计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (33): 18-21.DOI: 10.3778/j.issn.1002-8331.2008.33.005

• 博士论坛 • 上一篇    下一篇

一种用于多目标优化的混合粒子群优化算法

徐 刚1,瞿金平2   

  1. 1.南昌大学 数学系,南昌 330031
    2.华南理工大学 聚合物新型成型装备国家工程研究中心,广州 510640
  • 收稿日期:2008-08-06 修回日期:2008-09-02 出版日期:2008-11-21 发布日期:2008-11-21
  • 通讯作者: 徐 刚

Hybrid particle swarm optimization algorithm for multi-objective optimization

XU Gang1,QU Jin-ping2   

  1. 1.Department of Mathematics,Nanchang University,Nanchang 330031,China
    2.The National Engineering Research Center of Novel Equipment for Polymer Processing,South China University of Technology,Guangzhou 510640,China
  • Received:2008-08-06 Revised:2008-09-02 Online:2008-11-21 Published:2008-11-21
  • Contact: XU Gang

摘要: 将粒子群算法与局部优化方法相结合,提出了一种混合粒子群多目标优化算法(HMOPSO)。该算法针对粒子群局部优化性能较差的缺点,引入多目标线搜索与粒子群算法相结合的策略,以增强粒子群算法的局部搜索能力。HMOPSO首先运行PSO算法,得到近似的Pareto最优解;然后启动多目标线搜索,发挥传统数值优化算法的优势,对其进行进一步的优化。数值实验表明,HMOPSO具有良好的全局优化性能和较强的局部搜索能力,同时HMOPSO所得的非劣解集在分散性、错误率和逼近程度等量化指标上优于MOPSO。

关键词: 多目标优化, 粒子群算法, 局部搜索, Pareto最优

Abstract: Combining particle swarm search with local search,a hybrid multi-objective particle swarm optimization(HMOPSO) algorithm for multi-objective optimization is proposed.Aiming at the defect of local optimization for PSO,HMOPSO introduces multi-objective linearity search as a means of acceleration and refinement of the solutions of particle swarm search to improve search performance.It first runs the PSO in order to obtain approximative Pareto optimal solutions.Once the MOPSO is over,multi-objective linearity search is then run with each previously obtained solution to find a better solution.Simulation results show that HMOPSO,compared with MOPSO,can improve efficiency of optimization and ensure a better convergence,spacing and error ration to the true Pareto optimal front.

Key words: multi-objective optimization, PSO algorithm, local search, Pareto optimal solution