Alternative estimation procedure in SPC when the process data are correlated

Abstract

If the process observations are autocorrelated, the performance of control chart is influenced significantly. This autocorrelation leads to a large false-alarm rate. This article considers the problem of monitoring the mean of AR(1) process with random error. We provide a simple algorithm to improve the estimation results of process parameters. Simulation results show that the proposed method can produce stable and adequate estimates for the AR(1) process with random error, even though the sample size is small.

Keywords:

• Autoregressive moving average model
• Exponentially weighted moving average control chart
• First-order autoregressive model
• Maximum likelihood estimation
• Shewhart control chart