基于文化的多目标协同进化算法  

Multi-objective co-evolutionary algorithm based on culture

作　　者：赵学臣[1,2] 王洪国[1] 邵增珍[1,2] 苗金凤[1,2] 

机构地区：[1]山东师范大学信息科学与工程学院,济南250014 [2]山东省分布式计算机软件新技术重点实验室,济南250014

出　　处：《计算机应用研究》2011年第7期2494-2496,共3页Application Research of Computers

基　　金：山东省科技攻关资助项目(2009GG10001008);济南市高校院所自主创新资助项目(200906001)

摘　　要：为了改进NSGAⅡ算法中存在的分布性等不理想问题,在NSGAⅡ的基础上提出了基于文化的多目标协同进化算法。该算法提出评测信念空间多样性的指标,从信念空间中提取知识,利用知识来指导种群的进化;提出知识假说集,以现有知识为基础产生新知识,加强局部搜索,加速算法收敛。仿真实验表明该算法较NSGAⅡ在收敛性及分布性方面均有明显提高。In order to overcome the deficiency of NSGA Ⅱ on distribution,this paper proposed a multi-objective co-evolutionary algorithm based on culture.The algorithm proposed an index which was used for evaluating the diversity of belief space,extracted knowledge from belief space,and utilized the knowledge to guide the evolution of population.It proposed a concept called knowledge hypothesis set,created new knowledges based on knowledges in being to enhance local search in sparse area,which speeded up the convergence of the algorithm.Simulation results indicate that the algorithm improves significantly in terms of the ability of convergence and distribution.

关 键 词：多目标 文化 协同进化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP301.6[自动化与计算机技术—控制科学与工程]