Alternate search pattern-based brain storm optimization

Abstract

Brain storm optimization (BSO) groups population into several clusters and generates new individuals by using the information of these clusters. However, this mechanism limits the ability of exploration because it prevents new individuals from searching regions far away from current clusters. In this paper, we innovatively propose a grid-based search operator (GBS) to improve the exploration by dividing the given search space into smaller ones. Then, we modify the cluster, replacement, and mutation strategy of the original BSO for requiring a better exploitation. Besides, an alternate search pattern (ASP) strategy is designed for controlling the transformation between GBS and BSO to balance exploration and exploitation. Finally, two variants of BSO have been proposed based on the original BSO and a global-best BSO, and termed as ABSO and AGBSO, respectively. The proposed ABSO and AGBSO are tested on a number of widely used benchmark optimization problems. The comparative analysis shows that ASP strategy can significantly improve the performance of BSO in terms of solution quality and population diversity. Additionally, AGBSO can be considered as a state-of-the-art BSO among all its variants. The source code of all proposed methods can be found at https://toyamaailab.github.io/sourcedata.html.