使用相關係數及RBF類神經網路建立品管圖辨認系統

Control Chart Pattern Recognition Systems Using Correlation Coefficient and RBF Neural Networks

楊鎮槐(Jenn-Hwai Yang)

指導教授 ： 楊敏生

中原大學/理學院/應用數學系/博士(2006年)

https://doi.org/10.6840/cycu200600281

摘要

管制圖表中不尋常的圖形意指在統計製程管制中有一些不尋常的變異發生，這些圖形的產生通常反應了一些生產線上的異常，而這些異常必需要被移除。因此管制圖表的圖形辨識工作在品管中顯得格外重要。雖然圖形辨識技巧被廣範的應用在辨別管制圖表中不正常的圖形。但是一個完備的圖形辨認系統需要擁有二個能力，圖形辨別及參數估計。在這篇論文中，我們提出一個整合的管制圖圖形辨識系統，包括用相關係數的方法來處理圖形辨別，而且用輻射基底函數(RBF)類神經網路來做參數估計。而我們所提出的辨識系統可以完成前面所提的三個能力。我們同時也考慮了混合的圖形，其意義為兩種不正常的圖形可能同時出現在管制圖表。我們利用相關係數方法，在辨別單一或混合型的圖形都有很好的效果。RBF類神經網路可以被使用於找輸入與輸出資料間的關係，因此我們利用RBF類神網路來做不尋常圖形的參數估計。最後我們整合相關係數方法與RBF神經網路的管制圖圖形辨識系統可以做為一個即時的辨識系統。

關鍵字

圖形辨識 ； 類神經網路 ； 管制圖表

並列摘要

Abnormal patterns in control charts mean that there are some unnatural causes for variations in statistical process control (SPC) and need to be elimination. Hence control chart pattern recognition becomes more important in SPC. Although pattern recognition techniques have been widely applied to identify abnormal patterns in control charts, a complete pattern recognition system should have two abilities with pattern identification, parameter estimation. In this dissertation, we present an integrated control chart pattern recognition system which contains a correlation coefficient method for pattern identification and radial basis function (RBF) neural networks for the parameter estimation. The proposed control chart pattern recognition system can fulfill the previously mentioned two abilities. We consider with the concurrent patterns where two abnormal patterns may simultaneously occur in a control chart pattern. Our correlation coefficient method is used for the identification of abnormal control chart patterns with good performance for single and concurrent abnormal patterns. RBF neural networks can be used for constructing the relation among input-output data sets. Thus, RBF networks are adopted to accomplish the parameter estimation for abnormal patterns. Finally, the integrated control chart pattern recognition system with correlation coefficient method and RBF networks is proposed for real-time processing.

並列關鍵字

pattern recognition ； RBF ； control chart

參考文獻

Al-Ghanim, A.M., & Ludeman, L.C. (1997). Automated unnatural pattern recognition on control charts using correlation analysis techniques. Computers and Industrial Engineering, 32 (3), 679-690.

Behloul, F., Lelieveldt, B.P.F., Boudraa, A., & Reiber, J.H.C. (2002). Optimal design of radial basis function neural networks for fuzzy-rule extraction in high dimensional data. Pattern Recognition, 35 (3), 659-675.

Bishop, C.M. (1995). Neural Network for Pattern Recognition. New York: Oxford University Press.

Chen, S., Cowan, C.F.N., & Grant, P.M. (1991). Orthogonal least squares learning algorithm for radial basis function networks. IEEE Trans. Neural Networks, 2(2), 302-309.

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