一种动态环境下带有记忆的三岛粒子群算法  

Tri-island particle swarm optimization with memory in dynamic environment

作　　者：王洪峰[1] 汪定伟[1] 

机构地区：[1]东北大学信息科学与工程学院系统工程研究所,辽宁沈阳110004

出　　处：《系统工程学报》2008年第2期252-256,共5页Journal of Systems Engineering

基　　金：国家自然科学基金资助项目(70431003;70671020);国家创新研究群体科学基金资助项目(60521003);国家支撑计划资助项目(2006BAH02A09)

摘　　要：近些年来,求解动态环境下的优化问题已经引起越来越多的关注.借鉴并行进化计算中的孤岛模型思想,提出了一种特殊的粒子群算法,将整个粒子群分成三个不同的岛屿(子粒子群),分别具有探索、开发和记忆的功能,子粒子群间每经过一段时间进行一次交流.通过对一类标准的动态测试函数——移动峰问题的实验,能够证明该算法有效地解决某些动态优化问题.Recently, there has been increased interest in optimijation problem applied to dynamic environment. In this paper a new variation of particle swarm optimization （PSO） is presented for problem optimization in dynamic environments. Inspired by island model in parallel evolutionary computation, a swarm is divided into three islands which are used for exploration, exploitation and memory respectively. Each island works independently and communicates with others after a period of time. Experimental study over a benchmark dynamic problem showts that the three-island PSO with memory can solve some dynamic optimization problems efficiently.

关 键 词：粒子群算法 记忆 动态优化问题 岛屿模型 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]