作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (14): 152-154. doi: 10.3969/j.issn.1000-3428.2011.14.050

• 人工智能及识别技术 • 上一篇    下一篇

基于均匀设计的聚类多目标粒子群优化算法

刘衍民 1,2,牛 奔 3,赵庆祯 2   

  1. (1. 遵义师范学院数学系,贵州 遵义 563002;2. 山东师范大学管理与经济学院,济南 250014; 3. 深圳大学管理学院,广东 深圳 518060)
  • 收稿日期:2010-12-20 出版日期:2011-07-20 发布日期:2011-07-20
  • 作者简介:刘衍民(1978-),男,讲师、博士,主研方向:进化计算,运筹学;牛 奔,副教授、博士;赵庆祯,教授、博士生导师
  • 基金资助:
    广东省自然科学基金资助项目(9451806001002294);贵州省教育厅社科基金资助项目(0705204);山东省科技攻关计划基金资助项目(2009GG10001008)

Clustering Multi-objective Particle Swarm Optimization Algorithm Based on Uniform Design

LIU Yan-min 1,2, NIU Ben 3, ZHAO Qing-zhen 2   

  1. (1. Department of Math, Zunyi Normal College, Zunyi 563002, China; 2. School of Management and Economics, Shandong Normal University, Jinan 250014, China; 3. College of Management, Shenzhen University, Shenzhen 518060, China)
  • Received:2010-12-20 Online:2011-07-20 Published:2011-07-20

摘要: 为更有效地求解多目标优化问题,提出一种基于均匀设计的聚类多目标粒子群算法UCMOPSO。采用基于均匀设计的交叉操作尽可能地获得目标空间中均匀分布的非劣解,帮助种群跳出局部最优解,并通过一种新的聚类操作选择外部存档中有代表性的非劣解,从而控制外部存档规模,降低计算复杂度。对基准函数的测试结果表明,UCMOPSO算法相比同类算法在收敛性和分布性方面具有优势。

关键词: 均匀设计, 多目标优化, 聚类, 粒子群优化算法, 外部存档

Abstract: In order to solve multi-objective problems efficiently, this paper proposes a clustering multi-objective Particle Swarm Optimization (PSO) algorithm based on uniform design named UCMOPSO. Crossover operation based on uniform design is adjusted to get uniformly distributed solutions in objective space to help swarm to escape from local optima, and a new clustering operator is introduced to select the representative non-dominated solutions, which decreases the computation complexity and limits the size of the external archive. Experimental results based on benchmark functions indicate that UCMOPSO has superiority in convergence and distribution compared with other algorithms.

Key words: uniform design, multi-objective optimization, clustering, Particle Swarm Optimization(PSO) algorithm, external archive

中图分类号: