【作者】 吴翀；

【导师】 罗新星；

【作者基本信息】 中南大学 ， 管理科学与工程， 2010， 博士

【摘要】 敏捷供应链应具有高度的灵活性,能够快速的构建与分解,以应对快速变化的市场环境与客户需求。因此,随着供应链对自身敏捷性的要求不断提高,伙伴选择决策的决策频度在不断增加,伙伴选择决策的重要性更加显现。然而,敏捷供应链伙伴选择在供应链构建与重构过程中存在众多不确定性和模糊性,使得该问题的解决更加的复杂和困难。论文从动态的视角对敏捷供应链的伙伴选择问题进行了深入研究,提出了敏捷供应链伙伴选择四阶段动态反馈概念模型。四阶段包括：伙伴选择准备阶段,伙伴选择初选阶段,伙伴选择精选阶段和伙伴选择应用反馈与持续改进阶段。该概念模型以系统的方法为指导,针对敏捷供应链伙伴选择不同阶段的不同需求与特点,综合集成了多个定性与定量的决策方法和技术,提出了四个相应的子模型,以期准确高效地解决这个日益重要的问题。在敏捷供应链伙伴选择准备阶段,已有的研究发现：应用Dempster-Shafer信任函数理论可以在资源约束的条件下,解决敏捷供应链伙伴选择评价指标体系的构建问题。在此基础上,论文应用Dempster-Shafer信任函数理论提出了敏捷供应链伙伴选择评价指标体系构建模型。该模型既简化了已有模型的变量分类,又保持了模型评价的全面性；既全面客观地描述了决策过程,又不失其应用的可操作性。在敏捷供应链伙伴选择初选阶段,论文基于径向基网络技术和Kraljic经典伙伴分类理论,提出了敏捷供应链伙伴选择径向基网络初选模型。该模型能够帮助决策者解决敏捷供应链伙伴选择早先阶段,复杂的大规模数据分析处理问题。通过对该子模型的应用,组织和决策者可依据概念模型中选择准备阶段得出的评价指标体系,对潜在伙伴进行初步分类筛选。在敏捷供应链伙伴选择精选阶段,论文提出网络分析与混合整数多目标规划模型,以解决敏捷供应链伙伴选择的精选问题。敏捷供应链的一个关键要求是供应链上的各组成部分,如供应商、生产商和分销商等,能够迅速、高效地对市场变化做出响应,并按照最优的结构组织起来,将最优数量和质量的产品和(或)服务,投放到正确的市场区域中去。因此,在网络分析与混合整数多目标规划模型中,第一步是应用网络分析法,构建网络分析子模型以获取不同精选评价指标的相对权重。采用此权重,第二步应用混合整数多目标规划决策技术,构建最优敏捷供应链组织结构,并确定最优需求供给关系,以达到整条供应链的多目标最优。在敏捷供应链伙伴选择应用反馈与持续改进阶段,论文构建敏捷供应链伙伴选择应用反馈与持续改进模型。该模型应用Deming持续改进的理论方法和组织学习的理论,试图抽象概括敏捷供应链伙伴选择过程中,阶段间反馈提高的方式和方法。该模型帮助组织和决策者优化决策过程,提高决策质量,最终达到提高和改善敏捷供应链伙伴选择决策效率和效果的目标。理论模型构建完成后,论文还采用实际商业数据和经典算例,实验仿真了敏捷供应链伙伴选择的决策过程。同时论文还开发了模型的应用反馈量表,通过对量表收集到的数据进行的详尽分析,发现决策者在应用论文中所提出的模型辅助其决策的过程中,能够有效提高决策的效率和效果。更多还原

【Abstract】 Today’s more dynamic business environment increases the need for greater agility in supply chains, which increases both the importance and frequency of partner selection decision-making. Agile supply chains (ASCs) need to be highly flexible in order to reconfigure quickly in response to changes in their environment. The use of ASCs has become more common in today’s increasingly dynamic markets. However, partner selection decisions are inherently more complex and difficult under the conditions of uncertainty and ambiguity created as supply chains form and re-form. This study presents a four-phase conceptual model for supply partner selection in ASCs, which comprise Partner selection preparation, Pre-classification, Final selection and Application feedback. It draws on a range of quantitative and qualitative techniques. Building on existing literature the model draws on both quantitative and qualitative techniques in its four phases to present a comprehensive and systematic approach to this increasingly important task. These include application of the Dempster-Shafer and optimisation theories, radial basis function artificial neural networks (RBF-ANN), analytic network process-mixed integer multi-objective programming (ANP-MIMOP), Kraljic’s supplier classification matrix and the principles of continuous improvement. The resulting model offers a comprehensive and systematic approach to tackling this increasingly important task.During the Partner selection preparation phase, previous research has suggested that the application of the Dempster-Shafer and optimisation theories offers a way of solving this problem under conditions of resource constraints. This study advances this approach by offering a simplified yet thorough, rigorous yet still practical method for formulating criteria to use in partner selection decision-making in agile supply chains.During the Pre-classification phase, this study develops a model that helps overcome the information processing difficulties inherent in screening a large number of potential suppliers in the early stages of the selection process. Based on RBF-ANN, the model enables potential suppliers to be assessed against multiple criteria using both quantitative and qualitative measures.During the Final selection phase, this study proposes a two-stage approach, based on the application of ANP-MIMOP model, to solve the final selection problem of partner selection in ASCs. A key requirement of an ASC is that its constituents (suppliers, producers, distributors, etc.) can combine and react to fast changing customer demand as efficiently and effectively as possible. An ASC needs to adopt the most appropriate supply chain structure and assign the most suitable order quantities to the most appropriate supply partners in any given circumstance. In the first stage, an ANP methodology is applied to calculate the priorities of different criteria for partner selection. Secondly, using these priorities, a MIMOP method is used to determine the supply chain structure and optimize the allocation of order quantities.During the Application feedback phase, this study presents and tests a model designed to provide feedback and continuous improvement during the process of partner selection process in ASCs. The model seeks to capitalise on the increased application of the supplier section process by applying principles of continuous improvement and organisational learning. Its aim is to support organisational decision-makers in their efforts to optimise the performance of the supply chain by ensuring that the most appropriate suppliers are selected at all times.Empirical illustrative examples are used to demonstrate the models and approaches, obtain insights into its application. The results of a questionnaire, used to assess the efficacy of the model, showed that the participants found the model was likely to have significant benefits when used in practice.更多还原