分布式遗传算法求解TSP问题的CORBA实现研究  

Programming TSP With Distributed Genetic Algorithm and CORBA

作　　者：刘芹[1] 余一娇[1] 谭连生[1] 

机构地区：[1]华中师范大学计算机科学系,武汉430079

出　　处：《计算机工程与应用》2003年第18期83-86,共4页Computer Engineering and Applications

摘　　要：论文提出了基于CORBA的分布式遗传算法,描述了实现该算法的三层软件结构模型,并利用该模型对CTSP问题进行了广泛的计算。计算结果表明该模型存在以下三条规律:最优迁移策略与随机迁移策略对最优解影响几乎一致;当迁移率在20‰附近时,易获得最优解;繁衍代数在25000～30000范围内时,易获得最优解。文中还介绍了应用测试数据验证该分布式遗传算法求解最优性的方法。实验表明基于CORBA的分布式遗传算法是求解TSP问题的一种可行、高效的算法。This paper applies CORBA to parallel genetic algorithm.The CORBA-based distributed genetic algorithm is proposed and the three-layer software architecture model is subsequently described in detail.This model is used in large scale experiments of CTSP and three rules are found,which are that best migration policy is as good as random migra-tion policy,the optimized path is easily available both in the case when migration rate is about 20‰and when multi-ply number ranges from25000to30000.Finally several experiments are carried out to verify these rules.Experiments show the efficiency of our approach in solving Traveling Salesman Problem.

关 键 词：TSP 遗传算法 孤岛模型 分布式计算 CORBA 迁移率 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]