A performance comparison among the Hotelling T2 and the rank clasification control charts

Abstract

IN THE INDUSTRIAL SECTOR IT IS FUNDAMENTAL TO HAVE STATISTICAL METHODS THAT ALLOW EFFICIENT MONITORING OF PRODUCTION PROCESSES, PARTICULARLY WHEN THE CHARACTERISTICS OF QUALITY ARE MULTIVARIATE. THE PRODUCTION PROCESSES ARE GENERALLY MONITORED BY MEANS OF CONTROL CHARTS, WHICH GIVE GRAPHIC INFORMATION THAT TELLS IF THE PROCESS IS, OR NOT, UNDER CONTROL. IN THIS CONTEXT, THIS ARTICLE, COMPARES, THROUGH SIMULATION, THE PERFORMANCE OF TWO TYPES OF MULTIVARIATE CONTROL CHARTS: THE CLASSIC CHART BASED ON HOTELLING T² STATISTICS, AND A NON-PARAMETRIC CHART BASED ON THE CONCEPT OF DATA DEPTH. WE DECIDED TO PERFORM THE SIMULATION WHEN, STUDYING A PROCESS WITHIN THE AUTOMOTIVE INDUSTRY USING BOTH APPROACHES, THE CONCLUSIONS ARRIVED AT WERE DIFFERENT. RECOMMENDATIONS ARE GIVEN ABOUT THE USE OF ONE OR THE OTHER CONTROL CHARTIN ACCORDANCE WITH THE STATISTICAL BEHAVIOR OF THE MONITORED PROCESS