基于可进化性的快速遗传算法  

Fast Genetic Algorithm Based on Evolvability

作　　者：何大阔[1] 王福利[1] 

机构地区：[1]东北大学信息科学与工程学院,辽宁沈阳110004

出　　处：《东北大学学报（自然科学版）》2002年第7期628-631,共4页Journal of Northeastern University(Natural Science)

基　　金：教育部骨干教师基金资助项目;辽宁省自然科学基金资助项目(0 0 2 0 13 )

摘　　要：给出三类可进化性较强的个体的定义 ,提出充分重视个体的可进化性并对可进化性强的个体加以引导 ,来提高遗传算法搜索效率的思想·同时考虑算法全局收敛性能而采用一种并行群体进化结构 ,设计了旨在提高遗传算法寻优效率的基于可进化性的快速遗传算法 。The searching efficiency of the conventional genetic algorithms was investigated. The searching efficiency depends on the evolvability of individuals. The higher evolvability the individuals have,the higher searching efficiency the genetic algorithm has. three kinds of individuals with high evolvabioty were defined. the searching efficiency of the genetic algorithm can be improved by attaching importance to the evolvability of the individuals sufficiently and guiding the individuals with high evolvability. When global convergence is taken into account,a paralleled population heuristic structure is adopted,and the fast genetic algorithm based on evolvability is presented to improve the searching efficiency.

关 键 词：快速遗传算法(EGA) 个体 可进化性 加速群体 并行群体进化结构 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]