A Comparative Study Of Memory-type Control Charts Based On Robust Parameter Estimators

In the traditional quality control process,Cumulative sum(CUSUM)and exponen-tially weighted moving average(EWMA)charts are commonly used to detect moderate and small changes in the parameters of production processes.In order to improve the ability of control chart to detect shifts,recently,a new control structure was introduced,named as mixed EWMA-CUSUM control chart,which combines the advantages of a EW-MA chart with the CUSUM chart.In practice,process parameters are often unknown.In this paper,we investigate four different location estimators with the structures of these three memory type of control charts in phase 1 and phase 2 for monitoring the process dis-persion under normal and contaminated normal environments.The control limits based on Monte-Carlo simulation are provided in a table.The in-control and out-of-control per-formance properties of the chart are investigated in simulation studies in terms of some percentiles of the length distribution,average run length and standard deviation of run length.It turns out that none of the estimators have good performance in each case,when the phase 1 data are diffuse asymmetric variance disturbances,the control charts have the worse performance,and the charts based on median-mean perform better.On the other case,the charts based on mean-mean perform better.On the whole,Mixed EWMA-CUSUM control chart has the best ability to detect shifts.