基于精英选择和个体迁移的多目标遗传算法  

Multiobjective genetic algorithm based on elitist selection and individual migration

作　　者：祁荣宾[1] 钱锋[1] 杜文莉[1] 颜学峰[1] 

机构地区：[1]华东理工大学自动化研究所,上海200237

出　　处：《控制与决策》2007年第2期164-168,共5页Control and Decision

基　　金：国家973计划项目(2002CB3122000);国家863计划项目(2003AA412010;AA413130);2006上海市博士后科研基金项目

摘　　要：提出基于遗传算法求解多目标优化问题的方法,将多目标问题分解成多个单目标优化问题,用遗传算法分别在每个单目标种群中并行搜索.在进化过程中的每一代,采用精英选择和个体迁移策略加快多个目标的并行搜索,提出了控制Pareto最优解数量并保持个体多样性的有限精度法,同时还提出了多目标遗传算法的终止条件.数值实验说明所提出的算法能较快地找到一组分布广泛且均匀的Pareto最优解.A new method is proposed to solve multiobjective optimization based on genetic algorithm.The method is implemented to optimize each one of the multiobjective problem separately by parallel searching of genetic algorithm.At each generation of the evolving process,an elitist selection and individual migration to improve the parallel searching speed is applied.Finite precision method is applied to control the amount of the Pareto optimal solutions and keep the individual diversity.At the same time,a new stopping condition of multiobjective genetic algorithm is proposed.The numerical results demonstrate that the proposed method can fleetly find the Pareto optimal solutions which formerly are scattered extensively and uniformly.

关 键 词：多目标优化 遗传算法 PARETO最优解 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]