摘要
针对动态多目标优化问题,提出一种基于Pareto解集关联与预测的动态多目标进化算法(LP-DMOEA),设计了基于超块的Pareto解集关联方法.该方法能够动态维护若干描述Pareto解变化规律的时间序列,通过对新环境下的Pareto解集进行预测来生成初始种群.将LP-DMOEA应用于非劣分类遗传算法(NSGA2),并对3类标准测试函数进行了实验,所得结果表明该方法能够有效求解动态优化问题.
In order to solve dynamic multi-objective optimization problem（DMOPs）,a dynamic multi-objective evolutionary algorithm based on Pareto set linkage and prediction（LP-DMOEA） is proposed and a Pareto set linking method based on hyperbox is designed.In this scheme,several time sequences which present the trend of Pareto solutions can be dynamically maintained.Based on the prediction of these time sequences,the initial population is generated.The LP-DMOEA is applied to the NSGA2 algorithm to solve three benchmark problems.Computational results show the effectiveness of the LP-DMOEA to solve DMOPs.
出处
《控制与决策》
EI
CSCD
北大核心
2011年第4期615-618,共4页
Control and Decision
基金
国家自然科学基金项目(60875071
60774064)
水下信息处理与控制国家级重点实验室基金项目(9140C230503090C23)
关键词
动态多目标优化问题
动态多目标进化算法
Pareto解集关联与预测
超块
dynamic multi-objective optimal problem
dynamic multi-objective evolutionary algorithm
Pareto set linkage and prediction
hyperbox
