基于Powell搜索法的简化微粒群算法  

A Simplified Particle Swarm Algorithm Based on Powell Search Method

作　　者：夏桂梅[1] 苏长慧 

机构地区：[1]太原科技大学应用科学学院,山西太原030024

出　　处：《宁夏大学学报（自然科学版）》2015年第2期126-130,共5页Journal of Ningxia University(Natural Science Edition)

基　　金：山西省自然科学基金资助项目(2014011066-2)

摘　　要：利用Powell搜索法求解精度高、收敛速度快和局部强搜索能力强等优点,在简化微粒群算法的基础上,结合Powell搜索法,提出一种新型简化微粒群算法—Powell-SPSO算法.改进算法将Powell搜索法融合在简化微粒群算法中,让Powell搜索法与简化微粒群算法进行交替搜索.同时微粒的迭代利用了Powell搜索法的强搜索能力,使得算法改善了简化微粒群算法因每个微粒采用相同迭代公式进行进化而造成的微粒间的弱差异性,避免了易出现早熟、搜索速度慢等缺点.仿真结果表明,与标准微粒群算法(PSO)、简化微粒群算法(SPSO)、文献[10]算法相比较,不论是对高维函数还是低维函数,改进的算法都能够有效地避免早熟问题,并能显著地提高收敛速度和收敛精度.A new simplified particle swarm optimization--Powell-SPSO is put forward by using the advantages of Powell search method and simplifies particle swarm optimization. Improved algorithm let Powell search method and particle swarm optimization to search alternately. Meanwhile the new algorithm improves each particle of simplified particle swarm of algorithm premature, slowing search speed and the difference between the particles of weak by taking advantage of the strong search capabilities of Powell search method. The simulation results show that, compared with the PSO, SPSO and algorithms of literature 10, whether high or low dimensional function, the improved algorithm can effectively avoid premature convergence problem, and can significantly improve the convergence speed and accuracy.

关 键 词：微粒群算法 简化微粒群算法 POWELL搜索法 全局优化 

分 类 号：O221[理学—运筹学与控制论]