联系我们

论文查询结果	返回搜索

论文编号：	489
作者编号：	1120050759
上传时间：	2008/12/30 12:38:15
中文题目：	基于案例推理（CBR）的供应链过
英文题目：	A Systematic Approach Study on
指导老师：	严建援
中文关键字：	供应链;过程重组;基于案例推理;<
英文关键字：	Supply Chain;Process Reenginee
中文摘要：	随着供应链管理理念的深化，人们普遍认为，未来竞争焦点将不再发生在企业与企业之间，而在于供应链与供应链之间。无论是主动还是被动，企业融入供应链将不可避免。而业务过程又是供应链网络节点企业间的各种交易关系的直接纽带，因此，研究供应链过程重组问题便具有显著意义。不过，由于当前的社会、经济等因素已经远远超越了传统3C环境的挑战，因此研究不应仅限于重组动因及其关键成功因素的分析，而应努力探求有力的重组方法及工具支持。在丰富的企业实践案例客观存在的前提下，如果能够把它们组织成案例库并提供智能化访问机制，则将使复杂的重组工作事半功倍。 本文在业务重组与供应链管理基本理论支持下，开展了基于案例推理(CBR)的供应链过程重组系统方法研究，研究具体涉及供应链本体的构建，案例库的构建，案例的检索与知识学习，重组关键过程识别几个方面。其最终目的旨在为未来开发一个有足够应用性的供应链过程设计与优化决策支持系统提供重要的理论与技术框架。 本文的写作框架如下：第一章充分分析了选题背景，论证了该选题的必要性，并对研究动机、研究目的、研究方法及思路进行了阐述，同时给出文章框架。第二章通过相关文献回顾与评述，梳理了业务过程重组及供应链相关理论，分析了供应链过程重组中的常用方法、工具以及供应链的绩效认知方式，从而发现了研究空白并形成了本文研究点。第三章分析了供应链环境下业务过程重组决策的特征及传统业务过程重组决策支持方式的局限性，讨论了基于案例推理范式的特点及优势，提出了基于该范式的供应链过程重组框架。第四章阐述了本体的相关概念、功能与结构，提出了供应链知识本体的构建方法，给出了层次化的供应链本体模型。第五章讨论了供应链过程重组案例知识的获取问题，分析了案例内容的结构，给出了案例知识的形式化表示方法及案例库的组织方式。第六章给出了供应链过程重组案例的检索策略，研究了其中的案例相似性问题及特征项权重确定方法，讨论了相似度计算的两个特殊问题，分析了案例推理方法的学习机制。第七章开展了绩效原始数据获取系统架构的探索及供应链绩效评价决策支持系统结构的研究，同时在绩效驱动策略下，研究了供应链过程重组的关键过程识别。第八章以某饲料加工企业为例，将本研究提出的整体策略及相应工具应用于该企业的供应链过程重组之中，检验了基于案例推理方法在供应链过程重组中应用的可行性及有效性。第九章给出研究结论，指出研究局限，并提出未来进一步完善与改进的研究设想。 本研究做出了如下贡献：首先，在MIT《过程手册》相关研究启发下，借鉴SCOR模型思想，提出了一套基于CBR的供应链过程重组决策支持模型。该模型超越了传统过程参考模型过强的泛化性且使用中过多依赖用户专家经验的缺点，突破了现行纯文本管理类案例的非结构化缺陷导致的二义性及不易重用性，从方法上它还避免了最优化方法过高的计算复杂度，并且具有更符合人类思维模式特点的智能推理能力，从而可以更有效地辅助供应链过程重组。其次，为了改变传统CBR应用系统不易在语义级跨组织、跨系统充分共享核心术语的弊端，指出了其与供应链本体相结合的思路。在对重要知识本体概念化基础上，突破构建孤立领域本体的缺陷，提出了由企业通用本体、SCOR本体、行业本体及应用本体构成的继承性层次化供应链本体模型，一定程度上避免了本体的重复开发，提高了成熟本体的重用性。再次，针对案例库构建策略，提出了一种框架表示法与面向对象表示法相结合的混合案例表示法，其较之单一的案例表示方法更具灵活性；提出案例库纵向多级层次化组织与横向子域化处理相结合，它比单纯职能领域索引更便于案例的拆分、重构及后续检索，从而更好地满足了对复杂动态数据存储与访问的需求。第四，针对案例检索问题，在模板检索与最近邻检索相结合基础上，提出了一种适合供应链过程重组问题的两阶段检索策略，并专门探讨了管理类案例特征项赋权、时间序列类型特征项的相似性及案例时效性问题，此外，对CBR的学习机制中有关过程的相似性问题也提出了独特见解。这样，其功能便超越了简单的CBR分类器，体现了对管理问题复杂性的适应。 当然，本研究仍存在一定的局限性，未来可通过开发、集成成熟本体以及不断充实丰富案例库而使本研究得以持续推进。
英文摘要：	With the supply chain management getting wider and deeper, it is, generally thinking, the war between supply chains instead of that between firms that will be the competition focus in the future. The fact that any firm, positively or passively, becomes a supply chain member is inevitable. The business process in a supply chain network is a direct bridge to connect the relations between business transactions. Therefore, the academic study on Supply Chain Process Reengineering(SCPR) is of great significance. However, the social and economic elements are more challenging than what the traditional 3Cs environments ever bring, so we think research should not be restricted to critical success factors or why reengineering. Rather, we should pay more attention to powerful reengineering approaches and tools. If all the existing SCPR real case can be structured into casebase and intelligent access is provided, the hard work for SCPR will get twice the result with half the effort. Under the support of business process reengineering and supply chain management, our study on CBR-based SCPR is taken. It includes supply chain ontology construction, casebase construction, case retrieval, knowledge learning, and key process identification in detail. The purpose we pursue is to provide foundation for critical theory and technical framework for developing a supply chain process optimizing and reengineering DSS that is practical enough in the future. This thesis consists of the following contents: Chapter one in-depth analyzes the research background, argues the necessity and specifies the research motivation, goals, methodology, route, and the framework of the thesis. Chapter two, through literature review, discusses the related theory of BPR and SCM, the conventional methods and tools involved, and the supply chain performance cognitive style, which help finding the blind area that becomes our study spots. Chapter three analyzes the characters of the decision-making in SCPR and the limitation of traditional BPR decision making mode, discusses the characters and advantages of the CBR paradigm, and presents architecture of the CBR-based SCPR. Chapter four describes the concepts, functions and structure of the ontology, and presents the construction approach to supply chain ontology with a hierarchical supply chain ontology model followed. Chapter five discusses the issues about the SCPR knowledge acquirement, analyzes the case content structure, and presents the formal representation of case and the organizing format of the casebase. Chapter six presents the case retrieval strategy, argues the case similarity, feature weighting, and two special issues on similarity calculation in SCPR, and discusses the CBR learning mechanism. Chapter seven makes the exploration of the architecture of the original data collection system for performance evaluation, puts forward a study of the corresponding DSS structure, followed by a research on the approach to identify key process in SCPR under the performance-driven policy. Chapter eight verifies the feasibility and effectiveness through applying the whole solution and tools to the SCPR in a feed company which is shown as an example. Chapter nine specifies the research conclusions and limitations, and presents the prospects for further studies. The main contributions are summarized as follows: (1) Enlightened by MIT’s Process Handbook and the thinking from SCOR, a set of CBR-based SCPR decision-making support model are presented. These models not only surpass the defects that the traditional process reference models is over-generalized and dependent too much on expertise, break the ambiguity and not-easy-to-reuse character resulting from the flaws of non-structural of the pure-text case related to management, but also avoid the high calculation complexity of the optimization method and have the intelligent reasoning ability that is more similar to the mankind thinking style, so as to assist SCPR more effectively. (2) In order to change the fact that the traditional CBR applications are not easy to share core terms across organization and system boundaries in a semantic level, an idea to combine the SCPR with supply chain ontology is provided. Based on the conceptualization of key knowledge anthologies, in order to overcome the shortcomings of establishing isolated domain ontology, this thesis gives an inheritable and hierarchical supply chain ontology model made up of enterprise general ontology, SCOR ontology, industry ontology and application ontology. It avoids the re-development of ontology in some extent and improves the re-usability of the mature ontology. (3) For the casebase construction, a frame and object-oriented mixed case representation is provided. The mixed representation is more flexible than the single representation. A policy with the combination of casebase vertical hierarchization and horizontal decomposition is also presented. It is more convenient to take apart, re-combine and the following retrieve casebase than that of a pure function domain index, which satisfies better the requirements to store and access the complicated dynamic data. (4) For the case retrieval, under the integration of template-based approach and NN-based approach, a two-stage retrieval strategy suitable for SCPR issues is illustrated. Followed by a particular view on the process similarity in CBR learning mechanism, this study discusses case feature weighting, time-series type feature similarity calculation, and the time-effect of the case. Therefore, the function of the whole system is superior to a simple CBR classifier, which demonstrates its adaptability for management issues. Also, this study has its limitations, however. In the future, we can improve it by developing or integrating more mature ontology and fill the casebase with more powerful and diverse cases when possible.
查看全文：	预览  下载(下载需要进行登录)