頁籤選單縮合
題 名 | 整合平滑樣條法與決策樹於非線性剖面製程之研究=The Monitoring of Nonlinear Profiles Using Smoothing Spline and Decision Tree Model |
---|---|
作 者 | 李虹葶; 黃照雅; 鄭春生; | 書刊名 | 品質學報 |
卷 期 | 22:2 2015.04[民104.04] |
頁 次 | 頁77-87 |
分類號 | 494.56 |
關鍵詞 | 非線性剖面; 無母數迴歸; 平滑樣條法; 決策樹; Nonlinear profile; Non-parametric regression; Distance-based metrics; Decision tree; |
語 文 | 中文(Chinese) |
中文摘要 | 在傳統統計製程管制(statistical process control, SPC)之應用中,我們假設一個物件或製程之品質,可以由一個量測值或來自多變量分配之數個量測值來描述。但在許多實務應用中,我們需要監控一條由數個資料所構成之直線或曲線。這些直線或曲線被稱為剖面(profile)或函數。剖面資料可以利用一個線性或非線性之模型來表示。本研究之目的是建立監控非線性剖面製程之管制程序。此管制程序包含利用屬於無母數迴歸之平滑樣條法來建立參考剖面,接著再發展出以距離為基之特徵值。一個決策樹分類模型利用這些特徵值來進行剖面製程之監控。本研究所提出之管制程序是以彩色濾光片的銦鋅氧化物製程資料,來驗證其可行性和有效性。研究結果顯示,使用平滑樣條法可以有效地去除雜訊,其建立的平滑曲線可以作為參考剖面。深入的比較顯示,本研究所提出的特徵值,可以有效地提升決策樹之分類正確率,進而提升監控非線性剖面之效益。 |
英文摘要 | In traditional statistical process control (SPC) applications, it is assumed that the quality of a product or process can be characterized by a single measurement from a univariate distribution or multiple measurements from a multivariate distribution. However, in some practical applications, there is a demand in monitoring multiple measurements constituting a line or curve that is often referred to as a profile or function. Such profiles can be represented by a linear or nonlinear model. This paper focuses on the monitoring of nonlinear profiles. We propose using non-parametric regression method to construct a reference (baseline) profile. A set of relevant statistics based on distance-based metrics is used to construct a feature vector for a decision tree-based monitoring procedure. The implementation of the proposed approach is illustrated using the profile data obtained from industry. A comparative study shows that the proposed method is capable of detecting the changes in a profile. |
本系統中英文摘要資訊取自各篇刊載內容。