Nonparametric likelihood ratio-based EWMA control chart

Abstract

ABSTRACT Nonparametric control charts are among the most important tools of statistical process control. Such charts are useful when there is a lack of knowledge about an underlying distribution or its parameters. Most existing nonparametric control charts are used for monitoring location parameters: they may perform poorly when the scale parameters change arbitrarily over time. In this paper, we propose a new nonparametric control chart based on a powerful likelihood ratio test and the exponential weighted moving-average (EWMA) control chart. Its nonparametric property is applied via a consistent series density estimator in the exponential family. The proposed control chart does not require historical reference samples and can be monitored by fixed control limits. The Monte Carlo simulation results show that the proposed control chart performs well in monitoring both mean and variance shifts, especially in monitoring variance or large mean shifts. Furthermore, a real-data example is given to illustrate the effectiveness of the proposed method.