计算机科学与探索 ›› 2014, Vol. 8 ›› Issue (12): 1502-1510.DOI: 10.3778/j.issn.1673-9418.1409031

• 人工智能与模式识别 • 上一篇    下一篇

多种群多策略的并行差分进化算法

陈  颖1,林  盈2,胡晓敏3+   

  1. 1. 中山大学 计算机科学系,广州 510006
    2. 中山大学 心理学系,广州 510275
    3. 中山大学 公共卫生学院 卫生信息研究中心 广东省卫生信息学重点实验室,广州 510080
  • 出版日期:2014-12-01 发布日期:2014-12-08

Parallel Differential Evolution with Multi-Population and Multi-Strategy

CHEN Ying1, LIN Ying2, HU Xiaomin3+   

  1. 1. Department of Computer Science, Sun Yat-sen University, Guangzhou 510006, China
    2. Department of Psychology, Sun Yat-sen University, Guangzhou 510275, China
    3. Guangdong Key Laboratory of Health Informatics, Health Information Research Center, School of Public Health, Sun Yat-sen University, Guangzhou 510080, China
  • Online:2014-12-01 Published:2014-12-08

摘要: 为了更好地提高并行差分进化算法的求解精度和计算效率,实现适用于解决多种优化问题的鲁棒性算法,提出了一种多种群多策略的并行差分进化算法。该算法将种群划分为多个子种群,不同的子种群分别采用不同的差分进化策略。多个子种群各自独立进化,互不干扰,每隔一定代数才进行种群间的通信交流。通过利用多种群实现多种优化策略,并采用并行方式,使得算法可以采用不同的优化策略进行搜索,更加节省计算时间。数值实验结果表明,该算法在求解不同类型的优化问题时都具有良好的计算能力和效率。

关键词: 多种群, 多策略, 并行, 差分进化

Abstract: In order to improve the accuracy and efficiency of parallel differential evolution (DE), this paper proposes a parallel differential evolution with multi-population and multi-strategy, which provides a way to rebustly address various optimization problems. This algorithm divides an initial population into several sub-populations, and then they evolve with different DE strategies. The sub-populations evolve independently at first, and then communicate with each other at regular intervals. By using the proposed multi-population and multi-strategy, the parallel realization of the algorithm can save the computation time while searching with different optimization strategies. The experimental results show that the proposed algorithm is feasible and effective for solving different optimization problems.

Key words: multi-population, multi-strategy, parallel, differential evolution