计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (33): 62-65.DOI: 10.3778/j.issn.1002-8331.2008.33.020

• 理论研究 • 上一篇    下一篇

五种粒子群优化模型效率的研究

王娟勤,何东健,孙建敏   

  1. 西北农林科技大学 信息工程学院,陕西 杨凌 712100
  • 收稿日期:2007-12-12 修回日期:2008-05-17 出版日期:2008-11-21 发布日期:2008-11-21
  • 通讯作者: 王娟勤

Research of effectiveness of five particle swarm optimization models

WANG Juan-qin,HE Dong-jian,SUN Jian-min   

  1. College of Information Engineering,Northwest Agriculture & Forestry University,Yangling,Shaanxi 712100,China
  • Received:2007-12-12 Revised:2008-05-17 Online:2008-11-21 Published:2008-11-21
  • Contact: WANG Juan-qin

摘要: 粒子群优化算法按照认知部分和社会部分被区分为5种模型(完全模型、自认知模型、社交模型、非自身社交模型和非自身完全模型)。为了明确5种粒子群优化模型的效率,选用进化计算领域中常用的5种基准函数,分别对5种粒子群优化算法模型设置不同的参数,分析了它们在求解5种基准函数时的成功率、平均函数求值数、最佳适应度等。结果表明:PSO完全模型和非自身完全模型使用收缩系数K在某些参数设置下求解高维问题时即搜索问题的解时效率较高,社交模型和非自身社交模型在一些参数设置下求解Schaffer函数等二维问题的效率最好。

Abstract: The basic particle swarm optimization algorithm is identified as five types of PSO according to its cognition component and social component value,such as PSO Full-Model,PSO Cognitive-Only Model,PSO Social-Only Model,PSO Selfless Model and PSO Selfless Full-model.Compare five PSO models’ effectiveness and efficiency according to their success rate,average function evaluation and their best fitness by applying parameter set and using five benchmark functions.The result is that Full-Model and Selfless Full Model with K are effective in solving the functions with high dimension,Social Model and Selfless Model without K are also effective in solving the functions with less dimension such as Schaffer function.