遗传算法中群体多样性评价指标的研究  

Evaluation index of population diversity in Genetic algorithm

作　　者：何燕平[1] 何辉[1] 张筱磊[2] 

机构地区：[1]哈尔滨工业大学(威海)计算机科学与技术学院,山东威海264209 [2]哈尔滨工业大学航天学院,150008

出　　处：《哈尔滨工业大学学报》2009年第3期67-70,共4页Journal of Harbin Institute of Technology

摘　　要：本文针对遗传算法中早熟问题进行了分析,提出了评价群体多样性的两个性能指标,在此基础上结合模糊逻辑调节遗传算法的交叉和变异概率.对一组函数优化问题对标准算法和优化算法进行测试,测试结果表明基于模糊逻辑控制的遗传算法的性能要优于标准遗传算法.证明本文提出的群体多样性的评价指标是有效的.The premature convergence problem of genetic algorithm is analyzed, and two evaluation indexes are proposed in this paper. The fuzzy logic combined with these two evaluation indexes is applied to the adjustment of mutation and crossover probability of genetic algorithm. The experiments on functional optimization indicate that this kind of genetic algorithm can get fast convergence and find better results. It is proved that this kind of evaluation index is useful for analyzing the premature convergence problem of genetic algorithm.

关 键 词：群体多样性 评价指标 遗传算法 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]