製程改善能力是當前 TFT-LCD 製造商競爭力的決定性因素之一，但直到現今，還沒有任何適當的理論被提出用以改善 TFT-LCD 工業的良率問題，然而經驗證從良率模式所獲得的資訊（例如，domain knowledge 或parameter effect）對於 TFT-LCD 的製造商實能夠提供有用的建議和改善方案，也就是說，良率模式的建構與製程參數的影嚮性，對於 TFT-LCD 產業的良率分析而言，將會是一個必需被重視的課題。 在此篇論文中，我們提出了結合類神經網路與迴歸分析之技術以達成良率模式的建構，並在實例說明中，套用一臺灣臺南科學園區的TFT-LCD 製造商的實際生產資料，用以驗證我們所提出的模式。 The ability to improve yield in manufacturing process is an important competitiveness determinant for TFT-LCD factories. Until now, no any suitable theories were proposed to address the yield problem in TFT-LCD industry. However, the information (e.g. the domain knowledge or the parameter effect) obtained from the yield model will provide useful recommendations and improvements to those manufacturers. That is, the model construction and parameter effect for yield analysis will be a necessary issue to be addressed. In this study, we proposed a procedure incorporating the artificial neural networks (ANNs) and stepwise regression techniques to achieve the model construction and parameter effect. Besides, an illustrative case owing to TFT-LCD manufacturer at Tainan Science Park in Taiwan will be applied to verifying our proposed procedure.
