A0895
Title: A generalized distribution-free multivariate control chart with improved post-signal diagnostics
Authors: Chase Holcombe - University of South Alabama (United States) [presenting]
Abstract: Multivariate statistical process control (MSPC) charts are particularly useful when the need arises to simultaneously monitor several quality characteristics of a process. Most control charts in MSPC assume that the quality characteristics follow some parametric multivariate distribution. Distribution-free MSPC charts are attractive, as they can guarantee in-control (IC) or null performance of the control chart without the assumption of a parametric multivariate process distribution. Utilizing an existing tolerance interval, a simple phase II Shewhart-type distribution-free MSPC chart is constructed and proposed for individual and subgrouped observations. The proposed charting methodology preserves the original scale of measurements and can easily identify out-of-control (OOC) variables after a signal, which are both important practical advantages in the multivariate setting. The proposed control chart is attractive as it is easy to construct, visualize, and interpret, is exactly distribution-free, requires no complex parameter estimation calculations for implementation, comes with a natural and simple post-signal diagnostic mechanism, and only requires a modestly large reference sample size for small to moderate dimensions. IC and OOC performance is discussed. Effects of contamination and high dimensionality are also briefly discussed.
