【作者】 李琳；

【导师】 唐加福； 刘士新；

【作者基本信息】 东北大学 ， 系统工程， 2010， 博士

【摘要】 随着网络通信和信息技术的飞速发展,电子商务作为一种新型的商业运作模式正逐渐渗透到人们生活的各个方面。它从多方面动摇了传统企业经营和营销理论的根基,也改变了企业外部经营和购销的实务操作模式。物流配送是实现电子商务的关键环节,它决定着电子商务的发展水平。因此,研究电子商务环境下的企业物流配送系统及其管理对于企业的成败具有十分重要的意义。电子商务环境下的物流配送具有信息化、网络化、自动化、智能化、柔性化等特点。本文针对电子商务环境下物流配送的特点,对物流配送中的关键优化问题,即配送路径选择问题和配送中心选址问题进行了研究,具体内容包括以下五个部分：(1)查阅了大量的相关文献,综述了物流配送中车辆调度、路径优化、配送中心选址等优化问题的研究现状,探讨了物流配送在电子商务中的地位及作用,总结了电子商务环境下订单配送问题的新特点。(2)针对电子商务环境下订单配送问题的特点,建立了单阶段及多阶段带信息流的车辆路径问题的数学模型,设计了两阶段启发式求解算法。第一阶段用禁忌搜索算法求解计划日当天已到交货期的订单的配送问题；第二阶段考虑了订单的未来信息,用基于评价值的贪婪插入算法将未到交货期的订单与已有的配送方案进行整合配送。该方法能够提高车辆的使用率及配送企业的配送效率。(3)针对电子商务环境下带时间窗车辆路径问题的特点,分别建立了单目标及多目标配送问题的数学模型。对单目标问题研究了软时间窗及硬时间窗配送问题,用改进的蚁群算法进行求解。多目标问题的目标为最小化车辆使用数目、车辆行驶距离之和、车辆的平均等待时间及最大化顾客平均满意度。各目标具有相同的重要性,用近似Pareto最优解集来指导蚁群的信息素更新策略。该方法可使顾客具有较高的满意度、车辆等待时间较少,从而使配送企业具有较好的声誉及较高的配送效率。(4)针对电子商务环境下回程带货车辆路径问题的特点,分别讨论了静态及动态情形下回程带货问题的数学模型,设计了自适应记忆算法进行求解。对静态问题研究了无优先约束的混合回程带货问题。对于动态问题,首先,在每个工作日的开始制定配送中心当天的静态配送路线；然后,分时段检测新出现的取货需求,并将它们插入到现有的车辆行驶路线中的最佳位置(增加距离最少的位置),从而达到根据顾客需求及车辆行驶状况及时动态调整车辆行驶路线、满足顾客要求的目的。(5)以物流配送中配送中心选址问题研究的产生、发展及深入为主线,讨论了配送中心选址问题的基本模型,揭示了深入研究该问题的过程,探讨了电子商务环境下配送中心选址模型的建立。在配送中心选址问题已有基本模型的基础上,分析了其统一模型建立的可能性,将许多选址问题的基本数学模型统一成一个模型,为电子商务环境下配送中心选址问题优化系统的开发提供了数学基础。更多还原

【Abstract】 With quickly development of networking and information technology, Electronic Commerce, which is a new type of business operation, is penetrating every aspects of our life. It fluctuates the base of traditional business operations and marketing theory from many aspects. It also changes practice operation mode of enterprises’external management and trade. Logistics distribution is the key link of carry out Electronic Commerce, and it decides Electronic Commerce’s development level. Research about enterprises’distribution system and their management under Electronic Commerce is significant for enterprises’ success.Logistics distribution under Electronic Commerce has characteristics such as information, networking, automation, intelligence, flexibility, etc. Considering theses characteristics, this paper researches some key optimal problems in logistics distribution, that is, selection of distribution routes and distribution center location problem. The major work focuses on the following five sections:(1) The paper surveys a large number of correlative literatures, and reviews current researches about optimization problems about vehicle scheduling, routes optimization, and distribution center location problem. It discusses status and roles about logistics distribution in Electronic Commerce. It summarizes new characteristics about order delivery problem under Electronic Commerce.(2) Considering characteristics of order delivery problem under Electronic Commerce, this paper constructs single stage and multi-stage mathematical models about vehicle routing problem with information flow, and designs a two-stage heuristics algorithm to solve them. In the first stage, it uses Tabu Search algorithm to solve order delivery problem about orders that have to be delivered at the planning day, that is, basic delivery problem. In the second stage, it considers future information about orders and uses greedy insert algorithm based on evaluation to solve the problem. By using this method, it can delivery part of orders that not arrive the given due date with basic delivery plan and the given method could improve the ratio of usefulness about vehicles and delivery enterprise’s efficiency.(3) According to characteristics of VRPTW under Electronic Commerce, the paper constructs one objective and multi-objective mathematical models about delivery problem. For one objective problem, it researches soft time windows and hard time windows delivery problems, and uses improved Ant Colony algorithm to solve them. For multi-objective problem, objectives are minimization numbers of used vehicles, minimization sum of all vehicles’traveling distance, minimization vehicles’average waiting time, and maximization customers’average satisfaction. Each objective has the same importance. It uses similar Pareto optimal solution set to direct pheromone-updating tactics. This method can let customers have higher satisfaction, reduce vehicles’waiting time, and make delivery enterprises have better fame and higher delivery efficiency.(4) Considering characteristics of backhaul problems under Electronic Commerce, the paper discusses static and dynamic mathematical models about backhaul problem, and designs Adaptive Memory Programming algorithm to solve them. For static problem, it researches mixed vehicle routing problem with backhauls. For dynamic problem, firstly, it decides static delivery route at the beginning of each working day. Secondly, it detects the appearance of new pick-up demand at the end of each time division, and inserts them into the best position of current routes (the position where added distance is minimization). By using this method, it can dynamic adjust vehicle traveling routes according to customers’ demand and vehicles’traveling state, and it can satisfy customers’requirement.(5) The paper discusses the appearance, development of researches about distribution center location problem in logistic distribution. It reviews basic models about distribution center location problem, and reveals the process of intensive study about this aspect. It also discusses construction of mathematical model about distribution center location problem under Electronic Commerce. Based on the existing basic models, it analyzes feasibility of constructing a unitive model. It unifies a lot of basic mathematical models about location problem into one model and this work provides mathematical foundation for the development of distribution center location optimal system under Electronic Commerce.更多还原