童超尘作者KEUNPYOKimMAHMOUDA-工业工程与管理系.PPT

On The Monitoring of Linear Profiles 指導教授：童超塵 作者：KEUNPYO Kim, MAHMOUD A.  MAHMOUD, AND WILLIAM H. WOODALL 主講人：廖莉芳 Outline Introduction Phase II Methods ARL Comparisons Phase I Methods Conclusions Introduction The quality of a process or product is characterized by a relationship between two or more variables instead of by the distribution of a single quality characteristic. We propose statistical process control charts for monitoring in Phase II a process or product that is characterized by a linear profile. Phase II Methods When the process is in statistical control The first strategy is to use a bivariate T2 chart to monitor the regression parameters Phase II Methods When there are shifts from the in-control parameter values, Tj2 has a non-central chi-square distribution, one can evaluate the exact ARL of T2 chart using the following formula： Phase II Methods The second strategy is to apply some standard control chart schemes to the regression “residuals”. They suggested an R-chat be used in combination with the EWMA chat. Phase II Methods There are two reasons for an R-chat to be used in conjunction with EWMA chart. The EWMA chat is not sensitive to shift in the process variation. The EWMA chat based on the average residual is not sensitive to some shift in A0 and A1, for which the magnitudes of the residuals tend to be large, but the average residuals tend to be small. Phase II Methods For monitoring B0： Phase II Methods For monitoring the slope B1： Phase II Methods For monitoring the error variance σ2： ARL Comparisons ARL Comparisons ARL Comparisons ARL Comparisons Our method does not work quite as well as the EWMA /R method. ARL Comparisons ARL Comparisons ARL Comparisons We have Sxx=5 instead of Sxx=20： Phase I Methods We assume here that k samples of simple linear regression data Phase I Methods Conclusions Our methods are more effective. We can extend our methods to be more complicate. Polynomial regression model. Multiple regressi