New Robust MEWMA Control Chart for Monitoring Contaminated Data

Abstract

Multivariate Exponential Weighted Moving Average (MEWMA), E2 control chart is a popular multivariate control chart for monitoring the stability of time series data (non-random pattern). However, in this paper, we have shown that the existing MEWMA, E2 control chart is sensitive in contaminated data or in the presence of outliers. To address this problem, this paper proposed an alternative MEWMA E2 control chart using robust mean vector and covariance matrix instead of the classical mean vector and covariance matrix respectively. The classical mean vector in MEWMA E2 control chart is replaced by Winsorized Modified One-step M-estimator (WM) while the classical covariance matrix is replaced by the Winsorized covariance matrix. The proposed MEWMA E2 control chart known as robust MEWMA control chart, denoted as RE2 control chart. The control limit for the RE2 control chart was calculated based on simulated data. The performance of RE2 and existing MEWMA E2 control charts are based on the false alarm rate. The result revealed that the RE2 control chart is more effective in controlling false alarm rates as compared to the existing MEWMA, E2 control chart. The zinc-lead flotation data show that the RE2 performs better in application