遗传算法在蚁群算法中的融合研究  

Study on Fusing Genetic Algorithm into Ant Colony Algorithm

作　　者：肖宏峰[1] 谭冠政[2] 

机构地区：[1]湖南师范大学计算机教学部,湖南长沙410081 [2]中南大学信息科学与工程学院机器人研究所,湖南长沙410083

出　　处：《小型微型计算机系统》2009年第3期512-517,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(5027150)资助;教育部博士点基金项目(20040533035)资助

摘　　要：提出一种新的求连续空间最优值的蚁群算法.结合遗传算法和蚁群算法的各自优点以及两种算法融合的基础,提出遗传算法融入到蚁群算法的两种新策略:第一种策略是先利用遗传算法具有比较强的全局搜索能力,在大范围内寻找一组解,然后以此为基础,用蚁群算法快速寻找最优解X*best;另一种策略是利用遗传算法的交叉操作产生蚁群算法的新的旅行路径,以此提高蚁群算法的全局搜索能力.用上述两种策略构造了两个基于遗传算法的混合蚁群算法.文中用测试函数Rosenbrock和Shubert从收敛速度、命中率、计算精度等方面验证了混合蚁群算法的正确性.Propose a new ant colony system （ACS） for obtaining optimal value of continuous space. After comparing the advantages and disadvantages between genetic algorithm （GA） and ACS and analyzing their fusion basis, propose two new strategies of fusing GA into ACS： one is using genetic algorithm to obtain some rough solutions to a problem and then fast obtaining the more precise solutions X^＊best by ACS, the other is that using crossover operator in GA generates two new travel paths in ACS in order to improve the global search ability of ACS. Based on above two ideas, two new hybrid ant colony systems based on GA are built, which are called GA-HACS-Ⅰ and GA-HACS-Ⅱ. At last, verify the correction of GA-HACS-Ⅰ and GA-HACS-Ⅱ from converge speed, hit rate and computational accuracy by test function Rosenbrock and test function Shubert.

关 键 词：遗传算法 混合蚁群算法 算法融合 连续空间优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]