Back Issues 11-20

SELECTING PROPER POPULATION-BASED METAHEURISTIC ALGORITHMS FOR SOLVING THE DIFFERENT MODALS OF UNCONSTRAINED OPTIMIZATION PROBLEMS

Shou-Cheng Hsiung*, Department of Industrial Management, I-Shou University, Taiwan

*Corresponding Author: This email address is being protected from spambots. You need JavaScript enabled to view it.

I-Ming Chao, I-Shou University, Taiwan (ROC), This email address is being protected from spambots. You need JavaScript enabled to view it.

Hsiang-Chen Hsu, I-Shou University, Taiwan (ROC), This email address is being protected from spambots. You need JavaScript enabled to view it.

Abstract

Population-based metaheuristic (P-metaheuristic) algorithms are trendy for solving optimization problems. However, according to the NFL theorem, no metaheuristic is suited for solving all kinds of optimization problems. This study aims to determine the proper algorithms for the different modals of unconstrained optimization problems. We conducted a post-analysis of one-way analysis of variance (ANOVA) to test 14 P-metaheuristics, including evolutionary algorithms and swarm intelligence algorithms, on 23 unconstrained optimization benchmark functions. Experimental results show that the Harris hawks optimization (HHO) algorithm and gray wolf optimizer (GWO) are robust and more suitable for unimodal functions. In addition to the HHO being the best, the whale optimization algorithm (WOA) and GWO are also good choices for multi-modal functions. The cuckoo search (CS) algorithm dominated over fixed-dimension multi-modal functions. The study found that HHO, GWO, and WOA have similar mechanisms, such as searching (exploration), encircling, and attacking (exploitation) prey. The HHO and CS adopt the Lévy-flight-style random walk strategy to enhance the exploitation and exploration capabilities. Consequently, we acquired the proper P-metaheuristic to solve different modals and found superior mechanism to develop better P-metaheuristics in the future.

Key words: metaheuristic algorithms, population-based, swarm intelligent, exploration and exploitation capabilities, post-analysis, one-way ANOVA.

Attachments:

1241 Final.pdf

[ ]

633 kB