计算机应用 ›› 2013, Vol. 33 ›› Issue (07): 1926-1929.DOI: 10.11772/j.issn.1001-9081.2013.07.1926

• 人工智能 • 上一篇    下一篇

均衡分布性和收敛性的多目标粒子群优化方法

耿焕同1,2,高军1,2,贾婷婷1,2,吴正雪1,2   

  1. 1. 南京信息工程大学 计算机与软件学院,南京 210044
    2. 南京信息工程大学 江苏省网络监控中心,南京 210044
  • 收稿日期:2013-01-14 修回日期:2013-02-20 出版日期:2013-07-01 发布日期:2013-07-06
  • 通讯作者: 耿焕同
  • 作者简介:耿焕同(1973-),男,安徽绩溪人,教授,博士生导师,CCF高级会员,主要研究方向:计算智能与约束多目标优化、气象资料预处理与资料同化;高军(1987-),男,江苏如东人,硕士研究生,主要研究方向:进化计算、GIS气象应用;贾婷婷(1989-),女,江苏宿迁人,硕士研究生,主要研究方向:进化计算;吴正雪(1990-),女,江苏常州人,硕士研究生,主要研究方向:进化计算。
  • 基金资助:

    “青蓝工程”资助项目(2012);中国气象科学研究院灾害天气国家重点实验室2010年开放课题(2010LASW-A02)

Multi-objective particle swarm optimization method with balanced diversity and convergence

GENG Huantong1,2,GAO Jun1,2,JIA Tingting1,2,WU Zhengxue1,2   

  1. 1. College of Computer and Software, Nanjing University of Information Science and Technology, Nanjing Jiangsu 210044, China
    2. Jiangsu Engineering Center of Network Monitoring, Nanjing University of Information Science and Technology, Nanjing Jiangsu 210044, China
  • Received:2013-01-14 Revised:2013-02-20 Online:2013-07-06 Published:2013-07-01
  • Contact: GENG Huantong

摘要: 粒子群优化(PSO)算法是一种基于群体演化且非常有效的求解多目标优化问题的方法,但因经典算法中粒子进化存在趋同性导致算法易陷入局部Pareto最优前沿,使得解集收敛性和分布性不理想。为此提出了一种均衡分布性和收敛性的多目标粒子群优化(DWMOPSO)算法,算法中每个粒子根据自身在进化过程中记忆的个体最好适应度值构建进化速度,由进化速度的快慢动态调整各粒子惯性权重,增加粒子的多样性,从而提高粒子跳出局部最优解的概率。通过在5个标准测试函数上进行仿真实验,结果表明,与Coello的多目标粒子群优化(MOPSO)算法相比,DWMOPSO算法获得的解集在与真实解集的逼近性和解集的分布性两个方面都有了很大的提高。

关键词: 粒子群优化算法, 多目标优化, 局部最优, 动态惯性权重

Abstract: Particle Swarm Optimization (PSO) algorithm is population-based and it is effective for multi-objective optimization problems. For the convergence of the swarm makes the classical algorithm easily converge to local pareto front, the convergence and diversity of the solution are not satisfactory. This paper proposed an independent dynamic inertia weights method for multi-objective particle swarm optimization (DWMOPSO). It changed each particle's inertia weight according to the evolution speed which was calculated by the value of each particle's best fitness in the history. It improved the probability to escape from the local optima. In comparison with Coello's MOPSO through five standard test functions, the solution of the new algorithm has great improvement both in the convergence to the true Pareto front and diversity.

Key words: Particle Swarm Optimization (PSO) algorithm, multi-objective optimization, local optima, dynamic inertia weight

中图分类号: