摘要
为了更深入地分析探讨粒子群优化算法的性能,采用两种基本改进策略在MATLAB7.0中对几个典型测试函数的优化问题进行了实验,即单独采用线性递减惯性权重策略以及在其基础上再加入收缩因子法,给出了这两种策略下函数的在线性能、离线性能变化图。为指导参数选取,用图示方式给出了不同参数组合对收敛性的影响。结论是:采用线性递减惯性权重策略加上收缩因子法比单独采用线性递减惯性权重策略的收敛性能好。若取固定惯性权重w,则w越小,收敛速度越快。
To analyse the performance of particle swarm optimization method deeply,this paper uses two basic improved strategies to experiment several standard test functions optimization problem in the MATLAB 7.0 software,one of the strategies is linear inertia weight reduction only,the other is rejoining the constriction factor.The online and off-line performances are given to the two strategies.In order to guide the parameter selecting,we present the effect of convergence on different parameter combination through many charts.The conclusion is that the convergence of the strategy of inertia weight reduction adding the constriction factor is better than that of the strategies of linear inertia weight reduction only.And if we adopt fixed inertia weight w,the convergence rate is more rapid when the w is lesser.
出处
《计算机工程与应用》
CSCD
北大核心
2008年第28期53-54,90,共3页
Computer Engineering and Applications
基金
教育部科学技术研究重点项目(No.107106)
教育部高等学校科技创新工程重大项目培育基金。
关键词
粒子群优化
惯性权重
收缩因子
收敛性
Particle Swarm Optimization（PSO）
inertia weight
constriction factor
convergence
