Model-Based and Model-Free Control of Autocorrelated Processes

Abstract

Advances in automated sampling technology have made autocorrelated data commonplace. Positive autocorrelation degrades control charts designed by classical methods. If a correct time-series model of the autocorrelated process is available, many have advocated the use of control charts on the residuals from the model. Using the average run length criterion in an AR(1) model, we show that plotting averages of batches of the raw data can be an effective alternative to plotting residuals. We consider both weighted averages and the simple, model-free approach of arithmetic averages. We compare these statistics to residuals in both Shewhart and cumulative sum (CUSUM) control charts.

• Cumulative Sum Control Charts
• Residuals
• Shewhart Control Charts
• Simulation
• Statistical Process Control
• Time Series Analysis

Additional information

Notes on contributors

George C. Runger

Dr. Runger is an Assistant Professor in the College of Business and Management and a consultant in process improvement. He is a Member of ASQC.

Thomas R. Willemain

Dr. Willemain is an Associate Professor in the Department of Decision Sciences and Engineering Systems.