通信学报 ›› 2017, Vol. 38 ›› Issue (8): 66-78.doi: 10.11959/j.issn.1000-436x.2017165

• 学术论文 • 上一篇    下一篇

无惯性自适应精英变异反向粒子群忧化算法

康岚兰1,2,董文永1(),宋婉娟1,李康顺3   

  1. 1 武汉大学计算机学院,湖北 武汉 430072
    2 江西理工大学应用科学学院,江西 赣州 341000
    3 华南农业大学信息学院,广东 广州 510642
  • 修回日期:2017-03-30 出版日期:2017-08-01 发布日期:2017-09-07
  • 作者简介:康岚兰(1979-),女,江西赣州人,武汉大学博士生,江西理工大学讲师,主要研究方向为演化计算、机器学习等。|董文永(1973-),男,河南南阳人,博士,武汉大学教授、博士生导师,主要研究方向为演化计算、智能仿真优化、系统控制、机器学习等。|宋婉娟(1980-),女,湖北应城人,武汉大学博士生,主要研究方向为图像处理、机器学习等。|李康顺(1962-),男,江西兴国人,博士,华南农业大学教授、博士生导师,主要研究方向为智能计算、多目标优化、视频流图像识别等。
  • 基金资助:
    国家自然科学基金资助项目(61170305);国家自然科学基金资助项目(60873114);江西省教育厅科学技术研究基金资助项目(GJJ161568);江西省教育厅科学技术研究基金资助项目(GJJ151521)

Non-inertial opposition-based particle swarm optimization with adaptive elite mutation

Lan-lan KANG1,2,Wen-yong DONG1(),Wan-juan SONG1,Kang-shun LI3   

  1. 1 Computer School,Wuhan University,Wuhan 430072,China
    2 Faculty of Applied Science,Jiangxi University of Science and Technology,Ganzhou 341000,China
    3 College of Information,South China Agricultural University,Guangzhou 510642,China
  • Revised:2017-03-30 Online:2017-08-01 Published:2017-09-07
  • Supported by:
    The Nationa1 Natura1 Science Foundation of China(61170305);The Nationa1 Natura1 Science Foundation of China(60873114);Science and Techno1ogy Research in Department of Education of Jiangxi Province(GJJ161568);Science and Techno1ogy Research in Department of Education of Jiangxi Province(GJJ151521)

摘要:

为解决反向粒子群优化算法计算开销大、易陷入局部最优的不足,提出一种无惯性的自适应精英变异反向粒子群优化算法(NOPSO)。NOPSO算法在反向学习方法的基础上,广泛获取环境信息,提出一种无惯性的速度(NIV)更新式来引导粒子飞行轨迹,从而有效加快算法的收敛过程。同时,为避免早熟现象的发生,引入了自适应精英变异策略(AEM),该策略在扩大种群搜索范围的同时,帮助粒子跳出局部最优。NIV 与 AEM 这 2种机制的结合,有效增加了种群多样性,平衡了反向粒子群算法中探索与开发的矛盾。实验结果表明,与主流反向粒子群优化算法相比,NOPSO算法无论是在计算精度还是计算开销上均具有较强的竞争能力。

关键词: 无惯性速度更新式, 一般性反向学习, 自适应精英变异, 粒子群优化

Abstract:

Non-inertia1 opposition-based partic1e swarm optimization with adaptive e1ite mutation(NOPSO)was proposed to overcome the drawbacks,such as,s1ow convergence speed,fa11ing into 1oca1 optimization,of opposition-based partic1e swarm optimization.In addition to increasing the diversity of popu1ation,two mechanisms were introduced to ba1ance the contradiction between exp1oration and exp1oitation during its iterations process.The first one was non-inertia1 ve1ocity(NIV)equation,which aimed to acce1erate the process of convergence of the a1gorithm via better access to and use of environmenta1 information.The second one was adaptive e1ite mutation strategy(AEM),which aimed to avoid trap into 1oca1 optimum.Experimenta1 resu1ts show NOPSO a1gorithm has stronger competitive abi1ity compared with opposition-based partic1e swarm optimizations and its varieties in both ca1cu1ation accuracy and computation cost.

Key words: non-inertia1 ve1ocity equation, genera1ized opposition-based 1earning, adaptive e1ite mutation, partic1e swarm optimization

中图分类号: 

No Suggested Reading articles found!