Multivariate Monitoring of the Process Mean and Variability Using Combinations of Shewhart and MEWMA Control Charts

Control charts are considered for the problem of simultaneously monitoring the mean and variability of a multivariate process when the joint distribution of the process variables is multivariate normal. We investigate sets of univariate EWMA charts used in combination with sets of Shewhart charts or in combination with sets of EWMA charts based on the squared deviations of the observations from target. We also investigate the MEWMA control chart used in combination with a multivariate Shewhart control chart or with a form of MEWMA-type chart based on the squared deviations from target. We conclude that a combination of multivariate charts gives somewhat better average performance than a combination of sets of univariate charts, and that a combination of MEWMA charts that includes one based on the squared deviations from target gives the best overall performance.