It has been suggested that detecting large and small shifts in the mean when monitoring a process with multivariate normal variables can be better accomplished using a multivariate exponentially weighted moving average (MEWMA) control chart in combination with the Shewhart chart. This study investigates the performance of this combination in the more general problem of detecting changes in the mean or increases in variability. When considering both sustained and transient shifts, it can be shown that a combination of two MEWMA charts performs better than a combination of the MEWMA and Shewhart charts. Also considered is a three-chart combination of the MEWMA chart for the mean, a MEWMA-type chart of the squared deviation from target, and the Shewhart chart. This combination has significantly better performance for some mean shifts, but poorer performance for shifts in variability, when n is greater than 1.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados