No label defined (Q4373573)

In this paper, a method to monitor an autocorrelated process is discussed, for the univariate as well as the multivariate case. The method was introduced by the author [Statistical control charts: Performance of Shewhart and CUSUM charts. Ph.D. Thesis No. 63, Inst. Math. Stat. Oper. Res., Tech. Univ. Denmark (1992)]. The method extends the methods suggested by \textit{L. C. Alwan} and \textit{H. V. Roberts} [J. Bus. Econ. Stat. 6, No. 1, 87-95 (1988)] and \textit{D. C. Montgomery} and \textit{C. M. Mastrangelo} [J. Qual. Technol. 23, No. 3, 179-204 (1991)] in accounting for measurement error. The method builds on the well-known Kalman-filtering technique and standard control charts. We only consider observations from normal processes. In a case study where the distribution of process output from a filling head is being monitored, an AR(1)-model is considered. A MAR(1)-model is considered in a case study, where the distribution of process output from two filling heads on the same filling machine are being monitored simultaneously. Finally the extension to general models in the (M)ARIMA-class is indicated. For the entire collection see [Zbl 0867.00026].