【作者】 董延峰；

【导师】 何桢；

【作者基本信息】 天津大学 ， 管理科学与工程， 2008， 博士

【摘要】 随着产品质量在企业在参与市场竞争中占据越来越重要的作用,统计过程控制(SPC)也得到越来越广泛的应用,特别是6σ质量改进理论和方法开始进入各大生产企业。对于某些存在多变异源过程的质量控制,为企业所熟悉的建立在单一变异源基础上的传统休哈特控制图会出现虚发警报过高的问题,严重影响控制图的应用效果,因此需要研究多变异源工序控制方法。本文针对多变异源过程控制所急需解决的控制方法问题进行了理论研究,主要工作和成果如下:1.全面归纳了多变异源过程控制的现有理论,深入研究了多变异源过程控制方法,建立起一套集多变异分析、多变异源过程控制方法、控制图工具选择、抽样方案确定的多变异源过程控制流程方案。2.首次在多变异源过程控制中引入了指数加权移动平均控制图,实际结果表明,指数加权移动平均控制图要比改进的休哈特控制图的效果要好,提高了多变异源过程均值小偏移的检出力。3.在建立了多变异源控制图之后,又继续研究了多变异源控制图的操作特性曲线,给出了有关的公式,这有助于了解多变异源控制图的特性,分析控制图的效果和性能。4.在多变异源控制图操作特性曲线研究的基础上,给出了确定样本含量的方法,结果表明,通过给定的第二类风险概率,能方便的确定多变异源控制图的样本含量。5.通过多变异源过程参数的置信区间与样本含量的关系,为多变异源控制图的样本含量的确定找到了另一个办法。给定某个变异源方差分量的置信区间,就能确定相应的样本含量。为多变异源控制图节约抽样成本,扩展应用范围打下了基础。6.以存在多变异源的高精度主轴加工工序为实践对象,以实例说明本文所建立的一整套多变异源工序控制方法流程。更多还原

【Abstract】 Global market competition among manufacturers has prompted the rapid developing of statistical process control(SPC), especially for 6σquality improvement. But for some process with several source of variation, traditional shewhart control charts monitoring single sources of variation typically produce an unacceptably high number of false alarms, sometimes rendering the whole control system useless. So control system for multiple source of variation must be considered.This dissertation researches on theory and the methodology for quality control for machining process due to various sources.1. The dissertation sums up and develops systematically theory and the methodology for quality control for process due to various sources. A scheme for multi-vari process control is presented including multi-vari analysis, the choice of control charts and the sample size selection.2. The dissertation firstly use exponentially weighted moving average(EWMA) control charts are used to monitor the process with various sources of variation. It is proved that the power of EWMA control charts is more sensitive than improved shewhart control charts.3. The operating characteristic (OC) curves of the control charts with various sources of variation were calculated, presented and compared to those of the traditional shewhart control charts. This OC curves can assist in the evaluation of the power of control charts.4. The dissertation discusses the theory and methods to calculate sample size of control chart monitoring various sources of variation based on the OC curve of the control charts. An acceptableβwill depend on the sample size from OC curve, so the appropriate choice of sample sizes will usually be made on economic grounds.5. The dissertation also proposes another propoach to estimate the sample size based on confidence interval of the variance component due to various sources. An acceptable confidence interval of the variance component will determine the choise of sample size on economic grounds6. The process control of axis machining provides the applications example of process control with multiple sources of variation.更多还原