NSGA-Ⅱ中重复个体产生原因分析及影响研究  

Research on cause for overlapping solutions and on their influence in NSGA-Ⅱ algorithm

作　　者：谢炯亮[1] 郑金华[1] 

机构地区：[1]湘潭大学信息工程学院,湖南湘潭411105

出　　处：《计算机工程与应用》2008年第29期69-72,145,共5页Computer Engineering and Applications

基　　金：国家自然科学基金No.60773047; 国家高技术研究发展计划( 863)No.2001AA114060;教育部留学回国人员科研启动基金( 教外司留[2005]546 号);湖南省自然科学基金No.05JJ30125;湖南省教育厅重点科研项目( No.06A074)~~

摘　　要：进化种群中出现重复个体意味着搜索区域的重叠,使得算法探索新可行区域的效率降低。另外个体重复浪费了解集中的个体名额,且造成信息冗余,使得解集的有效代表性变差。这在用NSGA-Ⅱ处理低维问题时体现得较为严重。分析了NSGA-Ⅱ中出现重复个体的原因,测试了编码方式和变量维数与重复个体数量的关系;通过实验检验了重复个体对于算法性能和解集质量的影响。实验结果表明,去除重复个体的算法能获得分布性更好的解集,且具有更强的稳定性。The existence of overlapping individuals in the evolution populations means overlapping regions in the searching space,which makes the algorithm much less efficiently in exploiting new feasible region.Additionally,it wastes the positions in the population and leads to the information redundancy,which reduces the diversity of the obtained solution set.This phenomenon is quite obvious in the famous NSGA-Ⅱ algorithm when applied to low-variable dimension problems.In this paper,the cause of the overlapping solutions in NSGA-Ⅱ is analyzed and the relation between the number of them and the coding method is dis- covered,as well as and the variable dimension;then overlapping solutions have influence on the performance of the NSGA-Ⅱ algorithm and on the quality of the obtained solution set is illuminated.The experimental results demonstrate that eliminating overlapping solutions make the NSGA-Ⅱ algorithm more steady and gain a solution set with better diversity

关 键 词：NSGA—Ⅱ 重复个体 编码方式 进化操作 拥挤距离 变量维数 分布度 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]