A directional multivariate exponentially weighted moving average (MEWMA) scheme combining the EWMA scheme with the generalized likelihood ratio test scheme is proposed that can incorporate directional information based on the multistage state-space model and monitor the process mean shift. This scheme provides a solution that incorporates both interstage and intrastage correlations, and resolves the confounding issue in monitoring and diagnosing due to the cumulative effects from stage to stage. Simulated results show that the scheme outperforms most existing approaches to multistage processes.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados