獨立成份分析訊號分離技術於相關性製程監控之應用

Using Independent Component Analysis in Monitoring Correlated Process

黃柏凱(Po-Kai Huang)

指導教授 ： 邱志洲 ； 呂奇傑

國立臺北科技大學/管理學院/商業自動化與管理研究所/碩士(2009年)

https://doi.org/10.6841/NTUT.2009.00602

摘要

品質往往是最能抓住消費者的心之最關鍵因素，而為了能夠讓產品有良好的品質，在生產的過程中必須有嚴格的控管機制來監控製程是否失控。統計製程管制(statistical process control, SPC)常被用於製程管制，其中管制圖(control charts)是最常被用於監控的工具。然而管制圖有一個限制前提，即製程樣本值必須符合統計獨立，但由於今日製程自動化的緣故，頻繁的抽樣使得觀察值之間產生自我相關性(autocorrelation)，直接使用管制圖進行監控會讓製程產生假警訊(false alarm)，造成訊號誤判，浪費不必要的成本。有鑑於此，本研究期望藉由獨立成份分析(independent component analysis, ICA)將相關性製程中的自我相關性分離，並有效凸顯出製程干擾項(disturbance)，成為相互獨立的獨立訊號(independent component, IC)，再結合傳統的蕭華特管制圖(Shewhart chart)對此獨立訊號進行監控，以協助品管人員得以迅速找到可歸屬原因(assignable causes)，進而改善製程之監控能力。

關鍵字

獨立成份分析 ； 製程干擾項 ； 自我相關性

並列摘要

Qulity is often the most important reason that affect consumer, in order to keep the superior quality of products, we must have strict control mechanism to monitor processes during the produce step. Statistical process control (SPC) is often used in monitoring process, control charts is the most popular tool in monitoring process of SPC. Although control charts have a constrain that every observation must observe identical independent, sampling is too frequently to create autocorrelation among observations. If using control charts in autocorrelaive observations directly will create false alarm and cause signal error, even wasting unnecessary cost. For this reason, the proposed thesis expect separate autocorrelation signal through independent component analysis (ICA) and present disturbance signal obviously, letting the each signal become a independent component (IC) and make ICs mutually independent. Finally, combining Shewhart control chart to monitor the IC that including disturbance and help quality managers can find assignable causes quickly and then improve monitoring ability in process.

並列關鍵字

Independent Component Analysis (ICA) ； disturbance ； autocorrelation

參考文獻

[1] 林品杰，應用獨立成份分析(ICA)濾波器於背光版與TFT-LCD面板之瑕疵檢測，私立元智大學工業與管理工程研究所碩士，2005。

[1] A. Hyvärinen, "Fast and robust fixed-point algorithms for independent component analysis, " IEEE Transactions on Neural Networks, Vol. 10, 1999a, pp. 626-634.

[3] A. Hyvärinen, "Sparse code shrinkage: denoising of nongaussian data by maximum likelihood estimation, " Neural Computation, Vol. 11, 1999b, pp. 1739-1768.

[5] A. Hyvärinen, J. Karhunen and E. Oja, Independent component analysis, New York: Wiley press, 2001.

[6] A. Hyvärinen and E. Oja, "Independent component analysis: Algorithms and applications," Neural Networks, Vol. 13, 2000, pp.411-430.