Performance of self-starting control charts for autocorrelated data. Academic Article in Scopus uri icon

abstract

  • © 2018 Institute of Industrial Engineers (IIE). All rights reserved. Under the Statistical Process Monitoring label, control charts are powerful tools to determine whether or not a process is under statistical control. Nevertheless, their power relies on the validation of the assumptions made in their design such as prior knowledge of the in-control parameters and independent data, which are seldom met in practice. Even though there are solutions to overcome these issues, they have been applied apart: the self-starting control charts (SSCC) were used when parameters are unknown in order to avoid the necessity of large Phase I samples while residuals control charts have been used for autocorrelated data considering time series models. In this study, we connect these two approaches by considering self-starting control charts for monitoring the mean of stationary autoregressive processes of order 1, AR(1), under parameter estimation. The in-control performance of the Shewhart Q, CUSUM Q and EWMA Q control charts is evaluated by means of the average and standard deviation of the Average Run Length. With this study, it will be possible to determine if the self-starting methodology is a suitable approach for monitoring autocorrelated processes with estimated parameters.

publication date

  • January 1, 2018