【作者】 孙毅；

【导师】 谭建荣；

【作者基本信息】 浙江大学 ， 机械设计及理论， 2007， 博士

【摘要】 产品全生命周期的质量控制是企业参与市场竞争，提升产品生存与发展空间的重要基础，随着客户敏捷化与个性化需求日益增强的发展趋势，产品质量过程控制需要充分地利用产品形成各阶段的过程性质量数据，并加以智能化地应用，各种具有智能性、集成化的质量控制信息系统研究得到了人们的广泛重视，其相关技术正成为当前先进制造技术与质量保障体系的热点课题。本文针对产品过程质量数据的有效组织、处理与利用，系统地研究了以产品过程质量控制特征要素为对象，关联产品结构特征进行数据挖掘和知识发现的理论与方法。将质量数据与产品零部件结构有机结合，提出了类质量树(GBOM-QT，Generic Bill-Of-Materials for Quality Tree)的产品质量控制信息模型，系统地研究了类质量树的构建、演化与过程质量信息的数据挖掘方法以及相关知识重用等理论问题，并结合国家自然科学基金项目“设计知识演化原理、方法及应用研究”、浙江省科技计划项目“基于DM的质量控制与分析系统”、浙江省自然科学基金项目“基于案例与知识发现的MC产品信息挖掘与管理技术”等项目研究，将上述理论应用于企业的实际产品开发与生产过程中，取得了较好的效果，证明了模型及算法的正确性和先进性。全文的主要内容包括：第一章主要介绍了产品质量控制与信息管理技术的理论与方法。回顾了产品质量控制技术的发展历程和研究现状，重点说明了基于知识重用的质量控制组织模型、信息流以及质量控制建模技术的体系结构、QFD建模方法及其改进技术、制造过程的质量控制与分析模型等研究内容，还介绍了服务于质量控制的知识组织、知识发现与重用技术，讨论了利用数据挖掘技术发现质量控制的各因素关联关系与知识重用等问题的思路，给出了本文的研究背景、意义和主要工作内容。第二章从研究产品功能质量目标的实现方法出发，对质量过程控制知识的多态性、过程性、层次性和结构性等内容进行了深入研究，就质量规划、形成过程、检测、缺陷分析与反馈等方面归纳质量分类特征，给出了产品质量的结构特征、技术特征、检测特征、缺陷分析与反馈特征等知识特征的具体内涵。在设计质量、制造与装配质量等方面研究了质量控制知识的获取技术与方法。第三章通过产品功能质量与产品结构关联分析，构建具有结构关联映射的产品质量树(BOM-QT)数据模型，以及具有通用类特性的产品类质量树(GBOM-QT)数据模型，深入研究了建模原则、建模方法及形式化描述方法等内容。给出了质量树数据模型中描述层次结构的节点对象所内蕴的属性、约束与规则等内置信息的有效描述与维护方法，建立了基于多特征质量特性的数据结构及其组件对象演化方法，表达蕴含节点对象的显式结构知识与内置属性规则与约束的隐式知识。最后给出了具有扩展数据结构功能的通用型类质量数据模型与质量树模型的XML模式数据描述方法。第四章从类质量树表示的产品质量控制模型出发，讨论了节点对象BOM-QT的内置约束属性的数据挖掘方法。着重从BOM-QT的拓朴结构出发讨论了BOM-QT树的节点编辑操作方法和无序树进行有序化处理方法，在此基础上提出了BOM-QT树簇的相似度计算方法，结合BOM-QT的聚类处理，研究了BOM-QT对象间的最小异构度和加权异构度的计算方法，给出了GBOM-QT的具体归并算法，及其GBOM-QT转换成XML模式文件的相关约束条件等内容。最后讨论了基于类质量树数据模型进行产品综合参数设计的成本-利润优化计算实例。第五章以类质量树模型为基础研究了产品质量控制中的知识与案例重用等问题。主体研究了质量控制中的质控参数与质量计划等文档内容的自动生成与审核管理，将质量控制中的图档信息与数据信息有效地结合，针对过程控制中的控制图，缺陷判断等内容引用了基于类质量树数据模型进行CBR与RBR混合推理的知识发现与重用方法。第六章有别于长期质量控制中数据信息独立于产品的结构模型进行管理与控制的方式，分析了基于类质量树数据模型的质量控制系统的技术优势，改善了质量过程数据对产品缺陷形成的诱因分析，以及零部件局部状态对成品质量的影响等方面的数据支持。结合科研项目研究与应用，实现了系统软件的研发。第七章总结了全文的主要研究内容和成果，并给出了今后的研究方向。更多还原

【Abstract】 Product life cycle quality control plays an important role in helping enterprises create better market position in a competitive marketplace. As the trend of demanding more flexibility in product customization grows, product quality control needs more intelligent correlative data process and utilization on data gathered from product life cycle in all stages. Intelligent and integrated quality management information system has recently attracted a lot of research and industrial interests, subjects in related technology are becoming hotspots in advanced manufacturing and quality assurance systems.This dissertation aiming at the organization, transaction and utilization of the product process quality data model, date mining of the quality characteristic and interrelated attribute, combination of the quality data with construct of the parts or components. Then give out a product quality data modeling—GBOM-QT (Generic-Bill-Of-Materials for Quality Tree),systematically researches on the GBOM-QT modeling, evolvement ,data mining, knowledge discover and rule reuse etc. Support by Science and Technology fund "The quality control and analysis system based on Datamining",the theory and methods presented in the dissertation have proved to be correct and advanced by the applications in a motor enterprises.The main content is presented as follows:Chapter 1 introduced the theory and technique of product quality control and information management. A brief history of product quality control and latest development were reviewed first. Knowledge-reuse based organization model of quality control, methods of modeling, information flow and system architecture, QFD modeling and its improvements, quality control and analyzing modeling in manufacturing process were discussed in great details. Knowledge organization, discovery and reuse for quality control, using data mining to find data relevancy, etc were also discussed, thus gave the research background, major purpose and the importance of this dissertation.In chapter 2 illustrates the characteristic of quality control such as status,process, heristeristic and structure etc. Some quality classify character is induced, such as quality plan, manufacture processing, fault diagnose and feedback.Then connotation of many knowledge is presented as structure,technique, detect , fault diagnose and analysis etc. Over above, the technique of the knowledge catching on quality control is put out, which can be implemented to optimal the product design,manufacture and assembly in quality control.The 3rd chapter analysis the relationship of between customer requirement and quality control,Product quality control data mining model, which is aiming at the product quality data and based on the BOM-QT, is presented. In the GBOM-QT, generic product structure and quality parameter is used to describe structural knowledge and rules to represent the implicit knowledge of a part or a component, this reduce the complexity of product quality control data. Presents a new framework to formalize describe the data structure of the GBOM-QT,and then the data definition of the GBOM-QT unit in form of XML.The methodology of the BOM-QT edit operation and the algorithm of the data mining in BOM-QT is discussed in Chapter 4. The minimum topological symmetric difference is used in calculate the similarity measure between trees. The method to transfer the non-order BOM-QT tree to ordered BOM-QT tree is denoted, and a improved way for the weighted symmetric difference between BOM-QT tree is presented, which include the influence of the node weighted distance. Then the algorithm of GBOM-QT trees combination from cluster, the constraints relation in XML scheme file transferred from GBOM-QT is also given.In chapter 5, the resolution of knowledge discover and rule reuse in quality control based on GBOM-QT model is discussed,which include the quality plan,quality control parameter selection auto building and checking. Quality control information is illustrated in drawing correlated with the processing data, such as SPC, fault diagnose etc. A new framework integrate case-based reasoning (CBR) with rule-based reasoning(RBR) based on GBOM-QT is put out.Deference from the data apart with the product structure, chapter 6 presents the superiority of the quality control based GBOM-QT data model. On the basis of this, then present the BOM_QT data can affect the quality state of the product,compenent and parts and improve measure. Supported by national and province science & technology program, a quality management information system centered on the processing quality control based GBOM-QT data model is developed and has been proved to be practical and valuable.In chapter 7, all achievements of the dissertation are summarized and the further research work is put forward.更多还原