基于指标选择和密度评估删除的高维多目标进化算法
DOI:
作者:
作者单位:

1.东北大学信息科学与工程学院;2.东北大学

作者简介:

通讯作者:

中图分类号:

TP237

基金项目:

国家自然科学基金,国家自然科学基金项目(面上项目,重点项目,重大项目)


An indicator selection and density estimation deletion-based many-objective evolutionary algorithm
Author:
Affiliation:

College of Information Science and Engineering, Northestern University, Shenyang Liaoning

Fund Project:

The National Natural Science Foundation of China (General Program, Key Program, Major Research Plan)

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    研究表明随着目标维数的增加,大多数多目标进化算法的性能急剧恶化,无法使种群收敛且均匀分布 于Pareto前沿(PF).针对该问题,本文提出一种基于指标选择和密度评估删除的高维多目标进化算法(An indicator selection and density estimation deletion-based many-objective evolutionary algorithm, MaOEA/IS-DED).该算法在环 境选择过程中采用基于Iε+ 指标的选择策略和基于移动的密度评估删除机制协同指导种群进化.具体地,前者选 择Iε+ 指标值最小的一对个体,其在空间中表现为搜索方向最相似的个体;后者利用自身兼顾种群收敛性和多样 性的特性,比较被选的两个个体且删除收敛性和多样性较差的个体.实验结果表明MaOEA/IS-DED算法在处理高 维多目标优化问题时能获得较强的竞争性能.

    Abstract:

    Research indicates that with the increase of objective dimensions, performances for most multi-objective evolutionary algorithms rapidly deteriorate, resulting in the population fails to converge and evenly distribute in PF. In view of the issue, this paper proposes an indicator selection and density estimation deletion-based many-objective evolutionary algorithm (MaOEA/IS-DED). In this algorithm, selection strategy based on Iε+ indicator and deletion mechanism based on shifted-based density estimation (SDE) are used to guide the population evolution. More specifically, the former is designed to find a pair of individuals with the minimum Iε+ indicator values, which denotes these selected individuals have the most similar search directions in space. The latter, taking into account the convergence and diversity of the population, compares these selected individuals and deletes the worse one. Experimental results demonstrate MaOEA/IS-DED can gain the highly competitive performance when dealing with many-objective optimization problems.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2021-10-21
  • 最后修改日期:2022-04-05
  • 录用日期:2022-04-08
  • 在线发布日期:
  • 出版日期: