一种基于全局协同与局部进化的遗传算法  

A Genetic Algorithms Based on Global Coordination and Local Evolution

作　　者：王本年[1] 陈世福[1] 谢俊元[1] 

机构地区：[1]南京大学计算机软件新技术国家重点实验室

出　　处：《计算机工程》2005年第19期29-31,共3页Computer Engineering

基　　金：江苏省自然科学基金资助项目(DK2002081);安徽省教育厅资助项目(2002kj299)

摘　　要：改进了协同进化遗传算法中的协同操作,提出了一种基于全局协同与局部进化的两层框架模型和基于此模型改进的遗传算法(GCLEGA),在高层,采用基于邻域的局部进化算法,旨在加强局部搜索,加速收敛速度,在底层,采用改进的多种群协同进化算法,旨在改善群体的多样性,克服未成熟收敛,两层之间通过提升操作关联,使全局搜索与局部搜索、全局收敛性与收敛速度有机地统一了起来。实验结果显示,GCLEGA在改善未成熟收敛和提高收敛速度两方面都具有良好的性能。This paper improves coordinating operation in co-evolution genetic algorithms, a two-floor model based on global coordination and local evolution as well as an improved genetic algorithms are proposed, on the top floor, a local evolution method based on neighbor field is adopted for enhancing capability of local search and accelerating convergent speed. On the bottom floor, an improved multi-population co-evolution algorithm is used for the diversity of individuals in populations. Upgrading operation acts as the association between the two floors. Experimental results show the improved genetic algorithm can restrain premature convergence and accelerate convergent speed.

关 键 词：遗传算法 未成熟收敛 收敛速度 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]