论文部分内容阅读
Since Lowry et al. proposed a multivariate version of the exponentially weighted moving average (EWMA) control chart, the multivariate EWMA control chart has become more and more popular in monitoring production processes, especially in chemical processes. A major advantage of multivariate EWMA statistics is that it is sensitive to small and moderate shifts in the mean vector. However, when a multivariate EWMA chart issues a signal, it is difficult to identify which variable or set of variables is out of control. In this paper, we introduce a new approach to diagnosing signals from a multivariate EWMA control chart. The implementation procedure is that when the multivariate EWMA control chart issues a signal, we adopt a univariate diagnostic procedure to identify the variables or/and the principal components that caused the signal.
Since Lowry et al. Proposed a multivariate version of the exponentially weighted average moving (EWMA) control chart, the multivariate EWMA control chart has become more and more in monitoring production processes, especially in chemical processes. A major advantage of multivariate EWMA statistics is that it is sensitive to small and moderate shifts in the mean vector. However, when a multivariate EWMA chart issues a signal, it is difficult to identify which variable or set of variables is out of control. In this paper, we introduce a new approach to diagnosing signals from a multivariate EWMA control chart. The implementation procedure is that when when multivariate EWMA control chart issues a signal, we adopt a univariate diagnostic procedure to identify the variables or / and the principal components that caused the signal.