An Optimal-Parent-Crossover based Differential Evolution
Ruihan Zhao, Sichen Tao, Pengzhong Li · 2022
Differential Evolutionary (DE) is an important part of evolutionary algorithms (EAs) and it is efficient in solving complex optimization problems and has a straight- forward algorithmic structure. In order to further exploit the advantages of DE, we propose a new variant of DE called optimal-parent-crossover based differential evolution (OPCDE). Optimal-parent individuals participate in the crossover in a simple way, effectively increasing the convergence speed of the original algorithm without additional parameters. On the basis of the 30 optimization problems of the IEEE Congress on Evolutionary Computation (CEC) 2017 bench- mark, experimental results show that OPCDE significantly outperforms the previous algorithm and is competitive with one of the state-of-the-art DE variants, as well as outperforming two other state-of-the-art algorithms of two important branches of the meta-heuristics, GSA and PSO. At the same time, OPCDE is able to converge quickly when solving unimodal problems and always maintains a high degree of diversity when solving simple multimodal, hybrid and composition optimization problems.