摘要： 改善遗传算法中的“早熟”现象可以通过提供某种机制以恢复群体多样性。基于这种思想，该文参照其他遗传算法改进策略，从弥补丢失模式、提高模式浓度出发，提出了通过保持模式的多样性来保证群体多样性的方法。为了衡量改进策略的有效性，引入了模式再生期望值的概念，并利用模式再生期望值的分析方法分析了一种实用的改进策略。实验数据证明了该策略的有效性。

关键词: 遗传算法, 早熟, 模式权值, 模式再生期望值

Abstract: This paper presents an improved politics for premature to provide a mechanism for resuming population diversity. Population diversity is kept by keeping schema diversity. The means is regenerating schema that lose or less in population. To scale degree of validity, schema regenerate excepted value is introduced. An improved politic is analyzed by used schema regenerate excepted value, and examples show that the method is effective.

Key words: genetic algorithms, premature, schema weight, schema regenerate expected value

中图分类号: 

• TP18