【作者】 戴敏；

【导师】 易红； 史金飞；

【作者基本信息】 东南大学 ， 机械制造及其自动化， 2006， 博士

【摘要】 本文以多工序生产模式为背景,针对多工序生产系统加工工序多、工艺复杂、制造成本高、质量影响因素多和质量特性波动源辨识困难等特点,结合国内某半导体企业在质量控制方面存在的问题,对多工序生产系统的质量分析方法进行研究。并结合国家自然科学基金项目“动态质量信息的XML体系建模研究”(70272046),对多工序生产系统中质量信息的集成方法进行研究。论文的主要工作和研究成果如下:1).分析传统的过程能力指数只能衡量一个工序的加工能力,不能得知整条生产线的加工能力概况的不足,提出应针对多工序生产系统的特点,从整条生产线的角度出发,研究多工序生产系统的质量分析方法。结合半导体晶圆生产中质量损失的特点,对田口损失函数进行了修正,建立了适用于半导体晶圆生产的损失函数。在此基础上,推导出过程能力指数和质量损失函数之间的关系,提出了意义更广的过程能力指数——生产线过程能力指数。通过引入质量损失度的概念,建立了各工序过程能力指数和生产线过程能力指数之间的定量关系。针对多工序生产中质量特性存在多种变异的情况,在半导体晶圆生产中提出了基于方差分析的多变异抽样模型。2).误差是产品质量的一种衡量方式,本文分析了多工序制造过程中误差的来源、特点和分类。在对单工序误差传递进行研究的基础上,建立了多工序生产系统误差传递的迭代模型,通过研究多工序生产系统的误差传递规律来探测质量波动根源。针对实际生产中有些质量特性在几个工序完成后才能进行测量的情况,提出复合工序的概念;将误差传递与粗糙集理论结合,提出了基于优先关系的粗糙集方法,解决复合工序的质量分析问题。3).分析了多工序制造过程中的质量信息流,确定了工序质量信息的基本要素,在分析多工序制造过程质量信息特点的基础上,建立了基于XML(eXtensible Markup Language,可扩展标记语言)的质量信息体系。提出了采用XML语言来表达质量信息,建立了质量信息文档的MMP-XML(Multistage Manufacturing Process system XML,多工序制造系统可扩展标志语言)表达模式。4).分析了半导体企业信息管理系统的现状,提出了基于制造执行系统的多工序生产过程质量分析系统架构。结合江阴新顺微电子有限公司生产过程管理系统项目,开发了多工序生产过程质量分析系统。更多还原

【Abstract】 In multistage manufacturing systems, the processing is usually complex and high cost. The product quality is influenced by variant unknown factors. The methods for quality analysis are discussed in this thesis. The integration technique for quality information of Multistage Manufacturing Process (MMP) is also presented. This research is funded by the National Natural Science Foundation project“Study on XML System Modeling of Dynamic Quality Information”. The major part of research work can be included in following sections.The traditionary process capability index (PCI) can only be used to estimate simple process. The whole process capability can not be obtained. By the study of the relationship of PCI and Taguchi’s quality loss function (QLF), a new concept - Multistage Process Capability Index (MPCI) is introduced. On the basis of quality loss extent, a quantitative relationship between the MPCI and the PCI is discussed. An ANOVA (analysis of variance) sampling plan is set up in semiconductor manufacturing to resolve the problem of multi-variation of the product quality in MMP.Variation is a standard of the product quality. The variation source of MMP is studied. On the basis of single process variation transmission analysis, an iterative model of MMP variation transmission is presented. A method to detect the variation source is discussed via the study of variation transmission rules. By combining the rough set (RS) and variation transmission rules, a Dominance-based Rough Set Model is approached to analysis the MMP quality.Through the analysis of MMP information flow and the basic elements, an XML mode and the framework of MMP quality information is established. How to describe MMP quality information in XML is discussed. A new Procedure of information extracting from MMP-XML document is presented in detail.The actuality of enterprise information management system for semiconductor industry is analyzed. A framework of MMP quality analysis system (MMP-QAS) is established. It is well compatible with MES. In the project of XS-MES for Jiangyin Xinshun Microelectronic Co., MMP-QAS is integrated in the system.更多还原