粒子群优化算法的分析与改进  

Analysis and Improvement of Particle Swarm Optimization Algorithm

作　　者：张丽平[1] 俞欢军[1] 陈德钊[1] 胡上序[1] 

机构地区：[1]浙江大学智能信息工程研究所,浙江杭州310027

出　　处：《信息与控制》2004年第5期513-517,共5页Information and Control

基　　金：国家自然科学基金资助项目 ( 2 0 0 760 41)

摘　　要：分析了惯性权值对粒子群优化 (PSO)算法优化性能的影响 ,进而提出选择惯性权值的新策略 .在随机选取惯性权值的同时 ,自适应地调整随机惯性权值的数学期望 ,有效地调整算法的全局与局部搜索能力 .测试表明基于随机惯性权 (RIW )策略的PSO算法 ,其全局搜优的速率与精度有明显提高 .The effects of inertia weight on particle swarm optimization (PSO) performance are analyzed. A novel method of selecting inertia weight in PSO is developed, which can tune the expectations of inertia weights adaptively when the inertia weights are randomly selected and lead to effectively balance between the local and global search ability. Results of the two benchmark functions indicate that the PSO algorithm based on the strategy of random inertia weight (RIW) has been significantly improved on both optimization speed and computational accuracy.

关 键 词：粒子群 优化算法 随机 惯性权 策略 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]