《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (S1): 154-162.DOI: 10.11772/j.issn.1001-9081.2022060815

• 先进计算 • 上一篇    

基于自动快速密度峰值聚类的粒子群动态优化算法

李飞1,2,3,4(), 乐强2, 潘紫微1, 孙怡宁3, 余晓流4   

  1. 1.马鞍山学院 智造工程学院,安徽 马鞍山,243100
    2.安徽工业大学 电气与信息工程学院,安徽 马鞍山,243032
    3.中国科学院 合肥物质科学研究院,合肥230031
    4.特种重载机器人安徽省重点实验室(安徽工业大学),安徽 马鞍山,243032
  • 收稿日期:2022-06-04 修回日期:2022-08-12 接受日期:2022-08-13 发布日期:2023-07-04 出版日期:2023-06-30
  • 通讯作者: 李飞
  • 作者简介:李飞(1988—),男,安徽太和人,副教授,博士,CCF会员,主要研究方向:进化多目标优化、动态优化.lanceleeneu@126.com
    乐强(1998—),男,安徽池州人,硕士研究生,主要研究方向:动态优化、多模态优化
    潘紫微(1957—),男,江苏南京人,教授,博士,主要研究方向:机械设计及其自动化
    孙怡宁(1963—),男,安徽太湖人,教授,博士生导师,博士,主要研究方向:生物医学信息工程、智能感知
    余晓流(1962—),男,安徽岳西人,教授,博士生导师,博士,主要研究方向:特种重载机器人动态优化、机械设计。
  • 基金资助:
    国家自然科学基金青年项目(61903003);安徽省自然科学基金青年项目(2008085QE227);特种重载机器人安徽省重点实验室开放课题(TZJQR001?2021)

Dynamic particle swarm optimization algorithm based on automatic fast density peak clustering

Fei LI1,2,3,4(), Qiang YUE2, Ziwei PAN1, Yining SUN3, Xiaoliu YU4   

  1. 1.College of Intelligent Manufacturing Engineering,Ma’anshan University,Ma’anshan Anhui 243100,China
    2.College of Electrical and Information Engineering,Anhui University of Technology,Ma’anshan Anhui 243032,China
    3.Hefei Institutes of Physical Science,Chinese Academy of Sciences,Hefei Anhui 230031,China
    4.Anhui Province Key Laboratory of Special Heavy Load Robot (Anhui University of Technology),Ma’anshan Anhui 243032,China
  • Received:2022-06-04 Revised:2022-08-12 Accepted:2022-08-13 Online:2023-07-04 Published:2023-06-30
  • Contact: Fei LI

摘要:

针对常规多种群方法在求解动态优化问题时往往存在多样性缺失现象,提出一种基于自动快速密度峰值聚类的粒子群动态优化算法(DPCPSO)。首先,利用自动快速密度峰值聚类通过粒子的自身密度和相对距离创建无敏感参数子种群;然后,使用粒子群优化(PSO)来寻找最优解,在搜索过程中采用停滞计数器来判断粒子是否停滞,防止种群过早收敛;最后,采用最优粒子重定位策略响应环境变化。为了验证所提出算法的性能,在移动峰值基准(MPB)和广义动态基准生成器(GDBG)测试问题上进行了仿真实验。仿真实验中,所提算法性能与基于亲和传播聚类的动态优化算法(APCPSO)、基于聚类的动态优化(CPSO)算法等其他先进算法相比较,在峰值数大于20以及变化频率为2 000和3 000时均取得良好的结果。实验结果表明,所提算法更适合求解多模态和快变特性的动态优化问题。

关键词: 动态优化问题, 多种群方法, 快速密度峰值聚类, 停滞检测, 最优粒子重定位策略

Abstract:

Aiming at the lack of diversity in solving dynamic optimization problems by conventional multi-population methods, a dynamic Particle Swarm Optimization algorithm based on automatic fast Density Peak Clustering (DPCPSO) was proposed. Firstly, the automatic fast peak density clustering method was used to generate the subpopulations without introducing any sensitive parameter through the densities and relative distances of particles. Therefore, Particle Swarm Optimization (PSO) was used to find the optimal solution. In the search process, stagnation counter was used to determine whether the particles are stagnant to prevent premature convergence of the population. Finally, the optimal particle relocation strategy was adopted to respond to the environmental changes. To validate the performance of the proposed algorithm, a variety of experiments were carried out on Moving Peaks Benchmark (MPB) and Generalized Dynamic Benchmark Generator (GBDG) problems. In simulation experiments, the performance of the proposed algorithm was compared with other advanced algorithms such as Affinity Propagation on Clustering based PSO for dynamic optimization algorithm (APCPSO) and Clustering PSO (CPSO) algorithm. Good results were obtained when the peak number was greater than 20 and the change frequency was 2 000 and 3 000. Experimental results show that the proposed algorithm was more suitable for solving dynamic optimization problems with multimodal and fast change characteristics.

Key words: Dynamic Optimization Problem (DOP), multi-population method, Density Peak Clustering (DPC), stagnation detection, optimal particle relocation strategy

中图分类号: