EWNA charts for monitoring the mean and the autocovariances of stationary processes |
| |
Authors: | M Roso?owski W Schmid |
| |
Institution: | (1) Department of Statistics, Europe-University, 15207 Frankfurt (Oder), Germany |
| |
Abstract: | In this paper various types of EWMA control charts are introduced for the simultaneous monitoring of the mean and the autocovariances.
The target process is assumed to be a stationary process up to fourth-order or an ARMA process with heavy tailed innovations.
The case of a Gaussian process is included in our results as well.
The charts are compared within a simulation study. As a measure of the performance the average run length is taken. The target
process is an ARMA (1,1) process with Student-t distributed innovations. The behavior of the charts is analyzed with respect to several out-of-control models. The best design
parameters are determined for each chart. Our comparisons show that the multivariate EWMA chart applied to the residuals has
the best overall performance. |
| |
Keywords: | Statistical process control EWMA charts simultaneous control charts time series analysis |
本文献已被 SpringerLink 等数据库收录! |