TriBA拓扑结构的并行遗传算法研究与TSP问题的应用计算机软件与理论专业论文.docxVIP

关键词：遗传算法 巡回旅行商 TriBA 拓扑结构 数据迁徙策略 加速比 中北大学学位论文 Research of TriBA-topological master-slave parallel genetic algorithm and TSP application Abstract TSP(Traveling Salesman Problem) is a problem of combination optimization with simple definition but difficult to be solved, which attracts many researchers in various fields including mathematics, physics, biology and artificial intelligence. It has become a standard problem to test a new algorithm of combination optimization. Theoretically speaking, the enumeration not only can solve TSP, but also can get the best answer. But to all computers nowadays, it’s hardly to obtain its best answer in such huge researching space by using common enumeration. Therefore, all kinds of algorithms to solve TSP emerged because of demand. At present, there are more evolutionary algorithm have been applied to the TSP problem, such as simulated annealing algorithm, Artificial Fish School Algorithm\、Neural network algorithm、genetic algorithm. In this paper, the main topic is genetic algorithm solving the TSP problem. Through the analysis of advantages and disadvantages of the parallel genetic algorithm and concludes that genetic algorithm and parallel genetic algorithm is an effective algorithm for the TSP problem with constraint. Aiming at the defect of master-slave parallel algorithm model, we put forward a new kind of TriBA((triplet based architecture) topology structure parallel algorithm model .Based on the TrIBA parallel algorithm model, the now algorithm model is improved from two aspects: Chang the model of data migration. The mode of data migration in TriBA structure including single mode , center mode, parallel mode, nested mode, which changing traditional migrate mode of master-slave model .It helps the best individual spread to every sub-population of TriBA structure. 中北大学学位论文 中北大学学位论文 Change the topology. Using the new mode of population’s allocation, the TriBA structure model can solve the problem of load imbalance in master-sa