Robust Control Charting: A More Practical Way of Monitoring Statistical Processes

In recent times, the trending modification of control chart is a result of its effective application in various aspects of life. In this academic proposal, we intend propose some new charting scheme by dualizing the effects of cumulative sum and exponentially weighted moving average control charts. The purpose is to monitor the location parameter by enhancing the sensitivity of memory charts. The driving goal which inspired these schemes was enhancement of the sensitivity of the control scheme used by analyzers in the petro-chemical industry that identify different quantities of components present in different catalysts. The proposed schemes will be the combination of a Dual-CUSUM chart or a time-varying Dual-EWMA chart or both. We will also explore these schemes with robust estimators by evaluating their thresholds in the presence of disturbances and contaminations. This will be done by introducing both location and variance contaminants into the process environment. The average and standard deviation of the run lengths are to be used as the benchmark for performance evaluation through the various estimators. Finally, the proposed schemes will be applied on real life datasets from refinery laboratory statistics which serves as the case study.