- 作者: Chao-Yu Chou ; Bernard Chen-Chun Jiang ; Joseph Ching-Shihn Chen
- 中文摘要: Statistical techniques applied to industrial process control have been revitalized due in part to efforts by Demimg. One of these techniques is the analysis of process capability indices. Process capability indices are unitless and provide a common and easily understood language for quantifying the performance of a process. Various process capability indices have been applied to measure process performance in Japan, the U.S.A. and the other countries. In the applications of process capability indices, it is usually assumed that there is no correlation among the measurements within a sample; i.e., the measurements within a sample are independently distributed. However, in practice, this assumption may not be tenable. It would be more appropriate to assume that each sample is a realization of a multivariate normal random vector. The purposes of this article are: (1) to investigate the effect correlation on the process capability index Cp, and (2) to present the uses of the process capability index Cp, along with its sampling property and estimation procedure, when correlation within a sample exists. From the study, it can be shown that the capability index Cp will be over-estimated if the effect of positive correlation is neglected, and that Cp will be under-estimated if the effect of negative correlation is ignored.
- 英文摘要: --
- 中文關鍵字: process capability index, correlation, estimation, hypothesis testing, non-central chi-square distribution
- 英文關鍵字: --