求解TSP问题的自适应邻域遗传算法-计算机工程与应用.PDF

20 2010 ，46 （27 ） Computer Engineering and Applications 计算机工程与应用 求解TSP 问题的自适应邻域遗传算法 汪金刚，罗辞勇 WANG Jin-gang ，LUO Ci-yong 重庆大学 输配电装备及系统安全与新技术国家重点实验室，重庆 400044 State Key Laboratory of Power Transmission Equipment System Security and New Technology ，Chongqing University ， Chongqing 400044 ，China E-mail:jingang_023@163.com WANG Jin-gang ，LUO Ci-yong.Adaptive neighborhood method Genetic Algorithm for solving Travelling Salesman Problem.Computer Engineering and Applications ，2010 ，46 （27 ）：20-24. Abstract ：In this paper ，the travelling salesman problem （TSP ）is solved by using the adaptive neighborhood method ge- netic algorithm.In adaptive neighborhood method （ANM ）one mimics the traveller whose rule shows that the next city is not always the nearest as-yet-unvisited location but it ’s randomly selected from the unvisited cities which are further than the nearest city in adaptive neighborhood.While solving the TSP ，ANM is used to create the initial population at first ，then itera- tions are done through selection ，cross and mutation operation.In selection ，the proposed algorithm only keeps 90% samples from the previous generation ；the remaining is supplied by the new samples created by ANM.The results of simulation show that this approach is more competitive than other algorithms. Key words ：Genetic Algorithm（GA ）；Travelling Salesman Problem（TSP ）；nearest neighbor ；adaptive neighborhood method 摘 要：提出结合自适应邻域法与遗传算法来求解TSP 问题。在自适应邻域法中，从某个城市出发，下一城市不一定是其最近城 市，而是在比其最近城市稍远的邻域范围进行动态随机选取。在求解TSP 时，采用自适应邻域法对种群初始化，然后采用选择、 交叉、变异进行迭代，在选择中仅保留父代90 ％的样本，剩下的采用自适应邻域法产生新样本进行补充。仿真实验结果表明所提 算法与其他算法相比具有竞争能力。 关键词：遗传算法；旅行商问题；最近邻法；自适应邻域法 DOI ：10.3778/j.issn. 1002-8331.2010.27.005 文章编号：1002-8331（2010 ）27-0020-05 文献标识码：Ａ 中图分类号：TP301 1 引言 的遗传和进化过程而形成的一种自适应全局优化概率搜索算