Online monitoring of autocorrelated multivariate linear profiles via multivariate mixed models

Multivariate multiple profile monitoring has been studied extensively over the past few years. Most of these studies assumed that the observations are uncorrelated, which could be violated in practice. In this paper, multivariate linear mixed model is proposed to allow correlation among observations of the multivariate multiple linear profiles. In order to monitor random effects and process variability in phase II, three control charts are suggested. The results of performance comparisons with an existing method show the superiority of the proposed control chart. Finally, the applicability of the proposed method is illustrated using a real case.

Automated summary

Multivariate multiple profile monitoring has been extensively studied, with the proposal of a multivariate linear mixed model allowing correlation among observations. Three control charts are suggested for monitoring random effects and process variability in phase II, showing superiority over existing methods. The applicability of the proposed method is illustrated through a real case demonstration.