On developing robust adaptive approaches for monitoring location of non‐normal environments

Abstract

AbstractThe assumption of normally distributed data is emerged with several statistical inferences as well as in statistical process control. But various real practices on data in different fields like Biological sciences, health, production processes, and manufacturing industries exhibit non‐normal behavior. The current study is concerned with developing robust adaptive exponentially weighted moving average (AEWMA) control charts to monitor the location of non‐normal environments. In the current study, four estimators are considered and listed as Mean (, Mid‐range (), Median (and Trimean () for observing process target. Robust proposals of the said schemes are scrutinized towards symmetric non‐normal (t and Laplace) and skewed (Log‐normal and Gamma) environments. The average of run‐length and standard deviation of run‐length are taken as performance evaluation measures. Additionally, some percentile points of distribution run length are also reported for a better understanding of run‐length distribution. Corrected design constants of the proposed charts are also provided for mentioned distributions. Implementation of the proposed schemes is illustrated by providing examples related to real practice.