计算机应用 ›› 2018, Vol. 38 ›› Issue (2): 399-404.DOI: 10.11772/j.issn.1001-9081.2017071888

• 人工智能 • 上一篇    下一篇

基于反向学习的自适应差分进化算法

李龙澍, 翁晴晴   

  1. 安徽大学 计算机科学与技术学院, 合肥 230601
  • 收稿日期:2017-08-04 修回日期:2017-09-17 出版日期:2018-02-10 发布日期:2018-02-10
  • 通讯作者: 翁晴晴
  • 作者简介:李龙澍(1956-),男,安徽合肥人,教授,博士,主要研究方向:智能信息处理、智能软件;翁晴晴(1991-),女,安徽亳州人,硕士研究生,主要研究方向:软件测试、测试数据自动生成。
  • 基金资助:
    国家自然科学基金资助项目(61402005)。

Self-adaptive differential evolution algorithm based on opposition-based learning

LI Longshu, WENG Qingqing   

  1. School of Computer Science and Technology, Anhui University, Hefei Anhui 230601, China
  • Received:2017-08-04 Revised:2017-09-17 Online:2018-02-10 Published:2018-02-10
  • Supported by:
    This work is partially by the National Natural Science Foundation of China (61402005).

摘要: 为解决差分进化(DE)算法过早收敛与搜索能力低的问题,讨论对控制参数的动态调整,提出一种基于反向学习的自适应差分进化算法。该算法通过反向精英学习机制来增强种群的局部搜索能力,获取精确度更高的最优个体;同时,采用高斯分布随机性提高单个个体的开发能力,通过扩充种群的多样性,避免算法过早收敛,整体上平衡全局搜索与局部寻优的能力。采用CEC 2014中的6个测试函数进行仿真实验,并与其他差分进化算法进行对比,实验结果表明所提算法在收敛速度、收敛精度及可靠性上表现更优。

关键词: 差分进化, 自适应, 高斯分布, 反向学习

Abstract: Concerning premature convergence and low searching capability of Differential Evolutionary (DE) algorithm, the dynamic adjustment of control parameters was dicussed, and a self-adaptive differential evolution algorithm based on opposition-based learning was proposed. In the proposed algorithm, opposition-based elite learning was used to enhance the local search ability of the population and obtain more accurate optimal individuals; meanwhile, Gaussian distribution was used to improve the exploitation ability of each individual and increase the diversity of the population, which avoids premature convergence of the algorithm and achieves the balance of the global exploitation and local exploitation. Comparison experiments with some other differential evolution algorithms were conducted on six test functions in CEC 2014. The experimental results show that the proposed algorithm outperforms the compared differential evolution algorithms in terms of convergence speed, solution accuracy and reliability.

Key words: Differential Evolution (DE), self-adaptive, Gaussian distribution, opposition-based learning

中图分类号: