Application of Batch Means Control Charts for Mean Shift Detection in Autocorrelated Data of Machining Processes

High-frequency sensor data in modern manufacturing environments are typically autocorrelated, which violates the independence assumption of conventional Shewhart control charts and results in excessive false alarms or delayed shift detection. The batch means control chart offers a simple alternative by aggregating consecutive observations to mitigate autocorrelation without explicit time-series modeling. This study evaluates the performance of batch means control charts for detecting mean shifts in first-order autocorrelated processes. Through simulation, the effects of batch size and batch interval on the average run length are analyzed under varying autocorrelation levels, and practical design guidelines are proposed. A case study using sound signal data from a thread turning process demonstrates the applicability of the method. Results confirm effective false-alarm reduction and reliable mean shift detection in autocorrelated manufacturing processes.