| Summary: | 博士 === 國立臺灣科技大學 === 工業管理系 === 100 === To detect small shifts in the process mean or variability as early as possible, Sheu and Griffith, Sheu and Lin, and Sheu and Tai developed and applied the generally weighted moving average (GWMA) control chart. The addition of an adjustment parameter makes the GWMA control chart more sensitive than the exponentially weighted moving average (EWMA) control chart in detecting small process shifts.
However, recently, considerable attention has been drawn toward methods that use a single chart for monitoring the process mean and variability. The main problem addressed in this study is to develop the new control chart by GWMA techniques for monitoring the process mean and variability simultaneously. Two new control charts, called the maximum generally weighted moving average (MaxGWMA) control chart and maximum chi-square generally weighted moving average (MCSGWMA) control chart, are proposed to achieve this goal. They can effectively combine two generally weighted moving average (GWMA) control charts into a single one and can detect both increases as well as decreases in the process mean and/or variability simultaneously.
This thesis is divided into two major parts.
Firstly, we propose the MaxGWMA control chart to simultaneously detect both increases and decreases in the mean and/or variability of a process. Simulations are performed to evaluate the average run length, standard deviation of the run length, and diagnostic abilities of the MaxGWMA and maximum exponentially weighted moving average (MaxEWMA) charts. An extensive comparison reveals that the MaxGWMA control chart is more sensitive than the MaxEWMA control chart.
Secondly, a new control chart called the maximum chi-square generally weighted moving average (MCSGWMA) control chart is developed. This control chart can effectively combine two generally weighted moving average (GWMA) control charts into a single one and can detect both increases as well as decreases in the process mean and/or variability simultaneously. The average run length (ARL) characteristics of the MCSGWMA and maximum exponentially weighted moving average (MaxEWMA) charts are evaluated by performing computer simulations. The comparison of the ARLs shows that the MCSGWMA control chart performs better than the MaxEWMA control chart.
|
|---|
