【作者】 赵冰洁；

【导师】 李宗平；

【作者基本信息】 西南交通大学 ， 交通运输规划与管理， 2004， 硕士

【摘要】 随着电子商务的飞速发展，物流业发生了新的重要变革，货物配送工作是其中一个重要的环节。在配送中心的货物配送工作中，配送方案优化问题是一个极具魅力的问题，优质的配送方案将产生极其可观的经济效益。 论文在分析配送中心配送方案优化方法的发展过程的基础上，研究了不同情况下的VRP问题的算法设计。根据顾客对配送时间的要求，分别建立了在不考虑配送中心库存成本的有软硬时间窗的单周期配送方案优化模型与在考虑配送中心库存成本的多周期配送方案优化模型。 在建模过程中分析了配送中心配送资源在不同情况下模型建立与求解方法。如果配送中心的资源有限，配送成本受顾客的库存影响，随机库存分配问题的解决决定着整个问题的解决。论文根据相关文献设计了用于解决这一问题的近似算法，将禁忌搜索算法用到了这一问题中，取得了较好的优化效果；如果配送资源不受限制，总配送成本不受顾客的库存成本的影响，问题转化为带有时间窗的车辆路径问题(VRPTW)。论文将这一问题分解为两个阶段来设计求解，在第一阶段使用遗传算法，第二阶段使用快速有效的节约启发式算法。 最后，在单车配送方案优化管理系统的开发中，解决了几种模型中都涉及的核心问题，即带有软硬时问窗的配送方案的确定。在开发过程中，由于顾客信息的不确定性，完成了建立动态基础数据库。在配送方案优化过程中，充分考虑了顾客对软硬时间窗的要求，最终生成了车辆的最优配送方案一维显示与成本计算，能基本满足实际应用。更多还原

【Abstract】 Logistics trade has a newly important change, and the freight distribution is a important link with the rapid development of electronic business. In the course of freight distribution in a distribution center, the distribution plan optimization problem is a fascinating one. A reasonable distribution plan can bring extremely considerable economic benefit.In the base of analyzing the development process of distribution plan optimization in a distribution center, we research the algorithm design of VRP under different conditions in the thesis and set up the distribution plan optimization matrix including considering the single cycle in that inventory cost is not involved with soft and hard time windows and more cycles in that inventory cost is involved apart according to the customs’ desire to the distribution time.We analyze the method of setting up matrix and the solution in the different conditions. If the resource in the distribution center is finite, the custom inventory affects the distribution cost and the stochastic inventory allocation problem is the key to whole problem’s resolving. A approximate algorithm is designed in order to resolve this problem according to interrelated documents in this thesis in that the Tabu Search is used and a better optimization result is obtained; If the resource is infinite in the distribution center, the distribution cost is not affected by the custom inventory, the problem is a vehicle routing problem with time window (VRPTW). Then the problem is disintegrated into two parts that are solved respectively. Genetic algorithm is used in the first part and the saved heuristics algorithm is used in the second part.In the end, in the course of the management system development about single vehicle distribution plan optimization the key problem has been solved in the matrixes concerned that are the determination of the distribution plan with soft and hard time windows. In process of development, the dynamic basis database has been set up under the condition of the uncertainty about the custom’s information. After abundant consideration to the customs’ require to soft and hard time window, the vehicle optimized distribution plan has beenformed witch is a one-dimensional show and the total cost is computed at the same time. All of these can satisfy practical appliance.更多还原