Authors: Guohan Lin; Jing Zhang
Addresses: College of Electrical and Information Engineering, Hunan University, Hunan Changsha 410082, China; College of Electrical and Information, Hunan Institute of Engineering, Hunan Xiangtan 411101, China ' College of Electrical and Information Engineering, Hunan University, Hunan Changsha 410082, China
Abstract: The basic particle swarm optimisation (PSO) algorithm is easily trapped in local optima. To deal with this problem, a multi-subpopulation cooperative particle swarm optimisation (MCPSO) is presented. In the proposed algorithm, the particles are divided into several normal subpopulations and an elite subpopulation. The selected individuals in normal subpopulation are memorised into the elite subpopulation, and some individuals in normal subpopulation are replaced by the best particles from the elite subpopulation. Different subpopulation adopts different evolution model. This strategy can maintain the diversity of the population and avoid the premature convergence. The performance of the proposed algorithm is evaluated by testing on standard benchmark functions. The experimental results show that the proposed algorithm has better convergent rate and high solution accuracy.
Keywords: particle swarm optimisation; multi-subpopulation PSO; cooperative PSO; premature convergence; elite particle; mutation operator; diversity.
International Journal of Computing Science and Mathematics, 2015 Vol.6 No.1, pp.30 - 39
Received: 19 Jul 2014
Accepted: 24 Aug 2014
Published online: 19 Feb 2015 *