【作者】 江盛树；

【导师】 李平； 杨春节；

【作者基本信息】 浙江大学 ， 控制理论与控制工程， 2004， 硕士

【摘要】 在面对日益激烈的竞争和复杂的动态市场环境下，当今企业为了提高决策水平和市场应变能力，在实施企业内部ERP系统外，采用供需链管理技术也是重要的手段。流程企业是指产品生产过程为连续的或者是成批的工业，在ERP和供需链方面与离散企业相比具有不同的特点。本论文以制浆造纸流程企业供需链管理为对象，采用了MAS和神经网络等人工智能技术，讨论了流程企业供需链中的建模、预测结构、预测仿真等问题，介绍了课题组开发的供需链管理原型系统，论文的章节内容安排如下： 第一章是序论，综述了现代企业信息化的若干基本概念，包括企业ERP、供需链管理，比较了离散企业和流程企业的不同特点，提出了流程企业供需链对预测的功能需求； 第二章概述了建模的一般原理和供需链中建模的技术方法、关键因素和几种模型分类，并基于供需链的SCOR参考模型，建立了造纸企业供需链的网络模型、过程模型和信息模型； 第三章介绍了Agent和Multi-Agent System的概念和特征，提出了流程企业供需链预测中战略层、协调层和操作层递阶结构，定义了预测结构中的不同类型Agent，构建了Agent的协调消息、协调规则和处理流程； 第四章将现有的若干预测模型和技术分成强时间序列和弱时间序列两类模型进行介绍，在造纸企业供需链应用背景中对几种方法进行了比较，采用神经网络技术对某造纸企业的产品销售和木材采购数量进行了实例仿真； 第五章介绍了课题组开发的流程企业供需链原型系统，首先概述了一般的软件开发过程和相关建模和设计的技术，介绍了原型系统的建模过程、技术架构、功能结构和对象类的设计，重点介绍了该系统中的预测功能体系结构和预测算法实现，并在造纸企业进行了应用。 第六章是最后一章，对论文工作进行了总结，并指出了本课题下一步的研究工作和若干问题。更多还原

【Abstract】 Facing the increasingly competent and more dynamic market environment, nowadays the enterprises are aim to improve the level of decision and flexibility due to market demand fluctuate. Based on implementation of ERP System within factory, SCM is an important technology. Process enterprise, in which the process is continuous or batch, is different from the discrete enterprise in ERP and SCM. In our research, the modeling process, forecasting architecture and forecasting simulation, using Multi-Agent System and Artificial Network technology, are discussed. Also, the supply chain management prototype system in process enterprise is present. The following is about the content arrangement of each chapter:Chapter one summarizes several basic concepts in contempt enterprise information process, including ERP and SCM. After the concept and features are compared between process and discrete enterprise, the function requirements of forecasting for supply chain in process enterprise are proposed.Chapter two firstly introduced general modeling theory and some modeling technology, key elements and several model types. According to the SCOR reference model, the network model, process model and information model are present in pulp & paper enterprise.Chapter three is the important part of the thesis, in this chapter,’ we apply MAS technology into supply chain modeling in process enterprise. The forecasting hierarchical architecture composite of strategy, coordinate and operate layer are proposed and accordingly each category of function agent are defined, furthermore, the coordinate messages, rules and flows are present.In chapter four, the forecasting technologies are introduced as two types. We compared the application vision between some methods in pulp & paper supply chain and apply neural network method for data simulation of paper products and wood procurement quantities.Chapter five is group project system developing bulaletin in which a supply chainmanagement prototype system is introduced. The general software developing process and technology are summarized firstly, then the modeling process, technology architecture, function module and object-oriented class designing, especially the forecasting framework and algorithm implementation are introduced. There is a application instance in one pulp & paper enterprise.The last chapter summarized the whole research work and presented the work need to be done in the future.更多还原