【作者】 孙丽；

【导师】 王秀伦；

【作者基本信息】 大连交通大学 ， 机械制造及自动化， 2005， 博士

【摘要】 随着制造业向信息化、智能化、网络化、数字化的方向发展,制造企业中知识管理的地位与作用日益显著。对于制造企业而言,工艺知识贯穿在产品的规划、设计、制造、装配、维修等整个生命周期当中,对于企业的生存发展起着至关重要的作用。工艺知识管理已经成为企业一项基础性、关键性的工作,也是企业实施CIMS、CE、LP、AM、ERP等先进技术和管理手段的前提。 整个论文工作以企业的工艺知识管理为主线,深入系统的研究了网络环境下工艺知识管理系统的构建及其若干关键技术。本文工作的主要内容可以分为如下几个部分: 1.按照产品全生命周期所需的工艺知识,将工艺知识用面分类法进行分类编码,构造一套完整的工艺知识分类编码体系。 工艺知识纷繁复杂,涉及范围广泛,同时又是多学科的交叉,所以用传统意义上的线分类法存在一定的困难。本文作者在研究中首先致力于将工艺知识进行规范性的面分类,然后在每一层面上再采用线分类法,在线分类法中适当地填加包容项,使得所有的工艺知识都被包容且很少交叉,有效地解决了由于分类不明确所造成的不便管理的问题。同时,由于采用了四个码位的柔性编码技术,使得工艺知识进一步规范化和标准化,为后面的工艺知识表达、存储、挖掘创造了良好的条件,同时也为企业进一步实施信息化和数字化奠定了基础。 2.研究了数据仓库模式下的工艺知识信息模型和存储方式。 构造了一个企业级工艺知识数据仓库,该仓库采用雪花模型与星型模型相结合的方式,并按照部门需求创建了一定数量的数据集市,以备各部门进行数据分析和数据挖掘。数据仓库的存储采用集中式和分布式相结合的方式,将频繁被各种用户访问的大量知识存放在服务器上,而将仅被部门级个别用户访问的知识存储在客户端。该方法能大大提高检索速度并节省存储空间。 3.研究了工艺知识的推理和快速导引技术。 知识库与知识推理是研究知识管理必不可少的一项内容。本文主要研究了原型法机理及基于原型的工艺知识推理机制,为CAPP系统中设计工艺过程提供了一种新方法。另外,针对且前CAPP系统中普遍缺少辅助选择设备、工艺装备等功能的情况,研究子基于关系数据库的设备、工艺装备选择方法。针对工艺知识中涉及到许多工作流程管理的问题,研究了Petri网模型及其推理机制在时间工作流中的应用。 4.研究了数据挖掘技术在工艺知识管理中的应用。 有很多存在于大量数据中的工艺知识,它们是潜在的、隐含的,只有采用一定的数据挖掘技术才能使其显性化。本文采用灰色关联度分析法和灰色模型预测方法对某型号柴油机汽缸套磨损问题进行挖掘,得出了柴油机汽缸套磨损问题的数学模型和预测模型,对于柴油机的使用、维护和售后服务具有重要的指导意义。 5.提出了工艺知识管理水平评价指标体系。更多还原

【Abstract】 With the development of informationization, intelligentization, networkization and digitalization in manufacturing enterprises, the role of knowledge management (KM) in manufacturing enterprises are becoming more and more important. Process knowledge (PK) exists through the full life cycle of products, such as product planning, design, manufacture, assemble and service and it decides the living and developing of enterprises. Process knowledge management (PKM) is fundamental and key work in enterprises and is the premise of CIMS, CE, LP, AM and ERP etc.The thesis is centered on PKM of enterprises and deeply researches the framework of PKM system in network environment and its key techniques. It includes the following parts:1. PK is classified and coded using surface classification method and a PK classification and coding system is constructed according to full life PK of product.PK is very complicated and abroad, and at the same time it is the intersection of multi-subject, so it is difficult using traditional line classification method. In this thesis, PK is classified using traditional surface classified method firstly. Then line classification method is employed in each tier. Supplement is used in line classification method. Therefore, all PK is included without intercrossing. Meanwhile, four-part flexible coding technology is employed, which makes PK standardization and creates good conditions for future representation, storage and knowledge mining. So it does for the informationization of enterprises.2. PK information model and storage technology are researched under data-warehouse environment.An enterprise level PK data-warehouse is constructed based on the above two work. The data-warehouse employees the combination of snow model and star model. Some data marts are framed as required for data analysis and data mining. The storage of data-warehouse is the combination of integration and distribution. The great amount of knowledge that is often visited frequently is located in server. Others are located in clients. This method can increase retrieval speed greatly and save storage space.3. PK reasoning and quick inducting technology is researched.Knowledge base and knowledge reasoning is necessary work in studying KM. In this thesis, prototype theory and prototype based reasoning is researched. It provides a new method for CAPP in process planning. In addition, relational database based machine selection and rig selection technology is studied. At the same time, Petri net model is employed in modeling time workflow problem.4. The application of data mining in PKM is researched.There may be a great deal of knowledge existing in the data sea. They are latent and connotative. Only data mining can make them illustrious. In this thesis, grey association analysismethod and GM forecasting is used to mine abrasion problem of cylinder sheath of diesel engine. It is very important to guide the using, maintenance and after-sale service of diesel engine.5. PKM level evaluation indexes are put forward.As PKM is in underway state, many enterprises do not know how to set out. They do not know what level their enterprise in. A practical PKM evaluation indexes is put forward after great amount of investigation. It can evaluate enterprise in time and point out its developing direction, hi addition, fussy evaluation method and neural network method are combined together to make evaluation result more scientific and reasonable. As each enterprise has its own instance, the indexes can be deleted and added. Entropy weight concept is employed to decide indexes.6. A web-based framework of PKM is constructed.A web-based framework of PKM is constructed employing above research results. The system is developed using Asp.net and Sql-server. XML technology is used to communicate with other knowledge system. The system has such functions as PK collecting, PK communicating, expert consulting, PK inquiry, PK inducting, PK reasoning, manufacture site management, work-procedure quality management, machine management, after-sale service and so on. It also can evaluate the level of PKM and helps enterprise make it clear and locate its developing direction.更多还原