在品質管理中，統計製程管理是最為常用的工具。當製程越趨複雜時，各個品質特性會產生交互作用，因此常使用多變量管制圖作為主要監控的手法。傳統上常見的多變量管制圖，如Hotelling^' s 〖 T〗^2管制圖，在遇到擁有自相關特性的資料時可能會產生嚴重的錯誤訊號，也就是說傳統的管制圖沒辦法有效的監控此類型的製程資料。有學者提出可以減少抽樣頻率、修改管制圖的上下界線、使用殘差管制圖等等方法來解決，但上述這些方法皆有其缺點，不是過程中會流失過多的資訊就是沒有過多的使用上假設。例如殘差管制圖需要有良好的且準確地預測模型才可使用。本研究使用深度學習方法建立一套監控自相關製程的系統，該系統可以監控目前製程是否發生異常，若是異常產生則會再進一步的通知使用者可歸屬原因為何，訓練好的模型即可取代殘差管制圖的使用。本研究方法首先將時間序列資料轉換成AttLSTM-M所需的輸入型態並輸入，利用AttLSTM-M找出該筆資料視窗是否擁有發生異常的資料存在，若偵測到異常則將視窗資料投入AttLSTM-R以取得偏移資訊。在AttLSTM-M中本研究將會使用〖ARL〗_1作為指標，並與各種不同的深度學習方法比較，以期望本研究所提出的方法可以取得最佳的〖ARL〗_1結果。AttLSTM-R本研究將會使用深度學習中分類問題上常使用的分類指標為主，同樣與其他深度學習方法比較。結果顯示添加了注意力層的深度學習模型應用在自相關製程其偵測及辨識能力皆有提升，雖然在辨識的準確度仍有增長空間，但在詳細的辨識結果中顯示出其辨識結果的方向仍為正確，也就是能夠告知人員哪一個品質特性發生偏移但偏移幅度辨識錯誤，因此還是能夠使人員知道問題發生並問題排除。

In quality mangement, statistical process control (SPC) is one of the most effective ways to monitor processes. Currently, manufacturing processes are perceieved to be multivariate and autocorrelated. Thus, variables related to quality characteristics have to be taken into account simultaneously. If we use a multivariate control chart in the autocorrelation process a false alarm will occur. If so, we still can not diagnose the assignable cause when out-of-control signals occur. In this paper addresses the problem of multivariate autocorrelated processes is addressed, and the possible cause of an out-of-control signal is determined. To achieve this goal using, two deep learning models and two stages are used. First a model for the monitor satge is used to detect whether an abnormal shift occured or not, and a deep learning model is then designed called the Attention LSTM Monitor model (AttLSTM-M). If AttLSTM-M detects an abnormal shift , then this data is input into the next stage, which is the recognition stage. In the recognition satage, we design a deep learning model that is called the Attention LSTM Recognition model (AttLSTM-R) to determine which quality characteristic caused the shift and to determined its magnitude. Fianlly, practitioners can adjust the parameters and prevent defective products from being consistently produced. An out-of-control average run length and correct classification ratio were chosen to evaluate the performance of the designed models. In the comparison stage, LSTM, DNN and 1D-CNN were chosen to compare with AttLSTM-M and the AttLSTM-R model using 〖ARL〗_1and accuracy as indicator. The result of the experiments show that AttLSTM-M has the lowest 〖ARL〗_1 and that AttLSTM-R has the highest accuracy among these deep learning models.

[1] A. Kalgonda and S. Kulkarni, "Multivariate quality control chart for autocorrelated processes," Journal of Applied Statistics, vol. 31, no. 3, pp. 317-327, 2004.
[2] L.-H. Chen and T.-Y. Wang, "Artificial neural networks to classify mean shifts from multivariate χ2 chart signals," Computers & Industrial Engineering, vol. 47, no. 2-3, pp. 195-205, 2004.
[3] Y. Li, E. Pan, and Y. Xiao, "On autoregressive model selection for the exponentially weighted moving average control chart of residuals in monitoring the mean of autocorrelated processes," Quality and Reliability Engineering International, vol. 36, no. 7, pp.2351-2369, 2020.
[4] A. J. Hayter and K.-L. Tsui, "Identification and quantification in multivariate quality control problems," Journal of Quality Technology, vol. 26, no. 3, pp. 197-208, 1994.
[5] D. C. Montgomery, Introduction to statistical quality control. John Wiley & Sons, 2007.
[6] C. A. Lowry and D. C. Montgomery, "A review of multivariate control charts," IIE transactions, vol. 27, no. 6, pp. 800-810, 1995.
[7] D. C. Montgomery, Statistical quality control. Wiley Global Education, 2012.
[8] W. H. Woodall and M. M. Ncube, "Multivariate CUSUM quality-control procedures," Technometrics, vol. 27, no. 3, pp. 285-292, 1985.
[9] R. B. Crosier, "Multivariate generalizations of cumulative sum quality-control schemes," Technometrics, vol. 30, no. 3, pp. 291-303, 1988.
[10] S. Chakraborti, "Run length, average run length and false alarm rate of Shewhart X-bar chart: exact derivations by conditioning," Communications in Statistics-Simulation and Computation, vol. 29, no. 1, pp. 61-81, 2000.
[11] C. A. Lowry, W. H. Woodall, C. W. Champ, and S. E. Rigdon, "A multivariate exponentially weighted moving average control chart," Technometrics, vol. 34, no. 1, pp. 46-53, 1992.
[12] E. Vanhatalo and M. Kulahci, "The effect of autocorrelation on the Hotelling T2 control chart," Quality and Reliability Engineering International, vol. 31, no. 8, pp. 1779-1796, 2015.
[13] L. C. Alwan and H. V. Roberts, "Time-series modeling for statistical process control," Journal of Business & Economic Statistics, vol. 6, no. 1, pp. 87-95, 1988.
[14] E. Vanhatalo and M. Kulahci, "Impact of autocorrelation on principal components and their use in statistical process control," Quality and Reliability Engineering International, vol. 32, no. 4, pp. 1483-1500, 2016.
[15] D. E. Rumelhart, G. E. Hinton, and R. J. Williams, "Learning representations by back-propagating errors," nature, vol. 323, no. 6088, pp. 533-536, 1986.
[16] V. Vapnik, The nature of statistical learning theory. Springer Science & Business Media, 2013.
[17] S. Kumar, Neural networks: a classroom approach. Tata McGraw-Hill Education, 2004.
[18] G. E. Hinton, S. Osindero, and Y.-W. Teh, "A fast learning algorithm for deep belief nets," Neural Computation, vol. 18, no. 7, pp. 1527-1554, 2006.
[19] H. I. Fawaz, G. Forestier, J. Weber, L. Idoumghar, and P.-A. Muller, "Transfer learning for time series classification," in 2018 IEEE International Conference on Big Data (Big Data), 2018: IEEE, pp. 1367-1376.
[20] Y. LeCun, L. Bottou, Y. Bengio, and P. Haffner, "Gradient-based learning applied to document recognition," Proceedings of the IEEE, vol. 86, no. 11, pp. 2278-2324, 1998.
[21] S. Hochreiter and J. Schmidhuber, "Long short-term memory," Neural Computation, vol. 9, no. 8, pp. 1735-1780, 1997.
[22] J. Cheng, L. Dong, and M. Lapata, "Long short-term memory-networks for machine reading," arXiv preprint arXiv:1601.06733, 2016.
[23] D. Bahdanau, K. Cho, and Y. Bengio, "Neural machine translation by jointly learning to align and translate," arXiv preprint arXiv:1409.0473, 2014.
[24] B. K. Issam and L. Mohamed, "Support vector regression based residual MCUSUM control chart for autocorrelated process," Applied Mathematics And Computation, vol. 201, no. 1-2, pp. 565-574, 2008.
[25] M. E. Camargo, W. Priesnitz Filho, S. L. Russo, and A. I. dos Santos Dullius, "Control charts for monitoring autocorrelated processes based on Neural Networks Model," in 2009 International Conference on Computers & Industrial Engineering, IEEE, 2009. pp. 1881-1884.
[26] K. Harris, K. Triantafyllopoulos, E. Stillman, and T. McLeay, "A multivariate control chart for autocorrelated tool wear processes," Quality and Reliability Engineering International, vol. 32, no. 6, pp. 2093-2106, 2016.
[27] D. F. Cook and C.-C. Chiu, "Using radial basis function neural networks to recognize shifts in correlated manufacturing process parameters," IIE transactions, vol. 30, no. 3, pp. 227-234, 1998.
[28] R. Noorossana, M. Farrokhi, and A. Saghaei, "Using neural networks to detect and classify out‐of‐control signals in autocorrelated processes," Quality And Reliability Engineering International, vol. 19, no. 6, pp. 493-504, 2003.
[29] H. Brian Hwarng and Y. Wang, "Shift detection and source identification in multivariate autocorrelated processes," International Journal of Production Research, vol. 48, no. 3, pp. 835-859, 2010.
[30] A. Fountoulaki, N. Karacapilidis, and M. Manatakis, "Using neural networks for mean shift identification and magnitude of bivariate autocorrelated processes," International Journal of Quality Engineering and Technology, vol. 2, no. 2, pp. 114-128, 2011.
[31] C. C. Chiu, M.-K. Chen, and K.-M. Lee, "Shifts recognition in correlated process data using a neural network," International Journal of Systems Science, vol. 32, no. 2, pp. 137-143, 2001.
[32] G. E. Box, G. M. Jenkins, and G. C. Reinsel, Time series analysis: forecasting and control. John Wiley & Sons, 2011.
[33] C. Cheng, "Detecting changes in the process mean using artificial neural networks approach," Journal of Chinese Institute of Industrial Engineers, vol. 11, no. 1, pp. 47-54, 1994.