整合獨立成份影像重建技術與分類迴歸樹在製程監控上之應用

Process Monitoring with ICA-based Image Reconstruction Scheme and CART Approach

黃獻平(Shien-Ping Huang)

《台北海洋技術學院學報》 2卷1期 (2009/03) Pp. 59-75

https://doi.org/10.29770/JTCMT.200903.0005

摘要

生產製程的監測與管制一直是實務上常被用來於維持高品質產品的有效方法。然而，品質管制中的統計製程管制(statistical Process Control, SPC)及工程製程管制(Engineering Process Control, EPC)技術，卻又常無怯有效的偵測出製程的異常狀態，特別是在當製程數據具有某種程度的自我相關性質時。在本論文中，一個結合ICA影像重建機制與分類迴歸樹(Classification and Regression Tree, CART)的方法被提出來進行製程干擾項的辨識工作。研究結果顯示在辨識不同型態的干擾資料時，ICA影像重建技術可以成功的將干擾資訊明顯化。為了驗證在本研究中所提整合方法的有效性，我們針對兩種不同類型的常見干擾：平移式干擾(step-change disturbance)與線性式干擾(linear disturbance)進行驗證與測試。此外，我們也運用了傳統的蕭華特(Shewhart)管制圖與CUSUM管制圖來進行結果的比較。根據研究結果顯示，ICA影像重建技術與分類迴歸樹方法的應用可以成功的辨識不同類型的製程干擾。此外，論文中所提方法的辨識成功率，也較不會受到資料中相關性高低的影響。

關鍵字

統計製程管制 ； 工程製程管制 ； 獨立成份分析 ； 分類迴歸樹

並列摘要

Monitoring the producing process is always used as the most efficient way to maintain the high quality products. However, based on the Statistical Process Control, SPC, and Engineering Process Control, EPC technique can not quite accurately detect all exceptional situations; especially, when the producing process data have certain level self-related quality. In this paper, a integration of ICA image reconstruction and Classification and Regression Tree, CART, was addressed as the solution for identity of the unusual factors in producing process.
The result of this research showed ICA image reconstruction can distinct the inferences successfully when identifying different inference sources. The two different common inferences, step-change disturbance and linear disturbance, were aimed and tested as main factors. Furthermore, Shewhart restrain sketch and CUSUM restrain sketch were the methods used as gathering data in order to compare. Moreover, the identification rate mentioned in this paper will not be affected by the related date as well.

並列關鍵字

Statistical Process Control SPC ； Engineering Process Control EPC ； Independent Component Analysis ICA ； Classification and Regression Tree CART

參考文獻

Anderson, J. A.,E. Rosenfeld(1998).Neurocomputing: Foundations of research.MA MIT Press.

B. Ripley.(1994).Neural networks and related methods for classification.J. R. Statist. Soc. B. 2000.56,409-456.

Ben-Gal, I.,Morag G.,Shmilovici A.(2003).CSPC: A monitoring procedure for state dependent processes.Technornetrics.45,293-311.

Box, G. E. P.,Jenkins, G. M.,Reinsel, G. C.(1992).Time Series Analysis. Forecasting and Control.Englewood Cliffs. NJ:Prentice-Hall.

Box, G. E. P.,Kramer, T.(1992).Statistical process monitoring and feedback adjustment-a discussion.Technometrics.34,251-285.