Author Login Editor-in-Chief Peer Review Editor Work Office Work

Computer Engineering ›› 2010, Vol. 36 ›› Issue (20): 173-175. doi: 10.3969/j.issn.1000-3428.2010.20.061

• Networks and Communications • Previous Articles     Next Articles

Self-adaptive Genetic Algorithm Based on Evolvability

LIN Ming-yu, LI Ming, ZHOU Lin-xia   

  1. (Key Laboratory of Nondestructive Test, Ministry of Education, Nanchang Hangkong University, Nanchang 330063, China)
  • Online:2010-10-20 Published:2010-10-18

基于可进化性的自适应遗传算法

林明玉,黎 明,周琳霞   

  1. (南昌航空大学无损检测教育部重点实验室,南昌 330063)
  • 作者简介:林明玉(1983-),女,硕士研究生,主研方向:图像处理,模式识别;黎 明,教授、博士、博士生导师;周琳霞,副教授
  • 基金资助:
    国家自然科学基金资助项目“动态环境下的元胞遗传算法”(60963002);航空科学基金资助项目“基于图像处理与模式识别技术的疲劳断裂测试与分析”(2008ZD56003)

Abstract: Aiming at the problem of traditional genetic algorithm is easy to involve in local optima, this paper presents a self-adaptive genetic algorithm based on evolvability. The individual evolability as a parameter is put into the nonlinear fitness function which dynamically adjustment with the evolution algebra, and it adjusts dynamically the crossover and mutation probability to runaway the local optima. Experimental results show that this algorithm can improve the survival probability of the individuals with better evolvability but worse fitness, and enhances population diversity and search efficiency.

Key words: individual evolvability, self-adaptive Genetic Algorithm(GA), population diversity

摘要: 针对传统遗传算法容易陷入局部最优解的问题,提出一个基于可进化性的自适应遗传算法。将个体可进化性作为适应度函数的参数加入到随进化代数动态调整的非线性适应度函数中,动态调整整个种群的交叉与变异概率以逸出局部最优。实验结果表明,该算法可改善适应度不高但具有较好进化能力个体的生存概率,且提高了种群多样性与搜索效率。

关键词: 个体可进化性, 自适应遗传算法, 种群多样性

CLC Number: