检测变异系数的控制图

发布时间：2018-07-05 07:30

本文选题：变异系数 + CUSUM　；参考：《辽宁大学》2017年硕士论文

【摘要】：在传统质量控制应用中,控制图通常用来检测过程的平均值或者方差,有些时候,过程的平均值或方差不是一个常数,而他们的比值-变异系数,在过程受控时却是一个常数,因此,检验变异系数就更合理,更加符合实际需求.根据变量的维数,变异系数可分为一元变异系数和多元变异系数,本文从一元及多元变异系数两个方向,研究变异系数的检测问题.首先,文章将传统的累积和(CUSUM)控制图和指数加权移动平均(EWMA)控制图相混合,构造一种新的检测一元变异系数控制图-混合CUSUM-EWMA控制图,并和已有的一些方法进行比较,结果表明,本文提出的控制图对检测变异系数的中小漂移具有更好的检测能力.其次,在多元变异系数相等性的检测问题中,首先介绍了多元变异系数的定义方式,并根据定义提出检测多元CV相等性的EWMA控制图.模拟研究表明,本文提出的多元控制图对检测多元CV相等性的问题具有很好的表现.
[Abstract]:In traditional quality control applications, control charts are usually used to detect the average or variance of a process, sometimes the average or variance of a process is not a constant, and their ratio-coefficient of variation is a constant when the process is controlled. Therefore, the test coefficient of variation is more reasonable, more in line with the actual needs. According to the dimension of variables, the coefficient of variation can be divided into univariate coefficient of variation and multivariate coefficient of variation. In this paper, the detection of coefficient of variation is studied from two directions of univariate and multivariate coefficient of variation. Firstly, the traditional CUSUM control chart and the exponentially weighted moving average (EWMA) control chart are mixed to construct a new control chart for detecting univariate variation coefficient, which is a hybrid CUSUM-EWMA control chart. The control chart presented in this paper has a better ability to detect the small drift of coefficient of variation. Secondly, in the problem of detecting the equality of multivariate coefficient of variation, the definition of multivariate coefficient of variation is introduced, and the EWMA control chart is proposed to detect the equality of multivariate CV. The simulation results show that the multivariate control chart presented in this paper has a good performance in detecting the equality of multivariate CV.
【学位授予单位】：辽宁大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：F224;F273.2

【相似文献】