An Efficient Mixed-Memory-Type Control Chart for Normal and Non-Normal Processes

Abstract

Statistical process control techniques are commonly used to monitor process performance. Control charting technique is the most sophisticated tool of SPC and is categorized as memory-less and memory-type control charts. Shewhart-type control charts have low efficiency in detecting the small changes in the process parameters and named as memory-less control charts, and memory-type control charts (for example cumulative sum (CUSUM) and exponentially weighted moving average (EWMA) charts) are very sensitive to small persistent shifts. In connection with enhancing the performance of CUSUM and EWMA charts, an efficient variant of memory-type charts for the location parameter is developed based on mixing the double exponentially weighted moving average (DEWMA) chart and CUSUM chart by performing exponential smoothing twice. Performance of the proposed efficient variant is compared with existing counterparts under normal and non-normal (heavy tails and skewed) environments. The study also provides an industrial application related to the monitoring of weights of quarters made by mint machine placed into service at U.S. Mint. From theoretical and numerical studies, it is revealed that proposed variant of memory-type charts outperforms the counterparts in detecting shifts of small and moderate magnitude.