电子学报 ›› 2018, Vol. 46 ›› Issue (5): 1032-1040.DOI: 10.3969/j.issn.0372-2112.2018.05.002

• 学术论文 • 上一篇    下一篇

基于重新匹配策略的ε约束多目标分解优化算法

张磊1, 毕晓君2, 王艳娇3   

  1. 1. 长江大学电子信息学院, 湖北荆州 434000;
    2. 哈尔滨工程大学信息与通信工程学院, 黑龙江哈尔滨 150000;
    3. 东北电力大学信息工程学院, 吉林吉林 132012
  • 收稿日期:2016-12-29 修回日期:2017-03-28 出版日期:2018-05-25
    • 作者简介:
    • 张磊 男,1987年出生,湖北荆州人.长江大学电子信息学院讲师.研究方向为智能信息处理技术.E-mail:zl12306124@163.com;毕晓君 女,1964年出生,黑龙江哈尔滨人.哈尔滨工程大学信息与通信工程学院教授、博士生导师.研究方向为智能信息处理技术、数字图像处理.;王艳娇 女,1985年出生,吉林吉林人.东北电力大学信息工程学院副教授.研究方向为智能信息处理技术、进化算法.
    • 基金资助:
    • 国家自然科学基金 (No.61175126,No.61501107); 长江大学青年基金 (No.7011502105)

The ε Constrained Multi-objective Decomposition Optimization Algorithm Based on Re-matching Strategy

ZHANG Lei1, BI Xiao-jun2, WANG Yan-jiao3   

  1. 1. College of Electronics & Information, Yangtze University, Jingzhou, Hubei 434000, China;
    2. College of Information and Communication Engineering, Harbin Engineering University, Harbin, Heilongjiang 510000, China;
    3. College of Information Engineering, Northeast Electric Power University, Jilin, Jilin 132012, China
  • Received:2016-12-29 Revised:2017-03-28 Online:2018-05-25 Published:2018-05-25

摘要: 针对MOEA/D算法中权重向量与个体分配不合理,导致种群多样性降低的问题,提出基于重新匹配策略的ε约束多目标分解优化算法.首先,对Tchebycheff分解策略进行理论分析,推导出关于多样性和收敛性的定理,从而为研究MOEA/D算法奠定理论基础.其次,为有效解决由于随机为权重向量分配个体造成种群多样性降低的问题,提出权重向量和个体间的重新匹配策略,合理地为权重向量分配个体,改善种群多样性.最后,提出的个体比较准则较好地兼顾多样性和收敛性,提高了算法的约束多目标优化性能.通过与5种优秀算法的对比实验结果表明,该文算法所求得的近似Pareto最优解集的分布性和收敛性均得到一定提高,相比于对比算法具有一定的优势.

关键词: 约束多目标优化, 分解策略, 重新匹配, ε约束

Abstract: Aiming at the problem that unreasonable distribution between weight vector and individual in MOEA/D reduce diversity,the ε constrained multi-objective decomposition optimization algorithm based on re-matching strategy is proposed.Firstly,through theoretical analysis of the chebycheff decomposition strategy,two theorems about diversity and convergence are gained,which could provide a theoretical basis for the research of MOEA/D.Secondly,in order to solve the problem of diversity reduction caused by random assignment of individual to weight vector,the re-matching strategy is presented for reasonably assigning individual to weight vector,and then diversity is improved.Finally,the suggested individual comparison criterion has a good balance between diversity and convergence,and it increases optimization performance.Comparative experiment results with five excellent algorithms show that our algorithm achieves better diversity and convergence,and our algorithm has a certain advantage.

Key words: constrained multi-objective optimization, decomposition strategy, re-matching, epsilon constraint

中图分类号: