New monitoring strategies for multivariate processes

Abstract

When Shewhart proposed the control chart in 1927, his intention was to offer a very simple statistical tool for monitoring processes. After all these years, processes became more and more sophisticated and consequently, they are now requiring the use of more efficient statistical tools in the sense of detecting process disturbances faster. Moreover, the growing development in automation has led to an increased need of multivariate charts for the joint control of several quality characteristics. The computational resources available nowadays are allowing to redesign the control charts with better performance. So, a new generation of Shewhart charts with variable parameters, with more complex sampling schemes and special run rules are drawing the practitioner attention. The number of papers dealing with these new control charts is considerable in the case of univariate processes, but scant in the case of multivariate process, and in general, the simulation has been the technique used to study the properties of the multivariate control charts. In this project, we intend to extend the studies elaborated to obtain the control chart properties for the surveillance of univariate processes to the multivariate case. We expect to obtain the properties of the new generation of control charts for multivariate processes with better accuracy than the properties obtained with the simulation technique. (AU)