Affiliations: [a] School of Information Science and Technology, Jiujiang University, Jiujiang, China
| [b] School of Computer, Wuhan University, Wuhan, China
Correspondence:
[*]
Corresponding author. Hu Peng, School of Information Science and Technology, Jiujiang University, Jiujiang 332005, China. E-mail: [email protected].
Abstract: As a new and promising swarm intelligence algorithm, brain storm optimization (BSO) has drawn more attention of researches and has been successfully applied to solve the real-world optimization problems. However, too many parameters make the algorithm more complex and greatly limit the convergence performance. Thus, this paper proposed a novel BSO variant, named self-adaptive BSO with pbest guided step-size (SPBSO), in which a simple self-adaptive strategy is employed to choose a creating strategy in a random manner rather than depending on several adjustable parameters. In addition, the pbest guided step-size and dynamic clustering number are used to accelerate the convergence speed. The experimental studies have been tested on a set of widely used benchmark functions (including the CEC 2014 problems). Experimental results and comparison with the state-of-the-art BSO variants and some recently proposed PSO and DE algorithms, have proved that the proposed algorithm is competitive.
