电子学报 ›› 2018, Vol. 46 ›› Issue (10): 2430-2442.DOI: 10.3969/j.issn.0372-2112.2018.10.017

• 学术论文 • 上一篇    下一篇

GWO与ABC的混合优化算法及其聚类优化

张新明1,2, 王霞1, 康强1, 程金凤1   

  1. 1. 河南师范大学计算机与信息工程学院, 河南新乡 453007;
    2. 河南省高校计算智能与数据挖掘工程技术研究中心, 河南新乡 453007
  • 收稿日期:2018-01-10 修回日期:2018-06-27 出版日期:2018-10-25
    • 作者简介:
    • 张新明,男.1963年出生,湖北孝感人.教授、硕士生导师、CCF会员.主要研究方向是智能优化算法、数字图像处理和模式识别等.E-mail:xinmingzhang@126.com;王霞,女.1993年出生,河南新乡人.硕士研究生.主要研究方向是智能优化算法和数字图像分割.E-mail:wangxia0801@qq.com
    • 基金资助:
    • 河南省重点科技攻关项目 (No.132102110209); 河南省高等学校重点科研项目 (No.19A520026)

Hybrid Grey Wolf Optimizer with Artificial Bee Colony and Its Application to Clustering Optimization

ZHANG Xin-ming1,2, WANG Xia1, KANG Qiang1, CHENG Jin-feng1   

  1. 1.College of Computer and Information Engineering, Henan Normal University, Xinxiang, Henan 453007, China;
    2.Engineering Technology Research Center for Computing Intelligence & Data Mining of Henan Province, Xinxiang, Henan 453007, China
  • Received:2018-01-10 Revised:2018-06-27 Online:2018-10-25 Published:2018-10-25

摘要: 灰狼优化算法(Grey Wolf Optimizer,GWO)和人工蜂群算法(Artificial Bee Colony,ABC)是两种流行且高效的群智能优化算法.GWO具有局部搜索能力强等优势,但存在全局搜索能力弱等缺陷;而ABC具有全局搜索能力强等优点,但存在收敛速度慢等不足.为实现二者优势互补,提出了一种GWO与ABC的混合算法(Hybrid GWO with ABC,HGWOA).首先,使用静态贪心算法替代ABC雇佣蜂阶段中的动态贪心算法来强化探索能力,同时为弥补其收敛速度降低的不足,提出一种新型的搜索蜜源方式;然后,去掉影响收敛速度的侦查蜂阶段,在雇佣蜂阶段再添加反向学习策略,以避免搜索陷入局部最优;最后,为了平衡以上雇佣蜂阶段的探索能力,在观察蜂阶段,自适应融合GWO,以便增强开采能力和提高优化效率.大量的函数优化和聚类优化的实验结果表明,与state-of-the-art方法相比,HGWOA具有更好的优化性能及更强的普适性,且能更好地解决聚类优化问题.

关键词: 智能优化算法, 灰狼优化算法, 人工蜂群算法, 混合优化算法, 聚类优化

Abstract: Grey Wolf Optimizer (GWO) and Artificial Bee Colony (ABC) are two popular and efficient intelligent optimization algorithms. GWO has some features such as strong exploitation but weak exploration. ABC has other ones such as strong global search ability but slow convergence. In order to realize their complementary advantages, a hybrid GWO with ABC (HGWOA) was proposed. Firstly, a static greedy algorithm was used to replace the dynamic greedy algorithm in the employed bee phase to enhance the exploration ability, and a new search method was created to make up for the lost convergence quality. Secondly, the scout bee phase which affects the convergence speed was removed, and an opposition learning strategy was embedded into the employed bee phase to keep the algorithm from falling into the local optima. Finally, in order to balance the exploration ability of the employed bee phase, GWO was added to the onlooker bee phase to strengthen the exploitation and improve the optimization efficiency. Experimental results on many function and clustering optimization problems show that compared with state-of-the-art methods, HGWOA has better optimization performance and stronger universality and it can solve clustering optimization problems more efficiently.

Key words: intelligent optimization algorithm, grey wolf optimizer, artificial bee colony algorithm, hybrid optimization algorithm, clustering optimization

中图分类号: