Identifying the Time of Step Change in the Mean of an MA(1) Process

摘要

When applying statistical process control (SPC) charts to industrial processes, one often finds that the fundamental assumption of uncorrelated observations is violated. In this paper, a change point model based on the first-order moving average process is presented. The estimator is intended to be used as an add-on following signals from SPC charts that are appropriate for monitoring autocorrelated processes. The performance of the proposed change point estimator is evaluated following genuine EWMAST control chart signals using Monte Carlo simulation. Results indicate that the estimator provides process engineers with an accurate and useful estimate of the last sample obtained from the unchanged process.