Computer Engineering and Applications ›› 2007, Vol. 43 ›› Issue (20): 25-27.

• 博士论坛 • Previous Articles     Next Articles

Self-adaptive proportional selection strategy for genetic algorithm

YANG Xin-wu,LIU Chun-nian   

  1. Multimedia and Intelligent Software Technology Beijing Municipal Key Laboratory,the College of Computer Science,Beijing University of Technology,Beijing 100022,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-07-11 Published:2007-07-11
  • Contact: YANG Xin-wu

遗传算法中自适应的比例选择策略

杨新武,刘椿年   

  1. 北京工业大学计算机学院 多媒体与智能软件技术北京市重点实验室,北京 100022
  • 通讯作者: 杨新武

Abstract: The fitness-proportionate selection is the basic selection method for genetic algorithm,but it tends toward resulting in the premature convergence and the random walk phenomena.The paper analyzes causes of the two phenomenas by experiment,and argues for adopting a self-adaptive proportional selection strategy to adjust dynamically the selection intensity according to the change of the population state,so adjust dynamically the balance of the refining performance and the reforming performance of genetic algorithm.The analysis and comparative experiment show that the new selection strtegy can overcome the premature convergence and the random walk phenomena.

Key words: genetic algorithm, self-adaptive selection, premature convergence, random walk

摘要: 基于适应度比例的选择策略是遗传算法的基本选择方法,但采用该策略易出现未成熟收敛和随机漫游现象。通过实验分析了两种现象的成因,提出采用自适应的比例选择策略来依据种群性状的改变而动态地调整选择压力,进而调整算法求精和求泛能力的平衡。分析和对比实验证实,新的选择策略可有效克服未成熟收敛和随机漫游现象。

关键词: 遗传算法, 自适应选择, 未成熟收敛, 随机漫游