Crowding-measure-based multi-objective particle swarm optimization(CMPSO) is proposed,in which crowding measure is used to maintain external archive and select global best position for each particle from archive and some archive members are chosen to perform mutation operator.
设计出基于密集距离的多目标粒子群优化算法(CMPSO),该算法根据密集距离大小按轮盘赌方式为每个粒子从外部档案选取全局最好位置并采用基于密集距离的方法对外部档案进行维护。
Secondly, an external population maintenance method in terms of the crowding measure of individual is proposed, and then all individuals are categorized into four kinds according to Pareto dominance relationship and the crowding-measure based fitness function is defined.
外部种群维护和适应度赋值是多目标进化算法(MOEA)的两个重要部分,该文首先对这两个问题目前已有的处理方法进行了分析,然后提出了基于个体密集距离的外部种群维护方法,并在将所有个体根据Pareto支配关系分成四个层次的基础上,给出了一种由个体密集距离定义的适应度函数,最后将基于个体密集距离的多目标进化算法CMOEA应用于几个常用的测试函数,并和SPEA,SPEA-2进行了比较,计算结果表明CMOEA具有良好的搜索性能。