一种改进的动量粒子群算法及实验分析  

AN IMPROVED PARTICLE SWARM OPTIMIZATION ALGORITHM BASED ON MOMENTUM TERM AND ITS EXPERIMENTAL ANALYSIS

作　　者：黄福员[1,2] 

机构地区：[1]华南理工大学经济与贸易学院,广东广州510640 [2]湛江师范学院商学院,广东湛江524048

出　　处：《计算机应用与软件》2009年第10期57-59,共3页Computer Applications and Software

基　　金：高等学校博士学科点专项科研基金(20060561002)

摘　　要：为了克服粒子群算法存在的收敛缓慢、后期振荡等缺陷,在基本粒子群算法的基础上,引入动量项,提出一种新的改进型粒子群算法。新算法中动量项与微粒的历史修正量线性相关,典型复杂优化函数的实验结果表明:该算法不但保持了基本粒子群算法的简单、易实现等优点,而且提高了算法的收敛速度及部分地避免了算法的后期振荡。Particle Swarm Optimization （PSO） algorithm suffers from slow convergence and oscillations in later iterations. To overcome those drawbacks, an improved particle swarm optimization called IPSO was presented in this paper,in which a momentum term is embedded in- to fundamental PSO algorithm. In the proposed algorithm the momentum term is linear correlated with the historical correction value of particle. The experiment results of typical complex optimized function demonstrate that the algorithm keeps the characters of the fundamental PSO algorithm in simple and easy to be implemented, and also enhances convergence speed and partially gets rid of late-time oscillations in the algorithm.

关 键 词：群智能 粒子群算法 动量项 

分 类 号：O212.1[理学—概率论与数理统计] TS210.3[理学—数学]