當一個製程具有多個相關之品質特性需要管制時,我們稱其為多變量品質 管制問題。 由於先進之自動化量測設備可以同時在一件產品上量測多項 品質特性,多變量品質管制已成為一項重要之問題。在過去已有許多學者 提出以統計原理為基礎之管制程序,用於管制具有多項品質特性之製程 。 本研究是利用類神經網路 (artificial neural networks) 技術 ,發展用於管制具有多變量之製程, 並以偵測微量製程平均值移動為研 究重點。本研究以平均連串長度來評估類神經網路之偵測能力。模擬分析 之結果說明本研究所提出之類神經網路,在偵測製程平均值變化上,優於 傳統之管制程序。
Quality control problems in which several correlated quality characteristics are of interest are often referred to to as multivariate quality-control problems. This subject is particularly important today, as automatic inspection procedures make it relatively easy to measure many parameters on each unit of product manufactured. Various types of multivariate quality-control charts have been proposed to take advantage of the relationships among the variables. In this research we propose a neural network- based quality control procedure as an alternative means to traditional control charts. The primary focus is on the detection of changes in the process mean. The neural network performance is evaluated based on average run length. An extensive simulation study indicates that the proposed neural network approach is better than traditional control procedures.