Nonparametric Hampel Score‐Function‐Based Adaptive EWMA Charts for Monitoring Location of Manufacturing Industrial Process

Abstract

Control charts are applied to monitor changes in the process parameter(s). In the manufacturing industry, shifts in the process parameter(s) are generally unknown so the conventional monitoring techniques only perform well for the predefined shifts. To tackle the unknown range of shifts in the process parameter(s), adaptive charting techniques are preferred. The Hampel score‐function‐based adaptive exponentially weighted moving average ( chart has been recently investigated under the assumption of the normality of the process. However, in industrial processes, there is a lack of knowledge about the process distribution; in such cases, nonparametric charts are the better choice for practitioners. This study proposes nonparametric Hampel function‐based adaptive exponentially weighted moving average sign (NPHAEWMA‐SN) and Wilcoxon signed‐rank (NPHAEWMA‐SR) charts as alternatives to the chart for monitoring unknown changes in the location of the process. The proposed charts (NPHAEWMA‐SN and NPHAEWMA‐SR) proved robust and efficient alternatives to the counterparts against the symmetrical heavy‐tailed, extreme value, and contaminated normal distributions at certain and over the range of shifts. Three artificial datasets have been taken from different distributions to verify the robustness and detection ability of the proposals. An industrial dataset has also been taken from the piston rings manufacturing industry for the application of the proposals.