Page 57 - Fister jr., Iztok, and Andrej Brodnik (eds.). StuCoSReC. Proceedings of the 2016 3rd Student Computer Science Research Conference. Koper: University of Primorska Press, 2016
P. 57
rid Cuckoo Search for constraint engineering design
optimization problems
Uroš Mlakar
University of Maribor
Smetanova 17
Maribor, Slovenia
uros.mlakar@um.si
ABSTRACT is organized as follows. In Section 2 some of related work,
mostly that of SI-based algorithms is presented. Section 3 is
This paper investigates, how the hybrid self-adaptive Cuckoo dedicated to Cuckoo Search (CS) and the used self-adaptive
Search algorithm (HSA-CS) behaves, when confronted with hybrid Cuckoo Search algorithm, where an emphasis is on
constraint engineering design optimization problems. These describing the main differences from the original CS. Section
problems are commonly found in literature, namely: welded 4 deals with describing the design optimization problems,
beam, pressure vessel design, speed reducer, and spring de- and then in Section 5 the obtained results are presented. In
sign. The obtained results are compared to those found in Section 6 the results obtained by the HSA-CS is compared to
literature, where the HSA-CS achieved better or comparable those in the literature, and the paper is concluded in Section
results. Based on results, we can conclude, that the HSA-CS 7.
is suitable for use in real-life engineering applications.
Keywords 2. RELATED WORK
design optimization, hybridization, cuckoo search, self-adaptation Since the HSA-CS belongs to the SI-based algorithms, it
would only be reasonable to review the literature from this
1. INTRODUCTION point of view. Akay and Karaboga [1] presented an artificial
There is a increasing rate in the research community for de- bee colony (ABC) algorithm, with a very simple constraint
veloping new constrained optimization algorithms. There- handling method. This method is biased to choose feasi-
fore suitable problems must be used, to show the effective- ble solutions rather than those, which are infeasible. Gan-
ness, efficiency and convergence of these new algorithms. domi et al. [7] use a bat algorithm for solving constraint
Such problems are usually mathematical problems like the optimization problems. Their results indicate that their
CEC competition problems, but also engineering design op- method obtained better results, compared to those in lit-
timization problems are adopted in the specialized litera- erature. Another ABC algorithm was proposed by Braje-
ture. Many researchers have studied these problems, by ap- vic and Tuba [3]. The upgraded ABC algorithm enhances
plying a wide range of different optimization methods such fine-tuning characteristics of the modification rate parame-
as Quadratic Programming [5], Simulated Annealing [12], ter and employs modified scout bee phase of the ABC algo-
Genetic Algorithms [9], and Swarm Intelligence [2, 3, 6, 7, rithm. Baykasoglu and Ozsoydan [2] presented an adaptive
1]. The listed algorithms are among the most used in the firefly, enhanced with chaos mechanisms. The adaptivity is
literature. The design optimization problems usually have a focused on on the search mechanism and adaptive param-
non-linear objective function and constraints, while the de- eter settings. They report that some best results found in
sign variables are often a combination of discrete and con- literature, were improved with their method. Bulatovi´c [4]
tinuous. The hardest part of finding the optimal solution for applied the improved cuckoo search (ICS) for solving con-
these problems is directly related to the constraints, which strained engineering problems, which produces better results
are imposed on the problem. than the original cuckoo search (CS). Their improvements
lie in the dynamic changing of the parameters of probability
Since SI based algorithms are getting a lot of attention in and step size. Yang et al. [11] utilized a multi-objective CS
the past couple of years, our aim was to test the behaviour (MOCS) for the beam design problem and disc brake prob-
of a novel hybrid self-adaptive Cuckoo Search [8] (HSA-CS) lems. They conclude that the proposed MOCS is efficient
on the design optimization problems. The rest of the paper on problems with complex constraints.
3. CUCKOO SEARCH
Cuckoo search is a stochastic population-based optimization
algorithm proposed by Yang and Deb in 2009 [10]. It belongs
in the SI-based algorithm family, and it is inspired by the
natural behaviour of some cuckoo species in nature. To trap
the behavior of cuckoos in nature and adapt it to be suitable
for using as a computer program the authors [10] idealized
three rules:
StuCoSReC Proceedings of the 2016 3rd Student Computer Science Research Conference 57
Ljubljana, Slovenia, 12 October
optimization problems
Uroš Mlakar
University of Maribor
Smetanova 17
Maribor, Slovenia
uros.mlakar@um.si
ABSTRACT is organized as follows. In Section 2 some of related work,
mostly that of SI-based algorithms is presented. Section 3 is
This paper investigates, how the hybrid self-adaptive Cuckoo dedicated to Cuckoo Search (CS) and the used self-adaptive
Search algorithm (HSA-CS) behaves, when confronted with hybrid Cuckoo Search algorithm, where an emphasis is on
constraint engineering design optimization problems. These describing the main differences from the original CS. Section
problems are commonly found in literature, namely: welded 4 deals with describing the design optimization problems,
beam, pressure vessel design, speed reducer, and spring de- and then in Section 5 the obtained results are presented. In
sign. The obtained results are compared to those found in Section 6 the results obtained by the HSA-CS is compared to
literature, where the HSA-CS achieved better or comparable those in the literature, and the paper is concluded in Section
results. Based on results, we can conclude, that the HSA-CS 7.
is suitable for use in real-life engineering applications.
Keywords 2. RELATED WORK
design optimization, hybridization, cuckoo search, self-adaptation Since the HSA-CS belongs to the SI-based algorithms, it
would only be reasonable to review the literature from this
1. INTRODUCTION point of view. Akay and Karaboga [1] presented an artificial
There is a increasing rate in the research community for de- bee colony (ABC) algorithm, with a very simple constraint
veloping new constrained optimization algorithms. There- handling method. This method is biased to choose feasi-
fore suitable problems must be used, to show the effective- ble solutions rather than those, which are infeasible. Gan-
ness, efficiency and convergence of these new algorithms. domi et al. [7] use a bat algorithm for solving constraint
Such problems are usually mathematical problems like the optimization problems. Their results indicate that their
CEC competition problems, but also engineering design op- method obtained better results, compared to those in lit-
timization problems are adopted in the specialized litera- erature. Another ABC algorithm was proposed by Braje-
ture. Many researchers have studied these problems, by ap- vic and Tuba [3]. The upgraded ABC algorithm enhances
plying a wide range of different optimization methods such fine-tuning characteristics of the modification rate parame-
as Quadratic Programming [5], Simulated Annealing [12], ter and employs modified scout bee phase of the ABC algo-
Genetic Algorithms [9], and Swarm Intelligence [2, 3, 6, 7, rithm. Baykasoglu and Ozsoydan [2] presented an adaptive
1]. The listed algorithms are among the most used in the firefly, enhanced with chaos mechanisms. The adaptivity is
literature. The design optimization problems usually have a focused on on the search mechanism and adaptive param-
non-linear objective function and constraints, while the de- eter settings. They report that some best results found in
sign variables are often a combination of discrete and con- literature, were improved with their method. Bulatovi´c [4]
tinuous. The hardest part of finding the optimal solution for applied the improved cuckoo search (ICS) for solving con-
these problems is directly related to the constraints, which strained engineering problems, which produces better results
are imposed on the problem. than the original cuckoo search (CS). Their improvements
lie in the dynamic changing of the parameters of probability
Since SI based algorithms are getting a lot of attention in and step size. Yang et al. [11] utilized a multi-objective CS
the past couple of years, our aim was to test the behaviour (MOCS) for the beam design problem and disc brake prob-
of a novel hybrid self-adaptive Cuckoo Search [8] (HSA-CS) lems. They conclude that the proposed MOCS is efficient
on the design optimization problems. The rest of the paper on problems with complex constraints.
3. CUCKOO SEARCH
Cuckoo search is a stochastic population-based optimization
algorithm proposed by Yang and Deb in 2009 [10]. It belongs
in the SI-based algorithm family, and it is inspired by the
natural behaviour of some cuckoo species in nature. To trap
the behavior of cuckoos in nature and adapt it to be suitable
for using as a computer program the authors [10] idealized
three rules:
StuCoSReC Proceedings of the 2016 3rd Student Computer Science Research Conference 57
Ljubljana, Slovenia, 12 October
