Distribution-free triple EWMA control chart for monitoring the process location using the Wilcoxon rank-sum statistic with fast initial response feature |
| |
Authors: | Tokelo Irene Letshedi Jean-Claude Malela-Majika Philippe Castagliola Sandile Charles Shongwe |
| |
Affiliation: | 1. Department of Statistics, College of Science, Engineering and Technology, University of South Africa, Pretoria, South Africa;2. Université de Nantes & LS2N UMR CNRS 6004, Nantes, France |
| |
Abstract: | The exponentially weighted moving average (EWMA) control chart is a memory-type chart known to be more efficient in detecting small and moderate shifts in the process parameter. The double EWMA (DEWMA) chart is an extension of the EWMA chart that is more effective than the latter in the detection of small-to-moderate shifts. This paper proposes a new distribution-free (or nonparametric) triple EWMA (TEWMA) control chart based on the Wilcoxon rank-sum (W) statistic to improve the detection ability in the process location parameter. Moreover, a new fast initial response (FIR) feature is added to further improve the sensitivity of the new TEWMA chart. The performance of the proposed TEWMA chart with and without FIR features is compared to those of the existing EWMA and DEWMA W charts. It is observed that the TEWMA chart with and without FIR features is superior to the competing charts in most situations. A real-life illustration is provided to show the application and implementation of the new chart. |
| |
Keywords: | double EWMA chart EWMA chart fast initial response nonparametric chart overall performance run-length triple EWMA |
|