求解TSP的改进蚁群算法  

Improved ant colony algorithm for solving TSP

作　　者：侯文静[1] 马永杰[1] 张燕[1] 石玉军[1] 

机构地区：[1]西北师范大学物理与电子工程学院,兰州730070

出　　处：《计算机应用研究》2010年第6期2087-2089,共3页Application Research of Computers

基　　金：甘肃省自然科学研究基金资助项目(096RJZA115);甘肃省教育厅科研项目(0901B-08)

摘　　要：针对蚁群算法存在的搜索时间长、易限于局部最优解等缺陷,提出了一种改进的蚁群算法。通过在初始化信息素矩阵中采用候选城市列表减少劣质解,在局部搜索中采用聚类进行二次搜索,缩小了算法的搜索范围、改善了解空间的质量,提高了搜索速度。仿真结果表明,改进后的蚁群算法在TSP的求解中,收敛速度和全局寻优能力均得到较大的提高。Aimed at the shortcomings, which needing much time and easier to fall in local optimal solution in the ant colony algorithm, this paper proposed an improved algorithm. Through employing the list of candidate cities in the initial pheromone matrix to decrease inferior solutions and using cluster to do the second search in the local search, it could narrow the searching range of algorithm, could improve the quality of the solution space and raise the searching speed. The simulations result for TSP shows that the algorithm is improved greatly in convergence rate and ability of global optimization.

关 键 词：蚁群算法(ACA) 旅行商问题 候选城市列表 聚类 蚁群系统(ACS) 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]