estimation. TELKOMNIKA 2016; 14:
American Institute of Aeronautics and Astronautics. 245–253.
15. Mortazavi A and Toğan V. Simultaneous size, shape, and 30. He Q and Wang L. A hybrid particle swarm optimization
topology optimization of truss structures using integrated with a feasibility-based rule for constrained optimization.
particle swarm optimizer. Struct Multidiscip O 2016; 54: Appl Math Comput 2007; 186: 1407–1422.
715–736. 31. Wang J and Yin Z. A ranking selection-based particle
16. Joshua TB, Xin J and Sunil KA. Optimal design of cable- swarm optimizer for engineering design optimization
driven manipulators using particle swarm optimization. J problems. Struct Multidiscip O 2008; 37: 131–147.
Vib Acoust 2016; 8: 041003. 32. Lei J-J and Li J. A modified particle swarm optimization
17. Esmin AAA and Matwin S. HPSOM: a hybrid particle for practical engineering optimization. In: 2009 fifth inter-
swarm optimization algorithm with genetic mutation. Int national conference on natural computation, vol. 3, Tian-
J Innov Comput I 2013; 9: 1919–1934. jin, China, 14–16 August 2009, pp.177–180. New York:
18. Lovbjerg M and Krink T. Extending particle swarm opti- IEEE.
mizers with self-organized critically. In: Proceedings of 33. Coelho LDS. Gaussian quantum-behaved particle swarm
the 2002 congress on evolutionary computation. CEC’02, optimization approaches for constrained engineering
vol. 2, Honolulu, HI, 12–17 May 2002, pp.1588–1593. design problems. Expert Syst Appl 2010; 37: 1676–1683.
New York: IEEE. 34. Mahmoodabadi MJ and Bisheban M. An online optimal
19. Shi Y and Eberhart R. Empirical study of particle swarm linear state feedback controller based on MLS approxi-
optimization. In: Proceedings of the 1999 congress on evo- mations and a novel straightforward PSO algorithm. T I
lutionary computation-CEC99, Washington, DC, 6–9 July Meas Control 2014; 36: 1132–1142.
1999, pp.1945–1950. New York: IEEE. 35. Zhu H, Hu YM, Zhu WD, et al. Optimal design of an
20. Eberhart RC and Shi Y. Tracking and optimizing auto-tensioner in an automotive belt drive system via a
dynamic systems with particle swarms. In: Proceedings of dynamic adaptive PSO-GA. J Mech Design 2017; 139:
the 2001 congress on evolutionary computation, Seoul,
093302.
Korea, 27–30 May 2001, pp.94–97. New York: IEEE.
36. Zhang M, Luo W and Wang X. Differential evolution
21. Clerc M. The Swarm and The Queen: towards a determi-
with dynamic stochastic selection for constrained optimi-
nistic and adaptive particle swarm optimization. In: Pro-
zation. Inform Sciences 2008; 178: 3043–3074.
ceedings of the 1999 congress on evolutionary computation,
37. Coello CAC. Theoretical and numerical constraint-
Washington, DC, 6–9 July 1999. New York: IEEE.
handling techniques used with evolutionary algorithms: a
22. Ratnaweera A, Halgamuge SK and Harry CW. Self-orga-
survey of the state of the art. Comput Method Appl M
nizing hierarchical particle swarm optimizer with time-
2002; 191: 1245–1287.
varying acceleration coefficients. IEEE T Evolut Comput
38. Mezura-Montes E and Coello CAC. A simple multimem-
2004; 8: 240–255.
bered evolution strategy to solve constrained optimiza-
23. Shi Y and Eberhart RC. Parameter selection in particle
tion problems. IEEE T Evolut Comput 2005; 9: 1–17.
swarm optimization. In: Porto VW, Saravanan N, Waa-
39. Garg H. A hybrid PSO-GA algorithm for constrained
gen D, et al. (eds) Evolutionary programming VII. Berlin:
optimization problems. Appl Math Comput 2016; 274:
Springer, 1998, pp.591–600.
292–305.
24. He Q and Wang L. An effective co-evolutionary particle
40. Garg H. Solving structural engineering design optimiza-
swarm optimization for constrained engineering design
tion problems using an artificial bee colony algorithm. J
problems. Eng Appl Artif Intel 2007; 20: 89–99.
Ind Manag Optim 2014; 10: 777–794.
25. Krohling RA and Coelho LDS. Coevolutionary particle
41. Coello CAC. Use of a self-adaptive penalty approach for
swarm optimization using Gaussian distribution for sol-
engineering optimization problems. Comput Ind 2000; 41:
ving constrained optimization problems. IEEE T Syst
Man Cy B 2006; 36: 1407–1416. 113–127.
42. Coello CAC and Montes EM. Constraint-handling in mechanical design optimization problems. Comput Aided
genetic algorithms through the use of dominance-based Design 2011; 43: 303–315.
tournament selection. Adv Eng Inform 2002; 16: 193–203. 58. Coello CAC. Treating constraints as objectives for single-
43. Eskandar H, Sadollah A, Bahreininejad A, et al. Water objective evolutionary optimization. Eng Optimiz 2000;
cycle algorithm—a novel metaheuristic optimization 32: 275–308.
method for solving constrained engineering optimization 59. Siddall JN. Optimal engineering design: principles and
problems. Comput Struct 2012; 110: 151–166. applications. New York: Marcel Dekker, 1982.
44. Lampinen J. A constraint handling approach for the dif- 60. Deb K and Goyal M. Optimizing engineering designs
ferential evolution algorithm. In: Proceedings of the 2002 using a combined genetic search. In: Proceedings of the
congress on evolutionary computation. CEC’02, Honolulu, seventh international conference on genetic algorithms,
HI, 12–17 May 2002, pp.1468–1473. New York: IEEE. East Lansing, MI, 19–23 July 1997, pp.521–528. San
45. Wang L and Li LP. An effective differential evolution Francisco, CA: Morgan Kaufmann Publishers Inc.
with level comparison for constrained engineering design. 61. Mezura-Montes E, Velázquez-Reyes J and Coello CAC.
Struct Multidiscip O 2010; 41: 947–963. Modified differential evolution for constrained optimiza-
46. Zahara E and Kao YT. Hybrid Nelder–Mead simplex tion. In: 2006 IEEE international conference on evolution-
search and particle swarm optimization for constrained ary computation, Vancouver, BC, Canada, 16–21 July
engineering design problems. Expert Syst Appl 2009; 36: 2006, pp.25–32. New York: IEEE.
3880–3886. 62. Kannan BK and Kramer SN. An augmented Lagrange
47. Wang Y, Cai Z, Zhou Y, et al. Constrained optimization multiplier based method for mixed integer discrete con-
based on hybrid evolutionary algorithm and adaptive tinuous optimization and its applications to mechanical
constraint-handling technique. Struct Multidiscip O 2009; design. J Mech Design 1994; 116: 405–411.
37: 395–413. 63. Sandgren E. Nonlinear integer and discrete program-
48. Davoodi E, Hagh MT and Zadeh SG. A hybrid improved ming in mechanical design. J mech Design 1990; 112:
quantum-behaved particle swarm optimization-simplex 223–229.
method (IQPSOS) to solve power system load flow prob- 64. Osyczka A. Evolutionary algorithms for single and multi-
lems. Appl Soft Comput 2014; 21: 171–179. criteria design optimization (Studies in Fuzziness and Soft
49. Ray T and Liew KM. Society and civilization: an optimi- Computing). Heidelberg: Physica-Verlag, 2002.
zation algorithm based on the simulation of social beha- 65. Deb K and Srinivasan A. Innovization: innovating design
vior. IEEE T Evolut Comput 2003; 7: 386–396. principles through optimization. In: Proceedings of the
50. Kashan AH. An efficient algorithm for constrained glo- 8th annual conference on genetic and evolutionary compu-
bal optimization and application to mechanical engineer- tation, Seattle, WA, 8–12 July 2006, pp.1629–1636. New
ing design: league championship algorithm (LCA). York: ACM Press.
Comput Aided Design 2011; 43: 1769–1792. 66. Becerra RL, Coello CAC. Cultured differential evolution
51. Parsopoulos KE and Vrahatis MN. Unified particle for constrained optimization. Computer Methods in
swarm optimization for solving constrained engineering Applied Mechanics and Engineering 2006; 195: 4303–4022.
optimization problems. In: Wang L, Chen K and Ong YS 67. Gupta S, Tiwari R and Nair SB. Multi-objective design
(eds) Advances in natural computation. Berlin: Springer, optimization of rolling bearings using genetic algorithms.
2005, pp.582–591. Mech Mach Theory 2007; 42: 1418–1443.
52. Mezura-Montes E and Coello CAC. Useful infeasible
solutions in engineering optimization with evolutionary Appendix 1
algorithms. In: Gelbukh A, de Albornoz Á and Tera-
shima-Marı́n H (eds) MICAI 2005: advances in artificial Tension/compression spring design problem
intelligence (Lecture Notes in Computer Science, vol.
3789). Berlin: Springer, 2005, pp.652–662.
53. Liu H, Cai Z and Wang Y. Hybridizing particle swarm Minimize f ðxÞ = (x3 + 2)x2 x21
optimization with differential evolution for constrained Subject to
numerical and engineering optimization. Appl Soft Com-
put 2010; 10: 629–640. x32 x3
g1 ðxÞ = 1 ł0
54. Henendz S. Multiobjective structural optimization. In: 71785x41
Kodoyalam S and Saxena M (eds) Geometry and optimi- 4x22 x1 x2 1
zation techniques for structural design. Amsterdam: Else- g2 ðxÞ = 3 + 1ł0
vier, 1994, pp.341–362. 12566 x2 x1 x1 4 5108x21
55. Ray T and Saini P. Engineering design optimization using 140:45x1
g3 ðxÞ = 1 ł0
a swarm with an intelligent information sharing among x22 x3
individuals. Eng Optimiz 2001; 33: 735–748. x1 + x2
56. Deb K. Optimal design of a welded beam via genetic g4 ðxÞ = 1ł0
1:5
algorithms. AIAA J 1991; 29: 2013–2015.
57. Rao RV, Savsani VJ and Vakharia DP. Teaching–learn- where 0:05 ł x1 ł 20, 0:25 ł x2 ł 1:3, and
ing-based optimization: a novel method for constrained 2:00 ł x3 ł 15:00.
Zhu et al. 25
Symmetric three-bar truss design problem

pffiffiffi
Minimize f (x) = (2 2x1 + x2 )l
Subject to
pffiffiffi pffiffiffi
2 x1 + x2 x2 2
g1 ðxÞ = pffiffiffi 2 P s ł 0, g2 ðxÞ = pffiffiffi 2 P s ł 0, g3 ðxÞ = pffiffiffi P sł0
2x1 + 2x1 x2 2x1 + 2x1 x2 2x1 + 2x2
where 0 ł x1 ł 1, 0 ł x2 ł 1, l = 100 cm, P = 2 KN=cm2 , and s = 2 KN=cm2 .

Welded beam design
Minimize f ðxÞ = 1:10471x21 x2 + 0:04811x3 x4 (14 + x2 )

Subject to
g1 ðxÞ = tðxÞ tmax ł 0, g2 ðxÞ = sðxÞ smax ł 0, g3 ðxÞ = x1 x4 ł 0
g4 ðxÞ = 0:10471x21 + 0:04811x3 x4 ð14 + x2 Þ 5 ł 0, g5 ðxÞ = 0:125 x1 ł 0
g6 ðxÞ = dðxÞ 0:25 ł 0, g7 ðxÞ = P Pc ðxÞ ł 0
where 0:1 ł x1 , x4 ł 2 and 0:1 ł x2 , x3 ł 10.

Belleville spring design problem

Minimize f ðxÞ = 0:07075p D2e D2i t
Subject to
4Edmax
g1 ðxÞ = S ½bðh dmax =2Þ + gt ø 0
ð1 m2 ÞaD2e
4Edmax
g2 ðxÞ = 2 2
ðh dmax =2Þðh dmax Þt + t3 Pmax ø 0
ð1 m ÞaDe
g3 ðxÞ = dl dmax ø 0, g4 ðxÞ = H h t ø 0, g5 ðxÞ = Dmax De ø 0
h
g6 ðxÞ = De Di ø 0, g7 ðxÞ = 0:3 ø0
De Di
where

6 K 1 2
a=
p ln K K

6 K 1
b= 1
p ln K ln K

6 K1
g=
p ln K 2
Pmax = 5400 lb, dmax = 0:2 in, S = 200 kpsi, E = 30E06 psi, m = 0:3, H = 2 in, Dmax = 12:01 in, K = De =Di ,
dl = f (a)a, a = h=t.
Values of vary as detailed in Table 19.
Table 19. Variation of f (a) with a.
a <1:4 1.5 1.6 1.7 1.8 1.9 2.0 2.1 2.2 2.3 2.4 2.5 2.6 2.7 ø 2:8
f (a) 1 0.85 0.77 0.71 0.66 0.63 0.60 0.58 0.56 0.55 0.53 0.52 0.51 0.51 0.50
Appendix 2
Speed reducer design problem

Minimize f ðxÞ = 0:7854x1 x22 3:3333x23 + 14:9334x3 43:0934 1:508x1 x26 + x27

+ 7:4777 x36 + x37 + 0:7854 x4 x26 + x5 x27
Subject to
27 397:5 1:93x34 1:93x35
g 1 ðx Þ = 2
1 ł 0, g2 ðxÞ = 2
1 ł 0, g3 ð xÞ = 4
1 ł 0, g4 ð xÞ = 1ł0
x1 x2 x3 x1 x2 x23 x2 x6 x 3 x2 x47 x3
h i0:5 h i0:5
ð745x4 =x2 x3 Þ2 + 16:9 3 106 ð745x5 =x2 x3 Þ2 + 157:5 3 106
g 5 ðx Þ = 1 ł 0, g6 ðxÞ = 1ł0
110x36 85x37
5x2 x1
g7 ðxÞ = x2 x3 =40 1 ł 0, g8 ðxÞ = 1 ł 0, g9 ðxÞ = 1ł0
x1 12x2
1:5x6 + 1:9 1:1x7 + 1:9
g10 ðxÞ = 1 ł 0, g11 ðxÞ = 1ł0
x4 x5
where 2:6 ł x1 ł 3:6; 0:7 ł x2 ł 0:8; 17 ł x3 ł 28; 7:3 ł x4 ł 8:3; 7:3 ł x5 ł 8:3 2:9 ł x6 ł 3:9; 5:0 ł x7 ł 5:5.
Gear train design problem

2
1 x2 x3
Minimize f ðx Þ =
6:931 x1 x4
Subject to
12 ł xi ł 60, i = 1, 2, 3, 4
Multiple disc clutch brake design problem

Minimize f ðxÞ = p r02 ri2 ðZ + 1Þrt
Subject to
g1 ðxÞ = r0 ri Dr ø 0, g2 ðxÞ = lmax ðZ + 1Þðt + dÞ ø 0, g3 ðxÞ = Pmax Prz ø 0
g4 ðxÞ = Pmax vsr max Prz vsr ø 0, g5 ðxÞ = vsr max vsr ø 0, g6 ðxÞ = Tmax T ø 0
g7 ðxÞ = Mh sMs ø 0, g8 ðxÞ = T ø 0
where
2 r3 ri3
Mh = mFZ 02
3 r0 ri2
2 F
Prz =
3 p r02 ri2
3
2pn r0 ri3
vrz =
90 r02 ri2
Iz pn
T=
30 Mh Mf
Dr = 20 mm, Iz = 55 kgm2 , Pmax = 1 MPa, Fmax = 1000 N, Tmax = 15 s, m = 0:5, s = 1:5, Ms = 40 N m,

Mf = 3 N m, n = 250 r=min, vsr max = 10 m=s, lmax = 30 mm, 60 mm ł ri ł 80 mm, 90 mm ł r0 ł 110 mm,
1:5 mm ł t ł 3 mm, 600 N ł F ł 1000 N, 2 ł Z ł 9.
Zhu et al. 27
Pressure vessel design problem
Minimize f ðxÞ = 0:6224x1 x3 x4 + 1:7781x2 x23 + 3:166x21 x4 + 19:84x21 x3

Subject to
g1 ðxÞ = x1 + 0:0193x3 ł 0
g2 ðxÞ = x2 + 0:00954x3 ł 0
4
g3 ðxÞ = px23 x4 px33 + 1296000 ł 0
3
g4 ðxÞ = x4 240 ł 0
where 1 3 0:0625 ł x1 , x2 ł 99 3 0:0625 and 10 ł x3 , x4 ł 200.
Rolling element bearing design problem

(
fc Z 2=3 D1:8
b , if Db ł 25:4 mm
Maximize f ðx Þ = C d =
3:647fc Z 2=3 D1:4
b , else
Subject to
f0
g1 ðxÞ = 1
Z +1ø0
2sin ðDb =Dm Þ
g2 ðxÞ = 2Db KD min ðD d Þ ø 0
g3 ðxÞ = KD min ðD d Þ 2Db ø 0
g4 ðxÞ = jBw Db ø 0
g5 ðxÞ = Dm 0:5ðD + d Þ ø 0
g6 ðxÞ = ð0:5 + eÞðD + d Þ Dm ø 0
g7 ðxÞ = 0:5ðD Dm Db Þ eDb ø 0
g8 ðxÞ = fi 0:515 ø 0
g9 ðxÞ = f0 0:515 ø 0
where
8 " 9
< 1:72 0:41 #10=3 =0:3 " 0:3 # 0:41
1g fi ð2f0 1Þ g ð1 g Þ1:39 2fi
fc = 37:91 1 + 1:04
: 1+g f0 ð2fi 1Þ ; ð1 + lÞ1=3 2fi 1
!
½ðD d Þ=2 3T =42 + ðD=2 T =4 Db Þ2 ðd=2 + T =4Þ2
f0 = 2p 2 cos1
2½ðD d Þ=2 3T =4ðD=2 T =4 Db Þ
g = Db =Dm , fi = ri =Db , f0 = r0 =Db , T = D d 2Db , D = 160, d = 90, Bw = 30, 0:4 ł KD min ł 0:5,
0:6 ł KD max ł 0:7, 0:3 ł e ł 0:4, 0:02 ł e ł 0:1, 0:6 ł j ł 0:85, 0:5(D + d) ł Dm ł 0:6(D + d),
0:15(D d) ł Db ł 0:45(D d).

A Dynamic Adaptive Particle Swarm Optimization and

Uploaded by

A Dynamic Adaptive Particle Swarm Optimization and

Uploaded by

Advances in Nonlinear Dynamics and Vibrations on Mechanical Systems - Research Article

Advances in Mechanical Engineering

Hao Zhu1,2 , Yumei Hu3 and Weidong Zhu4

Date received: 18 August 2018; accepted: 12 December 2018

Handling Editor: Yunn-Lin Hwang

Introduction especially when the objective functions and constraints

GA-related algorithm Crossover and mutation operator. When the GA-selection

Figure 2. Procedures of the n-point mutation operator.

Figure 3. Flowchart of the DAPSO-GA.

fi ðxi , lÞ = fi ðround ðxi Þ, lÞ ð12Þ

Figure 4. Schematic of the tension/compression spring.

off approach is called a discrete DAPSO-GA and is Constrained engineering optimization

previously proposed algorithms. Statistical optimization

DV x1 x2 x3 g1 (x) g2 (x) g3 (x) g4 (x) f (x)

IAPSO 0.051685 0.356629 11.294175 21.97E–10 24.64E–10 24.05361 21.091686 0.01266523

Algorithms Worst Mean Best SD NFEs

IAPSO 0.01782864 0.013676527 0.01266523 1.57E–3 2000

Figure 6. Inertia weight versus number of iterations.

DV Hernendez DSS-MDE SC SIIS PSO-TVAC DAPSO-GA

x1 0.788 0.7886751359 0.788621037 0.795 0.7887058767 0.7886769887

Algorithms M S Worst Mean Best SD NFEs

DSS-MDE 10 300 263.8958498 263.8958436 263.8958434 9.72E–7 15,000

DV GA3 IAPSO APSO MBA WCA NM-PSO CAEP

x1 0.2489 0.2057296 0.202701 0.205729 0.205728 0.205830 0.205700

DV CPSO GA1 GA2 HPSO-GA ABC2 DAPSO-GA

x1 0.202369 0.2088 0.205986 0.2057296 0.2057245 0.205728318

Algorithm Worst Mean Best SD NFEs

GA3 2.64583 2.39203 (median) 2.38119 – 320,080

Figure 11. Schematic diagram of the Belleville disc spring.

reported algorithms and proposed DAPSO-GA. Note

DV Coello Gene AS1 Gene AS2 Siddall TLBO MBA DAPSO-GA

x1 0.208 0.205 0.210 0.204 0.204143 0.204143 0.20414335

Algorithm Worst Mean Best SD NFEs

ABC 2.104297 1.995475 1.979675 0.07 150,000

DAPSO-GA convergence faster than GA, while the

Constrained engineering optimization problems

Figure 13. Schematic diagram of speed reducer.

the reported algorithms including DEDS, DELC,

DV DEDS DELC HEAA MDE PSO-DE WCA

x1 3.5 3.5 3.500022 3.50001 3.5 3.5

DV MBA LCA APSO IAPSO DAPSO-GA

x1 3.5 3.5 3.501313 3.5 3.5

Algorithm Worst Mean Best SD NFEs

SC 300.964736 3001.758264 2994.744241 4.0 54,456

Figure 15. Schematic diagram of gear train.

DV Gene AS1 Gene AS2 SC ABC MBA AL BB APSO IAPSO DAPSO-GA

Algorithm Worst Mean Best SD NFEs

UPSO – 3.80562E-08 2.700857E–12 1.09E–07 100,000

Figure 17. Schematic diagram of the multiple disc clutch brake.

DV NSGA-II TLBO WCA APSO IAPSO DAPSO-GA

Algorithm Worst Mean Best SD NFEs

ABC 0.352864 0.324751 0.313657 – .900

and NFEs, and IAPSO and WCA ranks the second

Pressure vessel design problem. Figure 19 presents a sche-

vessel design optimization problem in terms of the

DV x1 x2 x3 x4 g1 (x) g2 (x) g3 (x) g4 (x) f (x)

Algorithm Worst Mean Best SD NFEs

PSO 14,076.324 8756.6803 6693.7212 1492.5670 8000

This design optimization problem was previously

Figure 21. Schematic diagram of rolling element bearing.

DV GA5 TLBO MBA PSO-TVAC DAPSO-GA

x1 125.7171 125.7191 125.7153 125.7191 125.7191

Algorithm Worst Mean Best SD NFEs

GA5 – – 81,843.3 – 225,000

support from the National Natural Science Foundation of

Symmetric three-bar truss design problem

where 0 ł x1 ł 1, 0 ł x2 ł 1, l = 100 cm, P = 2 KN=cm2 , and s = 2 KN=cm2 .

Minimize f ðxÞ = 1:10471x21 x2 + 0:04811x3 x4 (14 + x2 )

where 0:1 ł x1 , x4 ł 2 and 0:1 ł x2 , x3 ł 10.

Table 19. Variation of f (a) with a.

Speed reducer design problem

Gear train design problem

Multiple disc clutch brake design problem

Dr = 20 mm, Iz = 55 kgm2 , Pmax = 1 MPa, Fmax = 1000 N, Tmax = 15 s, m = 0:5, s = 1:5, Ms = 40 N m,

Pressure vessel design problem

Minimize f ðxÞ = 0:6224x1 x3 x4 + 1:7781x2 x23 + 3:166x21 x4 + 19:84x21 x3