一种基于动态拓扑结构的PSO改进算法  

Improved Particle Swarm Optimizer Based on Dynamic Topology

作　　者：温雯[1] 郝志峰[1] 

机构地区：[1]华南理工大学数学科学学院,广州510641

出　　处：《计算机工程与应用》2005年第34期82-85,共4页Computer Engineering and Applications

基　　金：国家自然科学基金(编号:10471045);广东省自然科学基金(编号:031360;04020079);霍英东教育基金会高等院校青年教师基金(编号:91005)

摘　　要：该文提出了一种新颖的PSO改进算法-PSO-DT。该算法通过动态调整粒子群的拓扑结构,在算法前期弱化全局最优粒子的影响力,以最大化地扩展寻优范围;在算法后期则强化全局最优粒子的影响力,以加快算法收敛速度。此外,文章还在PSO-DT中引入变异算子,获得MPSO-DT,大大减少了算法时间。通过对6个基准函数的测试及与另一改进算法MPSO-TVAC的对比实验,证实了该改进方案是有效而实用的。A novel version of PSO based on Dynamic Topology （PSO-DT）is proposed in this paper.In PSO-DT,the influence of Global-best particle is adjusted according to the number of iterations：in the initial stage,the influence of Global-best is weakened so as to expanding the particles＇ searching range,while in the late stage it is strengthened so as to guarantee the particles converge.Besides,mutation is introduced into PSO-DT,which makes MPSO-DT,so as to shorten searching time.By testing six well-known benchmark functions,it has been proved that both PSO-DT and MPSO-DT are effective.

关 键 词：粒子群优化算法 拓扑结构 模拟退火 变异算子 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]