Abstract: Based on deeply analyzing statistical shortcomings of applying generalized variance control charts to monitor covariance matrix changes of multiv ariate process,this paper puts forward two new control charts,C control chart and S control chart,which are respectively applicable to monitor the consistency of the coordinate system and the similarities among ellipsoids,which are composed by eigenvectors and eigenvalues of the covar iance matrix correspondingly.It can effectively improve statistical monitoring ability of covariance matrix changes of multivariate process.Simulation analysis showed that through combined use of C control chart,S control chart and regular mean T² control chart,systematic and efficient monitoring of mean and covariance of the multivariate process can be achieved.

Key words: multivariate process, normal distribution, statistical control chart, covariance matrix, latent structure