標題: | 田口方法於線外品管多重品質特性製程最佳化之應用研究 Generalization of Taguchi Method for Multi-Response Process Optimization |
作者: | 王春和 Chung Ho Wang 唐麗英 Lee Ing Tong 工業工程與管理學系 |
關鍵字: | 田口方法;訊號雜訊比;穩健設計;製程最佳化;Taguchi Method ; Signal-to-Noise Ratio ; Robust Design ; Process Optimization |
公開日期: | 1992 |
摘要: | 田口博士提出損失函數(Loss Function)的觀念,透過改善品質來降低生 產成本,以達到產品之穩健性。傳統田口方法應用於製程之實例,於已發 表的文獻中,大都只探討單一品質特性最佳化的問題,對於製程中同時有 多個品質特性最佳化的問題,也只停留在以工程師經驗判斷的階段,而在 實務上產品往往是具有多個品質特性的,且這些品質特性損失間均可能有 極高的相關性。本研究乃是針對多重品質特性製程,提出一多重品質績效 衡量的統計量,它是將個別品質特性損失正常化,以消弭損失函數因品質 特性或衡量單位的不同所造成之差異,使各品質特性之損失能處在同一水 平線上來比較,同時依品質特性重要程度分別賦予權重作加總,並延續田 口 SN比的精神與二步驟最佳化的分析程序,達到多重品質特性製程之最 佳化。本研究以IC 製程中,電漿輔助化學氣象沉積(PECVD)SiNx的製程, 來驗證此方法之有效性。 Dr. Taguchi developed the concept of loss function and used the Signal-to-Noise(SN) ratio to evaluate the performance of product quality. By finding the optimal combination levels of factors in the parameter design, the performance of product qua- lity could be made insensitive (robust) to noise factor. Most application of taguchi methods to-date have only been concened with the optimization of a single-response process or product. Handling the more demanding multi-response cases,which require the simultaneous optimization of more than one responses in the same process, is hardly ever considered, the usual recom- mendation was "to be left to engineering judgement". In pratice, we often have to deal with multi- response process or product and it is usually to optimize such case just by engineering ex- perinces. the diffucility will become even more pronounced if these responses are highly correlated. This thesis therefore developed a technique generalized the ideas of Taguchi's SN ratio and the two-step optimization pro- cedure to optimized simultaneously the multi- response in a pro- cess. An experiment of plasma enhanced chemical vapor deposition (PECVD) process was conducted and verify the effectiveness of this new multi-response optimization technique. |
URI: | http://140.113.39.130/cdrfb3/record/nctu/#NT810030009 http://hdl.handle.net/11536/56588 |
顯示於類別: | 畢業論文 |