在機械加工產業中，帶鋸床為金屬原料之前端設備，其將金屬原料切割成所需之形狀與大小以利後續加工。精密加工是機械產業的重要目標，若能夠在製程的前端掌握住加工之精度，除了能減少二次加工所需的時間與成本，也能確保最終產品品質。鋸帶的磨耗性能衰退是造成鋸床加工表面品質下降的主因，以往的檢測方法為等到機台發出明顯的振動或噪音時，才將機台停機進行加工品質檢查，但是此種方法通常發現得太晚。若能在機台出現輕微異常時就立即發出警報訊息，通知操作人員處理，不僅能確保產品品質，也能提高生產效率。
基於上述，本研究利用深度學習(Deep learning)方法分析帶鋸床運作之振動訊號與加工表面影像，並將結果整合至雲端，建立一個雲端智慧化診斷系統，目的在於不再仰賴人為的判讀而是透過建立智慧化的診斷流程來提高生產效率與品質。本研究使用帶鋸床切削S45C圓棒，總共切削1000片工件。接著分別使用有序神經元長短期記憶編碼器(ON-LSTM-ED)與自編碼器(Autoencoder)分析振動訊號與加工表面影像，得到振動訊號健康指數與加工表面健康指數。發現鋸帶的磨耗衰退行為主要會反應在側向振動(Lateral vibration)上，側向振動增大使得鋸帶無法完整的將材料從工件表面移除，因而開始有材料殘留在工件表面，使加工品質下降。分析結果顯示振動訊號健康指數對於有殘留材料在工件表面之情形可以即時反應出來，而加工表面健康指數則等到殘留材料有一定的量後才有明顯下降趨勢。因此，本研究將振動訊號健康指數之分析結果整合至雲端智慧化診斷系統，利用方程式對振動訊號健康指數作曲線擬合，並透過擬合之係數與振動訊號健康指數之變化趨勢作對應。在振動訊號健康指數有明顯下降趨勢時，擬合之係數會跟著劇烈上升，透過此特性作為智慧化診斷系統是否要發出警報的判斷依據。結果顯示本研究之雲端智慧化診斷系統可以在鋸帶有輕微異常時，即時的發出警報，將來能應用於真實的加工情況中。

Bandsaw machine is the front equipment of metal cutting process, which cut various materials into required dimensions. In order to lower the cost and prevent the waste of cutting materials, it is important to diagnose the working state of the bandsaw machine. The deterioration of blade is the main reason which causes low quality of the machined surface. Therefore, this study focuses on intelligent diagnosis for the wear degradation assessment of bandsaw blade by using deep learning methods and then establish an intelligent diagnosis cloud system. This paper uses two deep learning models named Ordered Neurons Long Short-Term Memory Network Encoder-Decoder (ON-LSTM-ED) and autoencoder to respectively process vibrational signals and machined surface images. With the data analysis ability of these two models, we can obtain vibrational healthy index
and machined surface healthy index which show the degradation of the blade and surface quality, respectively. We find that the wear degradation of the blade will cause higher lateral vibration on the blade. Because of large lateral vibration, the blade cannot completely remove the materials on the specimen, which lower the quality of the machined surface. The results show that vibrational healthy index has an earlier decline comparing to machined surface healthy index when lateral vibration of the blade becomes large. Thus, we establish an intelligent diagnosis cloud system utilizing the analysis processes of vibrational signals. The system quantifies the degradation rate of vibrational healthy index with an exponential function. By observing the fitting parameters of the exponential function, which reflect the trend of vibrational healthy index, the system can send out an alarm when there is an obvious decline of vibrational healthy index. With the result of this study, the intelligent diagnosis cloud system can issue an alarm immediately when the blade has minor abnormality and can be used in real-time cutting process in the future.

[1] S. Söderberg, L. Åhman, and M. Svenzon, "A metallurgical study of the wear of band-saw blades," Wear, vol. 85, no. 1, pp. 11-27, 1983.
[2] M. Sarwar, M. Persson, and H. Hellbergh, "Wear and failure modes in the bandsawing operation when cutting ball-bearing steel," Wear, vol. 259, no. 7-12, pp. 1144-1150, 2005.
[3] N. Zhu, C. Tanaka, T. Ohtani, and H. Usuki, "Automatic detection of washboarding in bandsaws," Journal of Wood Science, vol. 47, no. 2, pp. 102-108, 2001.
[4] C. Tanaka, Y. Shiota, A. Takahashi, and M. Nakamura, "Experimental studies on band saw blade vibration," Wood Science and Technology, vol. 15, no. 2, pp. 145-159, 1981.
[5] P. Gendraud, J.-C. Roux, and J.-M. Bergheau, "Vibrations and stresses in band saws: A review of literature for application to the case of aluminium-cutting high-speed band saws," Journal of Materials Processing Technology, vol. 135, no. 1, pp. 109-116, 2003.
[6] E. Gadelmawla, "A vision system for surface roughness characterization using the gray level co-occurrence matrix," NDT & e International, vol. 37, no. 7, pp. 577-588, 2004.
[7] R.-S. Lu, G.-Y. Tian, D. Gledhill, and S. Ward, "Grinding surface roughness measurement based on the co-occurrence matrix of speckle pattern texture," Applied Optics, vol. 45, no. 35, pp. 8839-8847, 2006.
[8] E. Kayahan, H. Oktem, F. Hacizade, H. Nasibov, and O. Gundogdu, "Measurement of surface roughness of metals using binary speckle image analysis," Tribology International, vol. 43, no. 1-2, pp. 307-311, 2010.
[9] S. Dutta, A. Datta, N. D. Chakladar, S. Pal, S. Mukhopadhyay, and R. Sen, "Detection of tool condition from the turned surface images using an accurate grey level co-occurrence technique," Precision Engineering, vol. 36, no. 3, pp. 458-466, 2012.
[10] S. Dutta, S. K. Pal, and R. Sen, "On-machine tool prediction of flank wear from machined surface images using texture analyses and support vector regression," Precision Engineering, vol. 43, pp. 34-42, 2016.
[11] R. Zhao, R. Yan, Z. Chen, K. Mao, P. Wang, and R. X. Gao, "Deep learning and its applications to machine health monitoring," Mechanical Systems and Signal Processing, vol. 115, pp. 213-237, 2019.
[12] S. Khan and T. Yairi, "A review on the application of deep learning in system health management," Mechanical Systems and Signal Processing, vol. 107, pp. 241-265, 2018.
[13] S. Tao, T. Zhang, J. Yang, X. Wang, and W. Lu, "Bearing fault diagnosis method based on stacked autoencoder and softmax regression," in 2015 34th Chinese Control Conference (CCC), 2015: IEEE, pp. 6331-6335.
[14] W. Sun, S. Shao, R. Zhao, R. Yan, X. Zhang, and X. Chen, "A sparse auto-encoder-based deep neural network approach for induction motor faults classification," Measurement, vol. 89, pp. 171-178, 2016.
[15] G. S. Galloway, V. M. Catterson, T. Fay, A. Robb, and C. Love, "Diagnosis of tidal turbine vibration data through deep neural networks," Proceedings of the European Conference of the PHM Society vol. 3, no. 1, pp. 172-180, 2016.
[16] D. Weimer, B. Scholz-Reiter, and M. Shpitalni, "Design of deep convolutional neural network architectures for automated feature extraction in industrial inspection," CIRP Annals, vol. 65, no. 1, pp. 417-420, 2016.
[17] N. Gugulothu, T. Vishnu, P. Malhotra, L. Vig, P. Agarwal, and G. M. Shroff, "Predicting Remaining Useful Life using Time Series Embeddings based on Recurrent Neural Networks," ArXiv, vol. abs/1709.01073, 2017.
[18] N. Gugulothu, V. Tv, P. Malhotra, L. Vig, P. Agarwal, and G. Shroff, "Predicting remaining useful life using time series embeddings based on recurrent neural networks," arXiv preprint arXiv:1709.01073, 2017.
[19] 陳冠宇, "運用非監督式深度學習方法分析滾珠螺桿傳動系統性能之智慧化診斷," 成功大學機械工程學系學位論文, pp. 1-144, 2020.
[20] 謝祐禾, "運用非監督式學習診斷刀具磨耗階段之研究," 成功大學機械工程學系學位論文, pp. 1-186, 2020.
[21] J. W. Cooley and J. W. Tukey, "An algorithm for the machine calculation of complex Fourier series," Mathematics of Computation, vol. 19, no. 90, pp. 297-301, 1965.
[22] 李天龍, "以 FFT 為架構建立之諧波參數建立方法," 中山大學電機工程學系學位論文, pp. 1-59, 2000.
[23] A. Eschler, "Stresses and vibrations in bandsaw blades," University of British Columbia, 1982.
[24] W. Wang, Z. Chen, X. Yuan, and X. Wu, "Adaptive image enhancement method for correcting low-illumination images," Information Sciences, vol. 496, pp. 25-41, 2019.
[25] E. H. Land and J. J. McCann, "Lightness and retinex theory," Josa, vol. 61, no. 1, pp. 1-11, 1971.
[26] R. M. Haralick, K. Shanmugam, and I. H. Dinstein, "Textural features for image classification," IEEE Transactions on Systems, Man, and Cybernetics, vol. SMC-3, no. 6, pp. 610-621, 1973.
[27] E. Gadelmawla, A. Eladawi, O. Abouelatta, and I. Elewa, "Investigation of the cutting conditions in milling operations using image texture features," Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture, vol. 222, no. 11, pp. 1395-1404, 2008.
[28] S. R. Safavian and D. Landgrebe, "A survey of decision tree classifier methodology," IEEE transactions on Systems, Man, and Cybernetics, vol. 21, no. 3, pp. 660-674, 1991.
[29] C. Cortes and V. Vapnik, "Support-vector networks," Machine Learning, vol. 20, no. 3, pp. 273-297, 1995.
[30] J. MacQueen, "Some methods for classification and analysis of multivariate observations," in Proceedings of the fifth Berkeley Symposium on Mathematical Statistics and Probability, 1967, vol. 1, no. 14: Oakland, CA, USA, pp. 281-297.
[31] T. K. Ho, "The random subspace method for constructing decision forests," IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 20, no. 8, pp. 832-844, 1998.
[32] W. S. McCulloch and W. Pitts, "A logical calculus of the ideas immanent in nervous activity," The Bulletin of Mathematical Biophysics, vol. 5, no. 4, pp. 115-133, 1943.
[33] D. O. Hebb, "The organization of behavior; a neuropsycholocigal theory," A Wiley Book in Clinical Psychology, vol. 62, p. 78, 1949.
[34] P. Baldi, "Autoencoders, unsupervised learning, and deep architectures," in Proceedings of ICML Workshop on Unsupervised and Transfer Learning, 2012, vol. 27, pp. 37-49.
[35] Z. C. Lipton, J. Berkowitz, and C. Elkan, "A critical review of recurrent neural networks for sequence learning," arXiv preprint arXiv:1506.00019, 2015.
[36] D. Dong, X.-Y. Li, and F.-Q. Sun, "Life prediction of jet engines based on LSTM-recurrent neural networks," in 2017 Prognostics and System Health Management Conference (PHM-Harbin), 2017: IEEE, pp. 1-6.
[37] Y. Shen, S. Tan, A. Sordoni, and A. Courville, "Ordered neurons: Integrating tree structures into recurrent neural networks," arXiv preprint arXiv:1810.09536, 2018.
[38] P. Malhotra, V. TV, A. Ramakrishnan, G. Anand, L. Vig, P. Agarwal, and G. Shroff, "Multi-sensor prognostics using an unsupervised health index based on LSTM encoder-decoder," arXiv preprint arXiv:1608.06154, 2016.
[39] O. Ronneberger, P. Fischer, and T. Brox, "U-net: Convolutional networks for biomedical image segmentation," in International Conference on Medical Image Computing and Computer-Assisted Intervention, 2015: MICCAI, pp. 234-241.
[40] ITREAD. (May 17). 顏色空間RGB與HSV(HSL)的轉換 [Online].Available:https://www.itread01.com/content/1549111148.html.
[41] P. Li, X. Jia, J. Feng, F. Zhu, M. Miller, L.-Y. Chen, and J. Lee, "A novel scalable method for machine degradation assessment using deep convolutional neural network," Measurement, vol. 151, p. 107106, 2020.
[42] Cosen. (May 17). Cosen saws [Online]. Available: https://www.cosen.com/product/band-saws/snc%EF%BC%8Fnc--programmable-automatic-mass-production/c-320gnc.
[43] PicoCoulomB. (May 17). PCB Model 352C22 [Online]. Available: https://www.pcb.com/products?model=352c22.
[44] PicoCoulomB. (May 17). PCB Model 356A33 [Online]. Available: https://www.pcb.com/products?model=356a33.
[45] M. E. s. I. Hub. (May 17). JIS S45C-Mild Steel-An Overview [Online]. Available: https://www.meadinfo.org/2010/03/s45c-jis-mechanical-properties.html.
[46] Amada. (May 17). SGLB Saw Blades [Online]. Available: https://amadamt.com/index.php?dispatch=products.view&product_id=16.
[47] M.Sarwar, M.Persson, and H.Hellbergh, "Proceedings of the 34th International MATADOR Conference," University of Manch-ester Institute of Science and Technology UMIST, pp. 103-110, 2004.