頁籤選單縮合
題 名 | Alpha Mistake of Taguchi's Robust Design with L□ and L□ Arrays for STB Type QCH by Simulation=利用模擬方法探討田口方法L□與L□直交表望小型品質特性值之型I錯誤的機率 |
---|---|
作 者 | 雷方奕; 李明賢; | 書刊名 | 品質學報 |
卷 期 | 18:1 2011.02[民100.02] |
頁 次 | 頁35-48 |
分類號 | 494.56 |
關鍵詞 | 型I錯誤的機率; 田口方法; 望小型; 模擬; Alpha mistake; Taguchi method; Smaller-the-better; Simulation; |
語 文 | 英文(English) |
中文摘要 | 摘 要 田口方法廣泛地應用於穩健設計,然而其統計方法卻遭到諸多的爭議與討論。本研究針 對L12(211) 和L32(231) 直交表,利用模擬方法探討田口方法對望小型產品質特性值參數設計 時,將不顯著因子誤判為顯著因子之型I 錯誤的機率。假設所有的反應變數均為常態分佈且 有相同的平均值與標準差,因此,所有的因子均不顯著之虛無假設為真。然而模擬的結果顯 示這兩種直交表之型I 錯誤的機率都很高。實務上,這樣的誤判將對不相關或不重要的因子 造成不必要的調查與推測。最後,田口品質工程的觀念雖然重要,然而,其統計手法包括信 號雜訊比、向上合併誤差 (pooling-up) 方法卻提供了高風險的參數設計。 |
英文摘要 | Abstract The Taguchi method has been widely used for parameter design. Nevertheless, its statistical techniques have been the subject of much debate and discussion among statisticians. This research investigates the alpha mistake of Taguchi method, or the probability of identifying insignificant factors as significant, with L12 (211) and L32 (231) arrays for the smaller-the-better (STB) type quality characteristic (QCH) using simulation. It is assumed that all response values are normally distributed with the same mean and standard deviation. The null hypothesis that all factors are insignificant is therefore true. Simulation results, however, showed high alpha mistake with both arrays. In reality, such mistake shall initiate unnecessary investigating and reasoning of helpless or unimportant factors. In conclusion, Taguchi’s quality engineering concepts are of main importance. However, his statistical tools, including signal-to-noise ratio, pooling-up technique, provide risky parameter design. |
本系統中英文摘要資訊取自各篇刊載內容。