Monitoring autocorrelated multivariate simple linear profiles

Recently, many researchers and practitioners have shown interest on profile monitoring as a relatively new subarea of statistical process control. One main reason for this interest, and perhaps a key factor for the contributions of many researchers to this field, is the various applications of profile monitoring in real life. Although one can easily encounter many univariate applications of profile monitoring in service and manufacturing environments, there exist situations where quality of a product or process needs to be modeled in multivariate terms. In this paper, we investigate monitoring of multivariate simple linear profiles in phase II when independence assumption of observations within profile is violated. A remedial measure based on a transformation method is proposed to remove the autocorrelation structure within multivariate profiles, and numerical simulation is used as a vehicle to evaluate the effect of the proposed solution. Finally, a case study is considered to show the application of the model in practice.