一种基于动态邻居和变异因子的粒子群算法  

Particle swarm optimizer based on dynamic neighborhood topology and mutation operator

作　　者：刘衍民[1,2] 赵庆祯[1] 隋常玲[2] 邵增珍[1] 

机构地区：[1]山东师范大学管理与经济学院,济南250014 [2]遵义师范学院数学系,贵州遵义563002

出　　处：《控制与决策》2010年第7期968-974,共7页Control and Decision

基　　金：山东省科技攻关项目(2009GG10001008);贵州省教育厅社科项目(0705204);遵义科技攻关项目([2008]21号)

摘　　要：提出一种基于动态邻居和变异因子的粒子群算法(DNMPSO).在该算法中,粒子的邻居是根据它的运行而动态变化.每个粒子的学习机制分为自己的历史经验和所有邻居的经验两部分.为了保证有效求解多峰问题,在每一次迭代,对当前解采用水平混合变异,使每个粒子能更好地进行局部搜索,提升粒子跳出局部最优解的能力.通过与其他算法比较,结果表明该算法求解多峰问题的能力最优.A variant of particle swarm optimizer（PSO） based on dynamic neighborhood topology and mutation factor（DNMPSO） is proposed in this paper. In DNMPSO,the neighborhood of a particle are not fixed but dynamically changed,and the learning mechanism of a particle includes two parts,the historical best experience of the particle itself,and the experiences of its all neighbors. To effectively solve multimodal problems,the parallel hybrid mutation is used to work for local search,which improves the ability of escaping from local optima. The results demonstrate good performance of the DNMPSO algorithm in solving complicated multimodal problems when compared with other PSO algorithms.

关 键 词：动态邻居 变异 粒子群 算法 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]