【作者】 吴洁；

【导师】 彭其渊；

【作者基本信息】 西南交通大学 ， 管理科学与工程， 2008， 博士

【摘要】 第三方物流服务项目中普遍存在着“效益背反”现象。它已经成为阻碍企业有效提升管理效益和企业可持续发展的管理陷阱。为此,论文以项目管理模式下的第三方物流服务项目多目标综合优化为宗旨,从时间—质量—成本(TQC)三要素的相互背反关系分析入手来研究服务项目的综合优化,分别从时间性和空间性上拓展了现有研究的纵横度。本文的主要工作包括以下几个方面:在对国内外相关研究现状的深入评述基础上,分别阐释了项目管理模式与第三方物流服务项目的一般规律,及两者相结合的特殊性。通过系统的分析得知,物流服务项目综合优化问题呈现出项目管理范畴内常见的多目标性、层次性和优先性的“主线”特征。这表明论文将物流服务项目纳入项目管理模式下进行研究是合理的。进而,本文抽象出具有代表性的时间—质量—成本三类变量,并研究其表现出的此消彼涨关系,以描述物流服务系统所特有的多目标效益背反现象。同时,结合企业分层次管理的实际和项目管理分阶段处理问题的思路,探求这一多目标优化问题的解。再则,基于首层决策环境相对于次层的模糊性和决策粒度相对偏粗等实际情况,将首层研究任务设定为推导多目标的优先级顺序,而将决策环境相对确定的次层的研究任务设定为解决多目标优化的数值问题。从而,确定了解决问题的空间复杂性基础框架模型。笔者在深入企业作调查研究时发现,决策者在战略管理层(即多项目层)决定企业的资源分配时,是按照物流服务项目的TQC的优先级,或者是要素的权重顺序来制定多目标决策的。因此,基于历史数据,客观地推导各个背反要素的排序即成为首层多目标优化的主要任务。论文根据粗糙集与条件熵原理来确定属性重要度,由此改进了经典的层次分析法中主观性强的不足之处,并基于可获得的数据驱动机制,确定了可信度较高的优先级顺序。在此排序的指导下不仅确定了主导目标,还确定了具有相对灵活度的其它次级目标,使得操作层有可能利用经典的多目标优化模型中“标杆学习法”进行项目过程管理。通过线性规划模型的求解,论文列举了TQC在可能的排序状况下各要素的优化值,并以此作为决策者操作的标杆值,在此范畴内根据要素重要程度来调整物流服务,使其达到项目最终要求。这一来既扩大了以往处理效益背反问题仅仅局限在某一层面的研究范围,又拓展了研究的层次化深度。鉴于项目管理和物流服务分别属于不同的管理范畴,目前尚未见到将两者较好结合的理论方法研究。故项目管理模式的“行业可移植性”是决定项目管理方法能否被物流企业所利用、能否促进此类企业可持续发展的关键所在。为了达到有效项目管理,促进企业发展的目的,论文将研究视角提升至具有时间性的项目评价管理决策规则的推导上来,由此给出了一个从过去的项目操作数据中挖掘决策支持客观依据的技术方法。主要工作是构建出实现上述目标的项目评价技术手段。这样论文便自然地引进了企业发展的时间维,对所研究问题进行了拓展,为项目管理模式的可移植性做出了一个关键性的突破,或者说是隐性知识转化为显性可编码知识在技术实现方式上给出了一个有益的探索,在一定程度上解决了该问题的时间复杂性。论文探讨方向是基于实际应用的理论研究,从实践中来,到实践中去,目的是通过更简便的方法提高决策效率。对于这一问题的解决思路是根据粒度原理,按照分层递阶的思路,并结合粗糙集理论,分解细化决策属性来分别评价项目,从而推导具有不同知识粒度即不同层次的多目标决策规则。由于决策信息表的指标群是由企业多目标优化的上下层共同构成的,该指标群的设计充分考虑了质量等属性粒度的细化指标及其可获得性数据源情况,因此它保证了物流服务项目管理中层次间扁平化的客观要求,同时还推导出通常容易被忽视的上下层之间的沟通信息,进一步将其归纳为约简后的属性,以指导项目管理模式下物流服务的多目标综合优化管理。这样不仅消除了上述组合算法中产生的大量冗余属性,使得这两个系列的算法集成为一个统一整体,并且还为企业的结构化沟通要点提供了决策支持,从而达成系统化物流管理的目的。最后,论文总结了全文的研究成果,就项目管理的两个发展方向,即技术实现手段和应用领域的拓展等若干创新点作出了总结,并对未来的研究工作作出展望。更多还原

【Abstract】 The phenomenon of trade-off stands out in service project of the Third Part Logistics (TPL). It has become a management trap, which embarrasses the enterprise’s development. Therefore, this paper presents a method about the multi-objective integrated optimization in TPL’s service project. Being the basic trade-off factors, the Time - Quality - Cost (TQC) of logistics service is taken into account. By the reason of the fact that this optimization has the spatiotemporal complex, this paper tries to extend the research from spatiality to temporality regulation. The main works include the following aspects.With the foundation of research about correlation between domestic and overseas studies, this paper explains the general rule of Project Management (PM) as well as the service project in TPL. This paper focuses on combinatorial innovation of them. Under the system analysis, this integrated optimization presents the characteristics of multi-objective, multi-layer, and relative priority. We can see that these three characteristics are the frame of this paper, which are familiar to us in the category of PM. That is to say it is reasonable by using PM method to study logistics service project. Moreover this paper presents the representative variabled of TQC, which are increasion or decreasion among them in logistics system. The multi-objective trade-off phenomenon is expressed. At the same time, this study is combined with PM idea of step-by-step. It searches out the result of multi-objective optimization. Furthermore, it is the fact that upper manage layer’s decision environment is more uncertain than lower manage layer’s. The task of upper layer should be supposed to deduce priority rank of objectives, because of relatively uncertain domain. While, the task of lower layer should be supposed to get the value of multi-objective optimization. On account of hiberarchy, this paper constructes the frame model about solution of spatial complex.According to deeply researches on TPLs, the auther found out that priority or weight of objective is the key to decided resources distribution by the decision maker of strategic layer (multi-project layer). Thus the main task of the upper layer is how to objectively deduce the rank of trade-off objectives based upon the history data. Significance of the attribute is confirmed by the theory of condition entropy. According to this method, this paper improves on the classical Analytic Hierarchy Process (AHP), which has the shortage of subjectivity. Thus it acquires the frame work based on data driven. The rank of priorities is reliability. Accorging to this, the objectives of primary and secondary are made sure. The classic method of benchmark can be applied in operation layer. This paper gives possible optimization value of each factor objective by linear programming. The decision maker deals with logistics service by significancy of factors in order to reach at the project’s requirement. In this method, this paper extends the study about trade-off problem from one layer to more layers. The scope of this study enlarges its views to space.Whereas PM and logistics service management belong to different category, there is few studies combined with both of them at present. The key of PM model application is the transplantation in industries, because the target of application PM is promoting TPLs’ business. In order to achieve this aim, this paper’s study takes dynamic viwe of decision rules’s deduced from evaluation PM. Thereby this method makes use of the technic of Data Mining (DM) based on history logistic service project data. It provides an objectively method to support of decision. This part of study is to constructe the frame work of carrying out PM evaluation. Naturally, this method focuses on the time dimension, which combines with TPLs’ development. Vertically spreading out the study, this paper breaks through the transplantation of PM. Hidden knowledge translates to obvious knowledge as a result of being coded. The problem of temporal complex had been settled down by some extents.The orientation of this paper is to study application theory, which comes from practise and goes to practise. The aim is to improve the efficiency of decision by simple ways. The route of solution is applying Granularity theory, which combines with Rough Set theory to propose multil-layer in the proper order thought. Theoretically the finer the decision attribute value of a decision table is, the lower the information granularity is. This paper provides a method of different decision attributes granularity according to different manage layer’s multi-objective decision rules. From this analysis, evaluation project should be made by different fine decision attribute. Because the indexes group is made up by both the upper and the lower layers objectives’ direction in TPLs, the indexes group must consider both information granularity and acquired data. Thus this method objectively ensures that logistics service project management’s requirement need flat to reduce management layer. Moreover it can deduce communication information between the upper and the lower layer, which is usually neglected at practise. Furthermore this information can be collected from the reduce attributs. The priorities of attributs can be solved by the reduction on PM in TPLs. The decision support will be provided by this reduction method, under the situation of structured communication for coordination between the upper layer and the lower’s.Finally, this paper sums up the whole result of study. Moreover it proposes some creative thinking in both technic implement and application domain. They are the main developing aspects. Furthermore it puts forward prospect for future researching work.更多还原