Desempeño estadístico de cartas de control con parámetros variables para procesos autocorrelacionados

Abstract

Chemical process and automation of data collection for industrial processes are known to result in auto-correlated data. The independence of observations is a basic assumption made by traditional tools that are used for the statistical monitoring of processes. If this is not adhered to then the number of false alarms and quality costs are increased. This research considers the use of variable parameters (VP) charts in the presence of autocorrelation. The objective is to determine the impact on the detection velocity and false alarms of different degrees of autocorrelation and their interaction with process conditions—the aim being to effectively select parameters. The VP chart showed improvements in its performance when detecting large average runs as the autocorrelation coefficient increased in contrast with the traditional variable sample interval (VSI) and x quality systems. This research demonstrates the superiority of variable parameters quality monitoring architectures over traditional statistical process monitoring tools.