Effect of autocorrelation on the performance of EWMA chart

Abstract

The outputs from the manufacturing processes are often assumed to be independent but actually the observations are correlated and when this correlation builds-up automatically in the entire process, it is called autocorrelation. The performance of a chart is measured in terms of average run length (ARL). The performance of the traditional exponentially-weighted moving average (EWMA) chart is studied under the effect of the positive correlation. ARL at various levels of correlation (Ф), weightage factor (λ) and at various width of control limits (K), are studied using simulation with MATLAB software. Optimal schemes of EWMA chart are proposed for each level of correlation and showed better performance compared to EWMA chart, suggested by Zhang (2000). Moreover, optimal schemes of EWMA chart at given weighting factor, λ = 0.2, are very much comparable with the EWMA stationary chart, proposed by Winkel and Zhang (2004) at the various levels of the correlation.