J4 ›› 2010, Vol. 37 ›› Issue (5): 846-851.doi: 10.3969/j.issn.1001-2400.2010.05.013

• 研究论文 • 上一篇    下一篇

免疫非支配自适应粒子群多目标优化

马晶晶;杨咚咚;焦李成   

  1. (西安电子科技大学 智能感知与图像理解教育部重点实验室,陕西 西安  710071)
  • 收稿日期:2010-03-11 出版日期:2010-10-20 发布日期:2010-10-11
  • 通讯作者: 马晶晶
  • 作者简介:马晶晶(1983-),女,西安电子科技大学博士研究生,E-mail: smallpig32@sina.com.
  • 基金资助:

    国家“863”计划资助项目(2009AA12Z210);陕西省“13115”科技创新工程重大科技专项资助项目(2008ZDKG-37);国家自然科学基金资助项目(60703107,60703108,60803098和60872135);陕西省“13115”科技创新工程重大科技专项资助项目(2008ZDKG-37);高等学校学科创新引智计划(111计划)资助项目(B07048);中央高校基本科研业务费专项基金资助项目(JY10000902001,JY10000903007,JY10000902033,JY10000902039,JY10000902040和JY10000902042)

Immune nondominated adaptive particle  swarm multi-objective optimization

MA Jing-jing;YANG Dong-dong;JIAO Li-cheng   

  1. (Ministry of Education Key Lab. of Intelligent Perception and Image Understanding, Xidian Univ., Xi'an  710071, China)
  • Received:2010-03-11 Online:2010-10-20 Published:2010-10-11
  • Contact: MA Jing-jing

摘要:

为了更加有效地利用粒子群优化技术来解决多目标优化问题,提出了非支配粒子群的概念,并根据当前代的非支配解的数量自适应地构建粒子惯性权,动态调节粒子进化过程.同时,利用人工免疫系统中的克隆选择机制来对非支配粒子进行增殖扩散,保持粒子种群的多样性.通过系统的实验验证,与当前多目标优化领域最有代表性的NSGA-Ⅱ, PESA-Ⅱ和SPEAⅡ相比,表明该算法在收敛性和多样性方面均取得了一定的优势,且时间复杂度明显较低.

关键词: 进化计算, 多目标优化, 人工免疫系统, 粒子群优化

Abstract:

Immune nondominated adaptive particle swarm multi-objective optimization(INPSMO) is studied. The nondominated particle swarm is proposed to efficiently apply particle swarm optimization into solving multi-objective optimization problems. Meanwhile, currently discovered non-dominated solutions are utilized to dynamically and adaptively adjust the inertia weights, which is critical to the evolutionary process of particles. Besides, the clone selection principle in the artificial immune system is employed for particle proliferation, which is beneficial to maintaining population diversity. Compared with three state-of-the-art multi-objective algorithms, namely, NSGA-Ⅱ, SPEA2 and PESA-Ⅱ, INPSMO achieves comparable results in terms of convergence and diversity metrics. Better computational complexity is also obtained by INPSMO.

Key words: evolutionary computation, multi-objective optimization, artificial immune system, particle swarm optimization