Abstract
The multivariate extension of control charts for process dispersion is not as straightforward as that for the process mean. A general model and techniques that would encompass a wide range of problems encountered in practice is not available. In most cases, particular problems need to be handled in a specific manner. In this paper we consider several special cases of a process displacement affecting the covariance matrix and we develop control charts (both Shewhart-type and CUSUM) to detect these process changes.
Acknowledgements
The authors wish to thank the two anonymous referees for comments and suggestions which improved the presentation of the paper.