摘要
种群维护是多目标进化算法的重要组成部分。针对维护方法和运行效率的矛盾,提出一种基于邻域的多目标进化算法(NMOEA)。定义了一个反映个体之间邻近程度的指标———邻域包含关系,利用此关系对个体进行分布适应度分级的赋值,并用动态方法快速地对种群进行维护。通过7个测试问题和3个方面的测试标准,结果表明新算法在较快速地接近真实的最优面的同时,拥有良好的分布性。
Population maintenance is an important issue in multi-objective evolutionary algorithms. For the deficiency that the maintenance methods of good distribution usually have a high time complexity, a multi-objective evolutionary algorithm based on neighborhood （named NMOEA） was proposed. This measure defined a criterion-neighborhood containing relation, which represented the close degree of individuals. And it was used to assign diversity fitness in a dynamic method that maintained the population rapidly. By examining three performance metrics on seven test problems, the new algorithm can approach the true Pareto front fast, and has good distribution.
出处
《计算机应用》
CSCD
北大核心
2008年第6期1570-1574,共5页
journal of Computer Applications
基金
国家自然科学基金资助项目(60773047)
国家863计划项目(2001AA114060)
留学回国人员科研启动基金资助项目(教外司留[2005]546号)
湖南省自科基金资助项目(05JJ30125)
湖南省教育厅重点科研资助项目(06A074)
关键词
多目标进化算法
多目标优化问题
种群维护
分布适应度
邻域
multi-objective evolutionary algorithm
multi-objective optimization problem
population maintenance
diversity rank
neighborhood
