Online monitoring of auto correlated linear profiles via mixed model. (1st January 2013)

• Record Type: Journal Article
• Authors: Soleimani, Paria; Narvand, Ali; Raissi, Sadigh
• Is Part Of: International journal of manufacturing technology and management. Volume 27:Number 4-6(2013)
• Page Start: 238
• Page End: 250
• Publication Date: 2013-01-01
• Subjects: multivariate process monitoring -- linear profile -- mixed model -- multivariate cumulative sum -- MCUSUM -- multivariate exponential weighted moving average -- MEWMA -- HottelingT2 -- auto correlation -- average run length; Manufacturing processes -- Periodicals; Manufacturing processes -- Automation -- Periodicals; Production management -- Periodicals; 670.427068
• Languages: English
• ISSNs: 1368-2148

Abstract

In statistical quality control a profile can be characterised by a given mathematical function between a quality characteristic and one or more explanatory process variables. Most existing control charts in the literature have been proposed for profile monitoring with the independence assumption of the observation within profiles. However in certain situation, this assumption can be violated. The present study focused on phase II of a linear profile monitoring and extends Jensen et al. (2008)'s work in applying linear mixed models on the presence of autocorrelation within profiles. Three methods namely Hotteling T 2, multivariate exponential weighted moving average (MEWMA) control chart and multivariate cumulative sum (MCUSUM) control chart are discussed and their performances are compared in term of average run length (ARL). These techniques are illustrated with a real data set taken from an agriculture field.