【作者】 苏海涛；

【导师】 杨世元；

【作者基本信息】 合肥工业大学 ， 精密仪器及机械， 2006， 博士

【摘要】 本论文是国家自然科学基金资助项目“基于质量信息技术集成的‘全质量’管理系统研究”(资助号:70272032,起止日期:2003.1-2005.12)研究内容的一部分。 本论文重点研究基于质量信息技术集成的“全质量”管理系统模型:建立以质量监控和质量管理相结合为主旨的“全质量”管理系统;研究现代质量管理的新理论和新方法;研究质量信息的获取、传输、处理等方法;研究质量管理、质量监控与质量评价方法;建立相关的多种数学模型(如灰色系统模型、神经网络模型等)。 提出了以质量信息技术集成为基础、以质量监控技术与质量管理相结合的“全质量”管理系统总体模型。重点建立了全质量管理概念模型、基于绩效目标的系统数学模型、基于系统工程理论的信息集成方法三维结构模型等9个模型,构建了系统总体框架。 研究了现代企业质量信息的获取、传输及处理方法与模型。建立了质量信息追溯3维结构模型和因果关系数学模型,研究了基于搜索引擎的因果关联强度分析方法;建立了质量信息3层(现场层、局域层和广域层)网络传输结构模型和广域网优化质量信息传输模型;研究了计量器具的5层信息编码方法,实现了基于BP神经网络字符智能识别,获得较好的网络识别精度和容错能力;研究了质量信息的动态灰色预测模型,应用于计量器具检定周期预测,获得较好的精度;研究了基于支持向量机的传感器动态补偿方法,明显缩短了传感器达到稳定的时间,提高动态测量系统的精度;研究了基于SFOM网络的质量信息聚类分析方法,实现了不同类型质量信息样本的区分、同种类型质量信息样本间的一致性程度分析、多检测样本与标准样本吻合性排序分析。 构建了质量管理、质量监控与质量评价模型。建立了产品质量创新绩效模型、因素关联模型、质量创新框架模型;研究了基于网络的光机电一体化质量监控体系功能结构模型,建立了远近程多传感器汽车故障监控诊断系统,构建了故障诊断数学模型,实现了基于BP神经网络故障仿真识别;研究了单光束照射细丝外径质量信息获取方法和数学模型,选用径向基函数神经网络构建在线预测优化控制体系,获得较好的的测量精度、较小的预测误差,适合实时优化监控;建立了产品质量创新绩效目标多层综合评价结构模型和数学模型、基于DEA方法的质量控制效率优化评价数学模型,结合Matlab语言实现了质量控制相对有效性排序评价。 本论文运用模糊集理论、灰色系统理论、神经网络理论等现代理论,建立“全质量”管理系统多种相关数学模型,并运用现代信息技术、计算机新技术对“全质量”管理系统进行实用研究,为研究成果转化为生产力提供了实现方法和技术路线;运用系统论、信息论、控制论及并行工程等理论,将现代企业的质量监控和质量管理融合在一起,并引入信息科学、计算机技术、测控技术、数学方法等开展多学科的交叉研究,建立了基于质量信息技术集成的“全质量”管理系统,为现代企业质量管理的研究引入新思路和新方法;结合企业实际进行了试验研究和部分实证研究,对发展我国现代企业的质量管理和质量控制技术,尽快缩小我国在适应高新技术领域的质量管理方面与国外的差距,更多还原

【Abstract】 This paper is supported by the project: INTEGRATIVE QUALITY Management System based on Quality Information Technology Integration, which is supported by National Natural Science Foundation of China (No.70272032).Integrative quality management system model based on quality information technology integration is the main study of the paper. Several new theories, new methods and new models were proposed: integrative quality management system based on quality monitoring and quality management;new theories and methods of modern quality management;methods of information acquirement, transmission and disposal;several mathematics models, methods of quality management, monitoring and evaluation;etc..By combining quality monitoring technology with quality management, overall model of Integrative quality management system based on quality information technology integration was proposed. Nine models were emphasized to build an overall frame of the system. These models are Integrative quality management conceptual model, mathematical model based on performance goal, three-dimension model of information integration methods based on system engineering, and so on.Quality information acquirement, transmission and disposal methods and models of modern enterprise were studied. Three-dimension quality information backward model and causal mathematical model were set up, to study causal intensity analysis method based on search engine. A three-layer network transmission architecture model and a WAN optimal transmission model for quality information were set up. A five-layer coding method of measuring instruments was studied to realize intelligent character recognition based on BP neural network. And it proved to be higher accuracy and fault-tolerance ability. Dynamic gray prediction model of quality information was studied, to predict verification cycle of measuring instruments, and it has satisfying accuracy. Dynamic sensor compensation method based on supported vector machine was studied. It shortens the time to stable and improves the accuracy of dynamic measuring system. A cluster analysis method of quality information based on SFOM network was also studied. It realized distinguishing different type of quality information samples, analyzing consistence among the same type of quality information samples, and analyzing consistence sequence between a large amount of detected samples and standard samples.Quality management, monitoring and evaluation models were set up. Quality innovation performance model, factor model and quality innovation structure model of products were set up. A function model of optic-mechano-eiectronic quality monitoring system based on web was studied, to set up mathematics fault diagnoses model and multi-sensor fault monitoring and recognition system of automobile. Intelligent recognition was realized by BP neural network. A filament quality acquirement method by single beam illumination and its mathematics model were studied. With some prediction simulation based on RBFNN (Radial Basis Functions neural network), it is proved to be satisfying accuracy and be suitable for real-time monitoring. An architecture and mathematics model of multi-level synthesis valuation on quality innovation performance goal and a mathematics model of efficiency optimal evaluation on quality control based on DEA were set up. Valuation on relative validity sequence of quality control combined with Matlab was realized.Modern theories such as Fuzzy Set Theory, Gray System theory, Neural Network Theory, etc. were used in the study to build mathematical models of Integrative quality management system. Modern Information Technology, Computer Technology were used to realize the system and to provide realistic methods and techniques. Besides System, Information, Control Theories, Concurrent Engineering, etc., there were some other subjects were introduced into the study to build integrative quality management system based on quality technology integration, such as Information Science, Computer Technology, Measurement and Control Technology, Mathematics, and so on. It introduces some new thoughts and methods into the research on quality management of Chinese modern enterprises. With experiments and empirical study on enterprises, it has scientific and wildly application significance. It will improve overall quality and core competence of enterprises, shorten the high-tech quality management gap between our country and developed country, and propel the development of Management Science.更多还原