一种基于向量空间模型的NSGA-II改进算法  

An Improved NSGA-II Algorithm Based on Vector Space Model

作　　者：黄超[1] 胡德敏[1] 余星[1] 

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093

出　　处：《小型微型计算机系统》2015年第2期391-396,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61202376)资助;上海市教委科研创新项目(13YZ075)资助

摘　　要：带拥挤距离排挤机制的非支配排序遗传算法(NSGA-II)在多目标优化领域具有广泛的应用,NSGA-II算法具有个体分布不均匀以及重复个体较多等缺陷.针对这些缺陷提出一种基于向量空间模型的NSGA-II改进算法VSMGA(Vector Space M odel Genetic Algorithm),VSM GA算法在NSGA-II算法的基础上引入了向量空间模型,利用目标权重向量之间的余弦距离代替原来的拥挤距离,提出一种距离排挤机制和重复个体排除规则.实验结果表明与NSGA-II算法比较,VSMGA算法具有更好的分布性和稳定性.Non-dominated Sorting Genetic Algorithm with crowded distance mechanism（NSGA-II） is frequently used in multi-objec- tive optimization. However, Poor distribution of population and negative impact of the overlapping individuals are main defects of the algorithm. In this paper, an improved algorithm named VSMGA （ Vector Space Model Genetic Algorithm ） is proposed against the de- fects. A crowding mechanism base on cosine distance between the target weight vectors and an eliminating overlapping solution is put forward in the algorithm. The experiments show that the improved algorithm has better distribution and stability.

关 键 词：多目标进化 NSGA-II 空间向量模型 分布性 重复个体 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]