Group Particle Swarm Optimization with Communication Strategy and Group Diversity

• Main Authors: Chia-hao Shen, 沈家豪
• Other Authors: Cheng-lung Huang
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/02654476881105870864

碩士 === 國立高雄第一科技大學 === 資訊管理所 === 95 === This study presents a group particle swarm optimization algorithm with three kinds of communication, and this method is extending from particle swarm optimization with subpopulation concept. We introduce particle swarm divided into four subpopulation called group pso, then each group of particles uses the standard particle swarm optimization to search the global best in the search place. Each particle group has searched local optimization. In order to improve each particle swarm to have communication channel, so we design three kinds of communication strategies between particle groups. The first communication strategy is to exchange the local best optimization of each particle groups; The second communication strategy is to exchange the particle members of each particle groups; The third communication strategy is to sort particles base on the fitness of each particles, then distribute all particle into four groups. Our study furthers to discuss each particle group with different activity model, in order to increase the search population of particle’s diversity. Each particle groups have different personal and social learning capability. To validate the proposed methods, numerical illustrations were conducted to compare the proposed PSO with traditional PSO for solving continuous problem. In this paper we propose particle swarm optimization with communication. From the results, we find that the Group particle swarm communicated to each other by changing their particle members (in this paper we called communication strategy 2) is better strategy.