【作者】 张赫；

【导师】 刘文煌；

【作者基本信息】 清华大学 ， 控制科学与工程， 2005， 硕士

【摘要】 随着制造业对成本降低需求的不断增长,物流已经成为企业提高竞争力所关注的焦点。第四方物流就是在这样的背景下产生的。第四方物流的研究目前还处于探索阶段,在其定义、概念、运作模式等多方面都还没有统一的认识和说法。本文通过充分综合第四方物流已有的研究成果,结合实际工程项目——“天悦”国际建材城的特点,将其定义为“方案集成商”的第四方物流供应商。本文提出了一个基于多Agent结构的第四方物流信息系统结构,并分析了其中各个代理的基本功能。由于第四方物流供应商在整个系统中处于核心关键的位置,因此,提出了一个第四方物流的决策支持系统框架,并分析了其中的关键问题——物流方案的设计与优化。对于这个关键问题,本文建立了“以城市为结点的多维多权有向图”和“以物流服务为结点的简单有向图的”模型,并结合前一种模型设计了第四方物流优化的全过程。由于客户需求(即约束)的存在,本文设计了一个“双遍历过程”来简化建立的物流网络,旨在减少下一步优化算法的计算量。对于物流方案的优化问题,本文将其描述为“在多约束条件下多个方案的搜索问题”。并设计了一个“基于A~*算法的剪枝策略”来解决该问题。虽然A~*算法从本质上并没有解决“指数爆炸”的问题,但是由于上一步的简化工作以及实际中3PL供应商的转换并不是很多,因而该算法在一定程度上仍然能保持较好的表现。优化方案中的最后一步是3PL供应商的评价与选择,层次分析法(AHP)和数据包络分析(DEA)是两种常用的供应商评价方法。本文结合这两种方法各自的优点,设计了AHP/DEA方法选择3PL供应商,并给出了一个简单的实例分析。论文的结尾分析了第四方物流的优化模块在具体实施中需要注意的一些问题,主要是物流信息的搜集与整理和优化模块的技术实现两个方面。最后结合研究中遇到的一些问题,提出了几点改进的意见。更多还原

【Abstract】 With the development of the demand for reducing cost in manufacturing,more and more enterprises have focused on their logistics. Fourth PartyLogistics (4PL) is presented under this background. There is no concept of4PL which can be admitted by most scholars. In this dissertation, presentresearch work is fully synthesized. Based on the practical project— “Tianyue”International Building Material Free City, this corporation is regarded as a4PL provider.With the knowledge of the information system, a framework based onmulti-agent is presented and each part of the system is analyzed for basicfunctions. Since the 4PL administrator functions as the core member, adecision support system is presented and the key problem— optimization ofthe logistics scheme, is analyzed.In order to model this problem, two models— “Multi-mension directedgraph using cities as its nodes” and “Simple directed graph using logsitcsservice as its nodes”, are presented and an optimization process is designedwith the first model. In this procedure, a method for network simplification isdesigned to reduce the complexity of the logistics graph in order to alleviatethe burden of the computation in the next step. The optimization problem canbe described as— “Searching the first K shortest Path with multi-constraints.”A~* algorithm is a method to search the target path in artificial intelligence.With this method, a procedure named A~* Prune is developed to search therequested K shortest paths. Although A~* algorithm can not solve the problemof “exponentially growing”, the burden of the calculation can be controlled bythe simplification step and there are not many exchanges in practical business.The last step is to evaluate and select 3PL vendors. AHP and DEA are twopopular methods applied to vendor selection. A method named AHP/DEA isdeveloped to solve the problem of 3PL selection and a simple case isanalyzed.In the end of this dissertation, some technical problems are discussed forrealizing the 4PL optimization procedure and some advice for improving theresearch work are also presented.更多还原