Abstract:
|
Control charts are widely used in industry to monitor process data and maintain quality levels. Here, a new EWMA-type control chart for continuous data is introduced. This control scheme is designed for instances when process data parameters are unknown and utilizes the exceedance statistic. Further, this chart maintains the same run length distribution regardless of the distribution of the process data. Unlike the EWMA exceedance chart, this new scheme has a chart statistic that only takes integer values. Compared to both the EWMA and CUSUM exceedance charts, it requires one less design parameter choice for the practitioner to evaluate the distribution of the chart's run length. Here, an algorithm and equations are provided to evaluate unconditional mean and standard deviation run length. In control and out of control chart performance is considered with comparisons made to the EWMA Exceedance chart. Evaluation of the chart reveals that parameters may be tailored to detect shifts of a larger or smaller magnitude.
|