Computer Engineering and Applications ›› 2020, Vol. 56 ›› Issue (15): 162-171.DOI: 10.3778/j.issn.1002-8331.1905-0348

Previous Articles     Next Articles

Adaptive Combine and Split Multi-population Differential Evolution Algorithm

WANG Hao, LI Jun, ZHOU Rong   

  1. 1.College of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan 430065, China
    2.Hubei Province Key Laboratory of Intelligent Information Processing and Real-time Industrial System, Wuhan 430065, China
  • Online:2020-08-01 Published:2020-07-30

自适应合并与分裂的多种群差分进化算法

王浩,李俊,周蓉   

  1. 1.武汉科技大学 计算机科学与技术学院,武汉 430065
    2.智能信息处理与实时工业系统湖北省重点实验室,武汉 430065

Abstract:

In order to solve the problem of premature convergence and search stagnation in DE algorithm, an adaptive combine and split multi-population DE algorithm is proposed. The algorithm divides the population into multiple subpopulations, and introduces the advantages and disadvantages factors of the subpopulations to evaluate the advantages and disadvantages of the population, so as to realize the adaptive combining and splitting among the populations. For each individual in the population, the mutation operator based on the elite pool learning is adopted, and the adaptive learning adjustment is carried out by combining the excellent individuals, so that the algorithm can achieve the balance between global search and local search ability. In the later stage of the algorithm, the perturbation strategy is introduced to ensure the fast convergence of the algorithm and effectively jump out of the local extreme point, so as to improve the accuracy of the algorithm. The experimental results of 30 standard test functions show that the improved algorithm can effectively solve the problems of precocity and local optimization.

Key words: differential evolution, multiple populations, population advantage and disadvantage factor, adaptive combining and splitting, elite pool learning, disturbance strategy

摘要:

针对差分进化(DE)算法存在的早熟收敛与搜索停滞问题,提出了自适应合并与分裂的多种群差分进化算法。算法将种群划分为多个子种群,引入子种群优劣因子来评价种群的优劣性,实现种群间的自适应合并与分裂;对于种群中的各个个体,采取基于精英池学习的变异算子,结合优秀个体进行自适应学习调整,使算法达到全局搜索与局部搜索能力的平衡;在算法后期引入扰乱策略,保证算法快速收敛的同时有效地跳出局部极值点,提高算法寻优的精度。在30个标准测试函数的实验结果表明,改进算法能有效解决早熟和陷入局部最优的问题。

关键词: 差分进化, 多种群, 种群优劣因子, 自适应合并与分裂, 精英池学习, 扰乱策略