Guidelines for Applying Statistical Quality Control Method to Monitor Autocorrelated Prcoesses

AbstractThe implementation of statistical control charts under autocorrelated situations is a critical issue since it has a significant impact on the monitoring capability of manufacturing processes. The objective of this study is to assess the performance of control charts under different scenarios and to optimize the design of control charts to best deal with autocorrelated processes. To achieve the proposed objective, two autoregressive integrated moving average models, ARIMA (1, 0, 1) and ARIMA (0, 1, 1), are utilized to characterize stationary and non-stationary processes. These process models were simulated to achieve the response, average run length (ARL), which is the performance measure of this study. The factorial design of experiment was conducted to quantify the effect of critical factors, i.e., ARIMA coefficients, types of charts (exponentially weighted moving average: EWMA and moving range: MR) and shift sizes on the ARL. The experimental results show that EWMA chart is the most appropriate control chart to monitor autocorrelated observations. Additionally, both AR and MA parameters along with shift sizes have a significant effect on the performance of control charts. Therefore, this study has pointed out a suitable tool for use under the different scenarios of autocorrelation. The validation of the above experimental results was conducted on another ARIMA model, ARIMA (1, 0, 0). If the performance of control charts under autocorrelated disturbances is correctly characterized, practitioners will have guidelines for achieving the highest possible performance potential when deploying SPC