Abstract: A novel multi-objective optimization differential evolution algorithm is proposed for solving constrained optimization problems. In the process of population evolution, the individuals generation based on good-point-set method is introduced into the evolutionary algorithm initial step. The constrained optimization problem is converted into a multi-objective optimization problem. The population is divided into Non-Pareto set and Pareto set based on multi-objective optimization technique. In order to improve global convergence of the proposed algorithm, DE/best/1 mutation scheme and DE/rand/1 mutation scheme are used to the Non-Pareto set and the Pareto set respectively. The experimental results show that the proposed algorithm can get high performance while dealing with various complex problems.

Key words: constrained optimization problems, differential evolution algorithm, multi-objective optimization, good-point-set

摘要： 提出一种新的多目标优化差分进化算法用于求解约束优化问题。该算法利用佳点集方法初始化个体以维持种群的多样性。将约束优化问题转化为两个目标的多目标优化问题。基于Pareto支配关系，将种群分为Pareto子集和Non-Pareto子集，结合差分进化算法两种不同变异策略的特点，对Non-Pareto子集和Pareto子集分别采用DE/best/1变异策略和DE/rand/1变异策略。数值实验结果表明该算法具有较好的寻优效果。

关键词: 约束优化问题, 差分进化算法, 多目标优化, 佳点集