An Adaptive Exponentially Weighted Moving Average Control Chart

• Activity Number: 430
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Quality and Productivity Section
• Author(s): Amitava Mitra* and Kang Bok Lee
• Companies: Auburn University and Auburn University
• Keywords: process mean shift; smoothing constant, average run length to first detection

Abstract

Exponentially weighted moving average (EWMA) control charts are typically used for faster detection of shifts in the process mean relative to a Shewhart control chart, when the degree of shift is small. While such control charts put more weight on recent observations than those further back in history, they choose a fixed value of the smoothing constant or weight. Here, an adaptive control chart is proposed where the selected weight varies based on the observed statistic. The selected statistic incorporates the squared deviation of the observations from a target value and utilizes a chosen look-back period. The performance of the proposed adaptive EWMA chart is studied through a simulation procedure where the degree of shift of the process mean and the probability of a Type I error is controlled. The chosen performance measure is the average run length to first detection. The proposed adaptive control chart performs well.