基于存档策略的多目标优化的遗传算法及其收敛性分析  

Archival-based Genetic Algorithm for Multi-objective Optimization and its Convergence Analysis

作　　者：张雅波[1] 刘庆怀[2] 罗长童[3] 

机构地区：[1]天津工程师范学院数理与信息科学系,天津300222 [2]长春工业大学应用数学研究所,长春130012 [3]吉林建筑工程学院基础部,长春130012

出　　处：《数学的实践与认识》2009年第2期37-42,共6页Mathematics in Practice and Theory

摘　　要：设计了一种用遗传算法求解多目标优化问题的有效方法——基于存档策略的多目标优化的遗传算法,并讨论了此算法的收敛性.首先给出档案的定义,设计出基于支配关系下的带有存档策略遗传算法,并通过算例检验了算法的有效性;然后引入了两档案间的距离的概念,在此距离定义的基础上证明了算法在概率意义下是收敛的.A strategy for solving multi-objective optimization problems by genetic algorithm is proposed--Archival-based genetic algorithm for multi-objective numerical optimization problems, and its convergence is discussed. First, An archive is defined, Archival-based genetic algorithm in the concept of Pareto dominance is presented. Its effectiveness is illustrated by an example. Second, the distance between two archives is defined, in the base of this distance, we demonstrate that the algorithm is probability convergence.

关 键 词：多目标优化 遗传算法 收敛性 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]