品質管制系利用統計手法為工具,藉著資料的收集、整理加以研判及分析 ,尋找出製程異常原因並加以改善,使產品的品質能夠符合我們所需的特 性。而統計製程管制是為品質中一項重要的技術,其主要的目的在儘快偵 測出可歸屬原因或製程之變動,以使在更多不良品製造出來之前能發現製 程之異常並進行改善。統計製程管制主要是利用管制圖來偵測製程是否有 可歸屬之異常原原及是否在管制狀態內。本研究是用類神經網路之技術, 發展一套偵測製程平均值變化之程序。我們同利用實驗設計之方法探討各 項參數對類神經網路成效之影響。在本研究中,我們是以平均連串長度來 衡量及比較類神經網路之成效。經深入之比較,本研究所提出之類神經網 路對於微量平均值變化之偵測,較修華特累和管制法更具有效率。
Control charts are important tool in statistical process cont- rol (SPC). They are useful in determining whether a process is bshaving as intended or if there are some unnatural causes of variation. In this paper we present an alternative approach to traditional control charts using artificial neural network technology and compare its performance with that of the combi- ned Shewhart-CUSUM schemes. The neural network developed here is concerned with the detection of sudden shifts in the process mean. In this study, we also conducted an experimental design to study the effect of various design parameters on the network performance. The performance of the neural network was evaluate by estimat- ing the average run lengths(ARL''s) using simulation. An exten- sive comparison shows that the proposed approach has 20-40% faster detection of small process changes than combined Shewh- art-CUSUM control schemes. The neural network approach presented in this paper appears to offer a competitive alternative to existing control schemes.