基于R2指标和目标空间分解的高维多目标粒子群优化算法
作者:
作者单位:

安徽工业大学

作者简介:

通讯作者:

中图分类号:

TP18

基金项目:

国家自然科学基金项目(面上项目,重点项目,重大项目)


R2 Indicator and Kuxing Guided Many-objective Particle Swarm Optimizer
Author:
Affiliation:

Anhui University of Technology

Fund Project:

the National Natural Science Foundation of China, under Grants 61903003

  • 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:

    基于R2指标和分解策略的多目标粒子群优化算法(R2-MOPSO)在求解两三个目标优化问题时具有较好的收敛性和多样性,但在求解高维多目标优化问题时难度较大,本文提出了一种基于R2指标和目标空间分解}的高维多目标粒子群优化算法R2-MOPSO-II).首先借鉴R2指标和目标空间分解策略综合权衡选择过程的收敛性和多样性设计双层档案维护策略;然后设计一种新的向导选择策略连接目标空间和决策变量空间,进而提出一种基于双层档案的速度和位置更新策略权衡粒子群优化算法的勘探和开采能力;}最后通过引入高斯学习策略和精英学习策略防止粒子陷入局部最优前沿.数值仿真实验结果表明所提出算法在求解DTLZ和WFG测试问题时具有较好的收敛性和多样性.

    Abstract:

    R2 indicator and decomposition based multiple particle swarm optimizer, named R2-MOPSO, is suitable for solving two and three objectives optimization problems in terms of the convergence and diversity. However, it is difficult for R2-MOPSO to address many-objective optimization problems (MaOPs).We propose an R2 indicator and objective space partition based many-objective particle swarm optimizer, named R2-MOPSO-II, to solve MaOPs. Firstly, a new bi-level archive maintainence strategy is introduced to balance the convergence and diversity after considering the R2 indicator and the objective space partition strategy.Secondly, a new leader selection strategy gives the bridge between objective space and decision variable space. The modified velocity updated equation based on bi-level archive is introduced to balance the exploration and exploitation. Finally, Gaussian learning strategy and elitist learning strategy are embedded into our proposed algorithm to help the algorithm jump out of local PF. The numerical simulation results have validated that the proposed algorithm has achieved better convergence and diversity.

    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:2020-02-05
  • 最后修改日期:2020-07-26
  • 录用日期:2020-05-29
  • 在线发布日期:
  • 出版日期: