计算机工程与应用 ›› 2012, Vol. 48 ›› Issue (6): 35-37.

• 研究、探讨 • 上一篇    下一篇

基于自适应学习搜索框架的混合分布估计算法

张庆彬1,2,刘 波3,田彦平1,贺媛媛1   

  1. 1.石家庄铁路职业技术学院 智能技术研究所,石家庄 050041
    2.大连大学 辽宁省先进设计与智能计算省部共建教育部重点实验室,辽宁 大连 116622
    3.河北省科学院,石家庄 050081
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2012-02-21 发布日期:2012-02-21

Hybrid estimation of distribution algorithm under scope of adaptive learning search

ZHANG Qingbin1,2, LIU Bo3, TIAN Yanping1, HE Yuanyuan1   

  1. 1.Center for Intelligent Systems, Shijiazhuang Institute of Railway Technology, Shijiazhuang 050041, China
    2.Key Lab of Advanced Design and Intelligent Computing, Ministry of Education, Dalian University, Dalian, Liaoning 116622, China
    3.Hebei Academy of Sciences, Shijiazhuang 050081, China
  • Received:1900-01-01 Revised:1900-01-01 Online:2012-02-21 Published:2012-02-21

摘要: 在元启发式算法自适应学习搜索框架下对分布估计算法和模拟退火算法的学习能力、深度搜索和广度搜索强度进行分析,针对分布估计算法广度搜索性能方面存在的问题,提出了一种将模拟退火算法融入分布估计算法的混合优化策略;以旅行商问题为例进行了仿真实验。实验结果表明,混合算法比分布估计算法和模拟退火算法具有更高的优化质量。

关键词: 自适应学习搜索, 分布估计算法, 单变量边缘分布算法, 模拟退火算法, 旅行商问题

Abstract: The ability of learning, intensification and diversification in Estimation of Distribution Algorithm(EDA) and Simulated Annealing(SA) under the scope of Adaptive Learning Search(ALS) is analyzed. Then a hybrid EDA integrated by SA is proposed with the aim of achieving an effective balance between diversification and intensification. Simulation results on TSP show that the proposed algorithm out-performs the standard EDA and SA.

Key words: Adaptive Learning Search(ALS), Estimation of Distribution Algorithm(EDA), Univariate Marginal Distribution Algorithm(UMDA), Simulated Annealing(SA), Trareling Salesman Problem(TSP)