A Novel Spherical Search Based Grey Wolf Optimizer for Optimization Problems

Abstract

Grey wolf optimizer (GWO) has shown to converge rapidly during the initial stage of a global search, but it still frequently stick into local optimal. In contrast, spherical evolution (SE) adopts a brand new spherical search style and has good abilities of local optimum avoidance. The focus of this research is on incorporating SE into GWO for optimization problems. This hybrid method generates a new generation of individuals by alternating the leadership hierarchy and hunting mechanism of GWO and the spherical search style of SE. The experiment results on CEC2017 benchmark functions indicate the effectiveness of this hybridization, suggesting that grey wolf search mechanism and spherical search style are complementary. This study gives not only more insights into both original algorithms, but also a novel construction method of merging different algorithms.