改进的蚁群算法求解tsp问题的分析-analysis of improved ant colony algorithm to solve tsp problem.docxVIP

中北大学学位论文 中北大学学位论文 将以上三步算法的改进结合起来，用 MATLAB 模拟，选取了 TSPLIB 中的十个 TSP 问 题进行实验，并与用基本蚁群算法做的结果以及现在应用比较广泛的遗传算法的结果做 了比较。然后运用改进的蚁群算法对以中国 100 城市为代表的中国旅行商问题进行了数 值模拟并与用基本蚁群算法做的结果以及神经网络算法求解的结果做了比较。结果表明 虽然运行时间变长了，但改进后的蚁群算法在全局搜索最优解能力上要好很多，在稳定 性和收敛性方面也有了较大提高。 最后对全文的研究工作进行了总结，并展望了蚁群算法进一步还要研究的课题。 关键词:蚁群算法，旅行商问题，自适应，信息素 ABSTRACT Ant colony optimization algorithm (ant colony optimization, ACO), which was first proposed by Italian scholars named Marco Dorigo, V Maniezzo, A. Colorni in 1992, is a groups intelligent optimization methods based on the development of the biological behavior of marching along the shortest path in the process of ant community grazing. The proposed algorithm is obtained encouraging effect in the combinatorial optimization with NP-hard characteristics problem which is difficult to work in traditional optimization method, and is widely attended by the academia and industry. At present, the ant colony optimization algorithm has become an important branch of the intelligent calculating method, and become the prosperous development of the hot research topic nowadays. The study of the theory of the ant colony optimization algorithm relatively lags behind, especially for that there still need further study for the algorithm parameter selection and the convergence proof. Although, this research method is in its infancy, but some research has shown the superiority of the ant colony system in solving complex optimization problem. At present, the ant colony system has been successfully applied to the solving of traveling salesman problem (TSP), quadratic assignment problem (QAP) and job-shop scheduling problem and has got good experimental results. TSP (Traveling Salesman Problem) is also known as the HuoLangDan Problem or the touring Salesman Problem, in operations research, and it has widely range of applications in operational research, management science and engineering practice. TSP is a famous problem of the combination optimization, and more and more researchers hav