动态选择与替换策略的多目标约束优化进化算法  

Multi-objective Constrained Optimizations Evolutionary Algorithm Based on Dynamic Selection and Replacement Strategy

作　　者：龙文[1,2] 梁昔明[1] 秦浩宇[1] 

机构地区：[1]中南大学信息科学与工程学院,长沙410083 [2]贵州财经学院贵州省经济系统仿真重点实验室,贵阳550004

出　　处：《小型微型计算机系统》2011年第9期1862-1866,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(60874070)资助;高等学校博士学科点专项科研基金项目(20070533131)资助;湖南省研究生科研创新项目(CX2009B038)资助

摘　　要：提出一种基于动态选择与替换策略的多目标优化进化算法用于求解约束优化问题.新算法首先将约束优化问题转化为两个目标的多目标优化问题,基于Parto支配关系,把初始种群分为Pareto子集和Non-Pareto子集,引入一种非劣个体保护偏好策略,动态选取一定比例的最优非劣个体直接进入下一代群体,剩下的非劣个体随机替代Pareto子集中的个体.Pareto子集和Non-Pareto子集分别进行单形交叉和多样性变异操作产生新的子种群.对13个标准测试问题的数值实验结果表明新算法的有效性.A muiti-objective constrained optimization evolutionary algorithm based on a dynamic selection and replacement strategy is proposed for solving constrained optimization problems. The constrained optimization problem is converted into a multi-objective optimization problem. During the evolutionary process, our algorithm is based on multi-objective optimization technique, where the initial population is divided into two sets, one is called by Pareto set, and the other is non-Pareto set. A non-dominated individual＇s conservation bias strategy is used to keep a certain number of infeasible solutions in each generation. The randomly selected individuals in Pareto sat are replaced by the resting non-dominated individuals. It simultaneously uses simplex crossover and diversity mutation operator to generate the offspring population. The proposed algorithm is test on 13 well-known constrained optimization problems, and the results indicate that it is very suitable than other algorithms from the literature for different formal of constrained optimization problems.

关 键 词：多目标优化 动态选择 约束优化问题 Pareto集 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]