The Residual T2 Control Chart of the Multivariate Heteroskedasticity Process with Trend Patterns

摘要

Statistical process control(SPC) is widely used to improve the product quality. The quality of a product can be attributed to several correlated quality characteristics, all of which need to be controlled and monitored simultaneously. Linear trends in processes are usually due to some type of process decay, such as the wearing out of a tool, where there is often heteroskedasticity. In this paper, the models of the residuals of the process with heteroskedasticity are fitted using Glejser test method which can test the increasing or decreasing heteroskedasticity of the process, which can distinguish three kinds of variables in the processes. Then, the identically independent random residuals after eliminating the linear trend and heteroskedasticity can be controlled by the multivariate T2 control chart. At the end, the axis diameter and surface roughness in the tool wear process are analyzed with the residual T2 control chart proposed.