應用小波分析法與類神經網路建構管制圖非隨機樣式之辨識系統

Control Chart Patterns Recognition using Wavelet Transfer and Neural Networks

黃啟原(Chi-Yuan Huang)

指導教授 ： 鄭春生

元智大學/工程學院/工業工程與管理學系/碩士(2006年)

摘要

管制圖是統計製程管制 (Statistical Process Control; SPC) 中之主要工具，它可用來判斷製程是否存在可歸屬原因所造成之變異。當製程存在可歸屬原因時，管制圖會呈現特定的非隨機性樣式變化，例如：趨勢樣式、偏移樣式、週期性樣式等。然而正確辨識出非隨機性樣式，可以縮小診斷製程可歸屬原因之範圍，有助於規劃改善對策。

本研究之主要目的為建立一個以類神經網路為基礎之辨識系統，用以偵測和辨識非隨機樣式的類型，以作為規劃製程矯正措施的依據。本研究首先探討如何利用Haar 離散小波多重解析擷取重要製程數據之特徵。製程數據經 Haar 離散小波轉換 (Discrete Wavelet Transfer; DWT) 後可獲得不同解析尺度下的係數，經由小波之數據轉換方式以呈現樣式的重要特徵。第二、本研究將建立一個監督式之類神經網路作為非隨機樣式的辨認系統。我們將考慮各類非隨機樣式同時存在與單一週期性樣式單獨發生之情形，評估原始數據和小波之特徵擷取方式，對於類神經網路之辨識績效的影響。

關鍵字

統計製程管制 ； 類神經網路 ； Haar 離散小波轉換 ； 特徵值擷取

並列摘要

Control chart pattern recognition is an important work in statistical process control. A control chart can present many unnatural patterns: trends, sudden shifts, and cycles. The presence of unnatural patterns indicates that a process is affected by assignable causes, and corrective actions should be taken. This research will develop a neural network-based recognizer for control chart pattern recognition. First, we apply a multi-resolution analysis approach based on Haar discrete wavelet transfer (DWT) to extract distinguished features from raw data. The extracted features are used as the components of the input vectors. Secondly, we will develop a supervised neural network for control chart pattern recognition. In addition, we will focus on single cyclic pattern and concurrent patterns that can be characterized using this classifier. The performance of the neural network using features extracted from wavelet analysis as the components of the input vectors will be investigated and compared.

並列關鍵字

statistical process control ； neural network ； Haar discrete wavelet transfer ； extracted features

參考文獻

1.Al-Assaf Y., “Recognition of control chart patterns using multi-resolution wavelets analysis and neural networks,” Computers & Industrial Engineering, 47, 17-29 (2004).

2.Aradhye, H. B., Bakshi, B. R., Strauss, R. A., and Davis, J. F., “Multisacle SPC using wavelets-theoretical analysis and properties,” AICHE Journal, 49, 939-958 (2003).

3.Assaleh K., and Al-Assaf, Y., ”Features extraction and analysis for classifying causable patterns in control charts,” Computers & Industrial Engineering, 49, 168-181 (2005).

4.Bakshi, B. R., “Multiscale analysis and modeling using wavelets,” Journal of Chemometrics, 13, 415-434 (1999).

5.Cheng, C. S., “A neural network approach for the analysis of control chart patterns,” International Journal of Production Research, 35, 667-697 (1997).