摘要
针对基本微粒群算法在处理高维度目标函数容易出现早熟的问题,提出了一种新的微粒群算法面向高维度目标函数的微粒群算法(HDOF-PSO).分析了基本微粒群算法难以处理高维度目标函数的原因.通过引入信心度和试探策略,算法的收敛速度得到提高;通过引入成功度,搜索过程中的变异概率能够自适应修正.在特定测试函数集上的实验表明,HDOF-PSO在处理高维目标函数时,比基本微粒群算法和一个改进的微粒群算法具有更快的收敛速度和更好的收敛性.
A higher-dimensional-object-function particle swarm optimizer HDOF-PSO algorithm is proposed for the prematurities which are easy to take place when dealing with the higher- dimensional object function by BPSO （basical particle swarm optimizer） algorithm. The reason why HDOF-PSO is difficult to be deal with hasical PSO algorithm is analyzal. The confidence level and trial-and-error strategy are introduced into the algorithm to accelerate its convergence rate with the probability of success also introduced in to enable the adaptive correction available to the probability of mutation in searching process. The experimental results of a specific set of benchmerk functions showed that the HDOF-PSO algorithm has better convergence and higher dimensional object functions.
出处
《东北大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2008年第5期649-652,共4页
Journal of Northeastern University(Natural Science)
基金
国家火炬计划项目(2002EB010154)
关键词
群体智能
微粒群算法
高维度
自适应试探
自适应变异
swarm intelligence
PSO algorithm
higher dimension
adaptive trial-and-error
adaptive-mutation
