當製程存在可歸屬原因,產品的品質特性可能產生某種變化,使得製 程處於管制外, 此時我們希望能儘快找出造成製程異常的原因進而加以 改善,而瞭解製程發生何種異常變 化將有助於尋找可歸屬原因。在傳統 的管制程序中,當管制對象為個別值時,通常是以個 別值管制圖或移動 全距管制圖分別來監視製程平均值及變異數的變化。過去的研究結果發 現,一般的統計製程管制程序並無法有效地分辨製程的變異是來自於何種 品質特性的改變 。 本研究是以類神經網路(artificial neural networks)為基礎發展一辨認 製程發生平 均值、變異數或相關性變化之管制程序。類神經網路的成效 是以蒙地卡羅模擬法產生數據 來評估。本研究所採用之評估指標為平均 連串長度和正確辨認率。 本論文所提出之辨認系統經實驗證明,其偵測速度及正確辨認率均優於過 去之研究結 果,且具備了整體系統即時、連線的可行性。
The individual chart and moving range chart are often used in statistical process control to monitor the individual observations. However,previous research has shown that these charts can''t effectively identify the cause in the process change. In this research, a control procedure based on artificial neural network for distinguishing the changes in mean, variance and lag one autocorrelation was developed. The performance of the proposed neural network has been evaluated based on the average run length (ARL) and the recognition accuracy. Experimental results show that the proposed control procedure is better than existing methods in terms of recognition accuracy.