摘要
针对遗传算法和蚁群算法存在运行时都会出现停滞、早熟等现象,且容易陷入局部最小的特点,提出了一种将两者结合协同演化运行的方法,通过建立对这两种算法状态的评估函数来动态判断其运行状态是否正常,进而动态调整运行的算法,从而最大程度地避免了这两种算法运行时的缺点.对TSP问题进行了实验测试,结果表明:此方法在收敛速度、寻优结果上都较上述两种算法单独运行有着明显的优势.
Genetic algorithm and Ant colony algorithm are classical evolution computing.They can solve the problem of combinatorial optimization.However,when they run,some phenomenons which represents local optimum will be appearing,such as stag nation and precocity.Through analyzing the characteristics of the two above algorithms,we propose a coevolution computing method based on combining the two algorithms.The method can automatically select one algorithm running according to the dynamic evaluation function to decide the status of algorithm.Therefore,the method can avoid the weakness of the two algorithms.Through the experiments on TSP problem,the method has advantages over the genetic algorithm or ant colony algorithm in convergence speed and search optimization results.
出处
《中南民族大学学报（自然科学版）》
CAS
2012年第1期97-100,共4页
Journal of South-Central University for Nationalities：Natural Science Edition
基金
国家自然科学基金项目资助(60803159)
关键词
遗传算法
蚁群算法
协同演化
genetic algorithm
ant colony algorithm
coevolution
