階段二多變量移動加權平均監控相關誤差的非線性 隨機效用剖面

Abstract

在現在的商業、化學、工業等，使用控制圖去監控產品的品質是很常見的。藉由把產品的特徵給特徵化，已便去監控製成。在文獻中，剖面內的變異常常假設獨立， 然而這個假設在很多情況下是不適合的，因此階段二的多變量移動加權平均監控有相關誤差的非線性隨機效用剖面。一筆真實資料在Walker 跟 Wright (2002)被用來說明，並且做一些模擬說明我們提出的方法的實用性。

The technique of control charts for monitoring the quality of products is used in the modern business, chemistry, manufacturing, etc. It characterizes one or more quality characteristics of products to monitor a process. In the literature, it is commonly assumed that observations are independent within each profile; however, this assumption is inappropriate in many actual situations and thus a Phase II multivariate exponentially weighted moving average monitoring scheme for nonlinear random-effects profiles with correlated errors is proposed in the paper. A real data set in Walker and Wright (2002) and a simulation study are utilized to illustrate the usefulness and applicability of the proposed methodology.