群知能と進化的計算の国際ジャーナル

概要

An Improved Multitracker Optimization Algorithm and Multiple Subpopulations

Rizk M. Rizk-Allah, Fatma Helmy Ismail, Aboul Ella Hassanien

Recently, a population-based optimization algorithm called the multitracker optimization algorithm (MTOA) was introduced based on the tracker con- cept. This paper proposes a novel variation of the original MTOA called the migration-based MTOA (MTOA1), which employs multiple subpopulations of trackers to achieve superior performance. The proposed algorithm differs from the traditional MTOA in that it splits the initial population into multiple sub- populations to enhance the search process in different areas of the search space. Furthermore, information is exchanged among the subpopulations in an iter- ative and cyclic manner. The best global trackers in the first subpopulation are used to update the global trackers of the second subpopulation, and this updating process continues for all subsequent subpopulations. Exploration and exploitation are balanced in this cyclic approach for multiple populations. The proposed MTOA1 is validated based on the CEC2017 benchmark problems, and an improvement over the original MTOA is observed. Furthermore, MTOA1 is used to solve the classical welded beam design problem and is compared with eight recently proposed optimization algorithms. The results confirm the superiority of the proposed algorithm.