基于多种群多模型协同进化的粒子群优化算法  

Particle Swarm Optimization Algorithm Based on Multi-swarm and Multi-model Cooperative Evolution

作　　者：徐冰纯[1] 葛洪伟[1] 王燕燕[1] 

机构地区：[1]江南大学物联网工程学院,江苏无锡214122

出　　处：《计算机工程》2013年第5期200-203,208,共5页Computer Engineering

摘　　要：为克服标准粒子群优化(PSO)算法易陷入局部极值和优化精度较低的缺点,提出一种多种群多模型协同进化的粒子群优化(MSM-PSO)算法。将整个粒子群分成大小相等的3个分群,各分群采用不同的进化模型,分群间相互影响促进。同时采用自适应动态惯性权重,以保持种群多样性,降低陷入局部极值的概率。测试结果表明,该算法全局性能好、寻优精度高。Aiming to improve the performance of standard Particle Swarm Optimization（PSO） algorithm, an improved PSO algorithm based on Multi-swarm and Multi-model cooperative evolution（MSM-PSO） is proposed. The particles are divided into three swarms in the same size, different swarms adapt different evolution models, and three swarms interact and affect each other. Besides, adaptive dynamic inertia weight is adopted in the algorithm, which can maintain diversity of population and reduce the possibility of local minimum. Simulation result shows that this method has better global performance, and higher optimization precision.

关 键 词：粒子群优化算法 多种群 多模型 自适应动态惯性权重 协同进化 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]