摘要
简要介绍了粒子群算法(PSO),对算法中的重要参数惯性权重进行了系统的实验,分析了固定权重与时变权重的选择问题,并从问题依赖性、种群大小和拓扑结构等方面详细分析了惯性权重对于算法性能的影响.结果表明,惯性权重的问题依赖性较小,随着种群的增大,其取值应适当减小,局部版本下,惯性权重的选择具有更大的自由度.
The basic principle of particle swarm optimization (PSO) is introduced briefly in the paper. Sufficient experiments are done on inertia weight that is an important parameter in the algorithm. The selection difference between fixed and time-variant inertia weight is analyzed. And the influence of inertia weight on the algorithm performance is analyzed in detail from different aspects, including the problem dependence, swarm size and topologic structure. The result indicates the inertia weight has little problem dependence, and with the increase of the population its value should decrease properly. The selection of inertia weight has more freedom in the local version of PSO.
出处
《系统工程学报》
CSCD
北大核心
2005年第2期194-198,共5页
Journal of Systems Engineering
基金
国家自然科学基金重点资助项目(70431003).
关键词
粒子群算法
进化计算
惯性权重
particle swarm optimization
evolutionary computation
inertia weight