多物流配送中心选址及求解

Multi distribution Center Location Problem and Its Resolution

摘要:

经典蚁群算法不能直接用于求解多配送中心选址问题（MDLP），据此，将MDLP映射为扩展K TSP过程并设计了改进的蚁群算法.改变了经典蚁群算法禁忌表的设置方式，算法运行时，给蚁群建立一个共享禁忌表，里面存放所有蚂蚁访问过的客户点，任何蚂蚁只能选择共享禁忌表未曾记录的客户点，从而增强蚂蚁间的信息交流，促进它们的分工与协作，使蚂蚁无遗漏无重复地遍历各配送点并找出问题的最优解.为提高算法的求解性能，在蚂蚁的选择规则里加入了代价引导函数，使用2 opt策略优化可行解并优化了信息素的更新方式.仿真算例及算法对比表明，模型和算法可以有效地表达和求解MDLP.

Abstract:

To solve multi distribution center location problem（MDLP) directly, MDLP was mapped into the process of expanded K TSP and improved ants colony algorithm was designed. The setting mode of tabu for classic ants colony algorithm was changed, specifically, when algorithm run, a sharing tabu for ants colony was constructed and all the distribution points the ants had traversed were put into it, any ants could only choose the customer point unrecorded in it, thereby this intensified exchange of information among ants and promoted their division and cooperation, so that the ants visited all the distribution sites exhaustively without repetition, and finally found the optimal solution. In order to improve the algorithm’s performance, price guided function was added to ants’ selection rule, and 2 opt strategy was used to optimize the feasible solution, and optimized pheromone update policy was designed. The simulation example and the algorithm comparison show that the model and the algorithm may express and solve MDLP effectively.