摘要
针对基本粒子群算法具有容易陷入局部极值、对多维搜索空间精度不高等缺陷,提出了一种位置扰动的粒子群算法。算法通过对粒子个体最优位置的一个或多个随机维上的计算,产生对群体最优位置对应维上的扰动,使群体最优位置可以从个体最优位置搜索经验中更直接的学习,并且跳出局部最优。通过几个常用测试函数的测试结果表明,位置扰动的粒子群算法比标准PSO算法在处理多峰值、多维搜索空间问题时有更高的寻优能力。
To solve the problems such as the basic particle swarm algorithm＇s easy to fall into local extremum and its low accura- cy of multi-dimensional search, a new algorithm named position disturbed particle swarm optimization （PDPSO） is provided. Al- gorithm uses one or more random dimensions of the best positions of individual to produce disturbance for corresponding dimen5 sions of the best position of the group, the best position of the group learns from the best positions of individual more directly, and the algorithm jumps out of the local optimal. Finally, the results of some commonly used test functions presented shows that, PDPSO algorithm has a better global optimization capability than standard PSO algorithm in situations such as multi-peak and multi-dimensional searching problems.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第3期1037-1040,共4页
Computer Engineering and Design
基金
山西省自然科学基金项目(2013011017-7)
关键词
粒子群算法
个体最优位置
群体最优位置
扰动
随机维
particle swarm optimization
the best position of individual
the best position of group
disturbance
randomdimension
