The exact statistical properties of a signed‐rank‐based adaptive exponential weighted moving average chart.

The adaptive exponential weighted moving average (AEWMA) monitoring technique has recently received increasing attention as a robust alternative requiring no knowledge of the actual shift magnitudes. In this paper, we develop a discrete state nonparametric AEWMA chart based on the Wilcoxon signed‐rank statistics. The major characteristic of this monitoring scheme is that it takes into account the discrete nature in terms of the statistics being used as well as the chart structure design. Moreover, an accurate evaluation of the chart's statistical properties is provided using an appropriate Markov chain approach. Simulation results show that this AEWMA signed‐rank scheme performs robustly across a diverse variety of distributions, and it can also provide an effective performance for detecting both small and large shifts. Finally, a real data example is cited for showing practical implications of our monitoring scheme. [ABSTRACT FROM AUTHOR]