Integrating multivariate engineering process control and multivariate statistical process control

Abstract

Multivariate engineering process control (MEPC) and multivariate statistical process control (MSPC) are two strategies for quality improvement that have developed independently. MEPC aims to minimize variability by adjusting process variables to keep the process output on target. On the other hand, MSPC aims to reduce variability by monitoring and eliminating assignable causes of variation. In this paper, the use of MEPC alone is compared to using the MEPC coupled with MSPC. We use simulations to evaluate the average run lengths (ARL) and the averages of the performance measure. The simulation results show that the use of both MEPC and MSPC can always outperform the use of either alone. To detect small sustained shifts of the mean vector, combing MEPC with a multivariate generally weighted moving average (MGWMA) chart (MEPC/MGWMA) is more sensitive than the MEPC/multivariate exponentially weighted moving average (MEWMA) chart and MEPC/Hotelling’s χ2 chart. An example of the application, based on the proposed method, is also given.