在現代工業中，設備的穩定性是工廠生產和機器利用的重要因素。機械振動可能會導致旋轉機器中的主軸偏心或軸承損壞，進而導致設備停機。近年來，由於深度學習模型具有自動特徵學習能力，因此它們被用於故障診斷。然而，仍存在一些限制，例如：無法應用於不同工作條件下的機器的不平衡數據問題。本論文提出了一種基於多通道輸入深度遷移學習模型之旋轉機械故障診斷，該模型可以利用源領域學到的知識來解決目標域相似的問題。其原理藉由輸入原始時域訊號經預處理得到的時域灰階圖像（64x64）、兩張時頻圖像（64x64）所組成的三通道圖像，再從目標領域遷移學習模型的全連接層調整隱藏層而得到模型參數。訓練集、驗證集和測試集的樣本比例分別為49％、21%和30％。對兩個類似的開源數據集CWRU、Mendeley依照軸承的內環、外環、滾珠故障等九種故障與正常狀況進行測試，並使用深度學習的指標準確率、回報率、混淆矩陣、P-R曲線進行模型的驗證。在CWRU下兩種不同負載間相互遷移的結果中，單通道、雙通道及三通道的準確率分別為92.8%、93.13%、96.74%；而兩種數據集（CWRU(A)、Mendeley(B)）的相互遷移的結果中，單通道、雙通道及三通道的結果分別為76.12%、81.27%、86.4%。實驗結果顯示多通道輸入相較於單通道輸入有較佳的表現並且相較於未遷移學習具有良好的準確率與更少的訓練時間。

In modern industry, the stability of equipment is a critical factor for factory production and machine utilization. Mechanical vibrations can lead to spindle eccentricity or bearing damage in rotating machinery, resulting in equipment downtime. In recent years, deep learning models have been utilized for fault diagnosis due to their automatic feature learning capabilities. However, there are still limitations, such as the inability to handle imbalanced data from machines under different operating conditions. This thesis proposes a method based on deep transfer learning model with multi-channel inputs for fault diagnosis of rotating machine. This model can solve similar problems in target domain through the knowledge learned from source domain. The input consists of a three-channel image, which is composed of one time-domain grayscale images (64x64) and two time-frequency images (64x64) obtained from the raw time-domain signals. The model parameters are obtained by adjusting the hidden layers of the target domain transfer learning model's fully connected layers. The sample ratios for the training, validation, and testing sets are 49%, 21%, and 30%, respectively. Testing is conducted on two similar open-source datasets, CWRU and Mendeley, in which normal condition and nine types of faults based on inner race, outer race, and ball faults. Finally, the model is validated by four metrics including accuracy, recall rate, confusion matrix, and P-R curve. In the case of transferring between two different loads of CWRU, the accuracy of single-channel, dual-channel, and three-channel are 92.8%, 93.13% and 96.74%. In the case of transferring between the two datasets, CWRU(A) and Mendeley(B), the accuracy of single-channel, dual-channel and three-channel are 76.12%, 81.27% and 86.4%. These experimental results demonstrate that multi-channel input outperforms single-channel input, and in comparison to non-transfer learning, it exhibits both superior accuracy and reduced training time.

[1]Y. Ju, X. Tian, H. Liu, and L. Ma, "Fault detection of networked dynamical systems: A survey of trends and techniques," International Journal of Systems Science, vol. 52, no. 16, pp. 3390-3409, 2021.
[2]C. Li, J. V. De Oliveira, M. Cerrada, D. Cabrera, R. V. Sánchez, and G. Zurita, "A systematic review of fuzzy formalisms for bearing fault diagnosis," IEEE Transactions on Fuzzy Systems, vol. 27, no. 7, pp. 1362-1382, 2018.
[3]A. Rai and S. H. Upadhyay, "A review on signal processing techniques utilized in the fault diagnosis of rolling element bearings," Tribology International, vol. 96, pp. 289-306, 2016.
[4]B. Assaad, M. Eltabach, and J. Antoni, "Vibration based condition monitoring of a multistage epicyclic gearbox in lifting cranes," Mechanical Systems and Signal Processing, vol. 42, no. 1-2, pp. 351-367, 2014.
[5]S. Shen et al., "A physics-informed deep learning approach for bearing fault detection," Engineering Applications of Artificial Intelligence, vol. 103, p. 104295, 2021.
[6]Y. Lei, B. Yang, X. Jiang, F. Jia, N. Li, and A. K. Nandi, "Applications of machine learning to machine fault diagnosis: A review and roadmap," Mechanical Systems and Signal Processing, vol. 138, p. 106587, 2020.
[7]P. Konar and P. Chattopadhyay, "Bearing fault detection of induction motor using wavelet and Support Vector Machines (SVMs)," Applied Soft Computing, vol. 11, no. 6, pp. 4203-4211, 2011.
[8]H. Shao, H. Jiang, H. Zhao, and F. Wang, "A novel deep autoencoder feature learning method for rotating machinery fault diagnosis," Mechanical Systems and Signal Processing, vol. 95, pp. 187-204, 2017.
[9]J. Tian, C. Morillo, M. H. Azarian, and M. Pecht, "Motor bearing fault detection using spectral kurtosis-based feature extraction coupled with K-nearest neighbor distance analysis," IEEE Trans. Ind. Electron., vol. 63, no. 3, pp. 1793-1803, 2015.
[10]Y. Yang, Y. Liao, G. Meng, and J. Lee, "A hybrid feature selection scheme for unsupervised learning and its application in bearing fault diagnosis," Expert Systems with Applications, vol. 38, no. 9, pp. 11311-11320, 2011.
[11]S. Chen, R. Yang, and M. Zhong, "Graph-based semi-supervised random forest for rotating machinery gearbox fault diagnosis," Control Engineering Practice, vol. 117, p. 104952, 2021.
[12]Z. Zhao et al., "Applications of unsupervised deep transfer learning to intelligent fault diagnosis: A survey and comparative study," IEEE Transactions on Instrumentation and Measurement, vol. 70, pp. 1-28, 2021.
[13]B. Yang, R. Liu, and E. Zio, "Remaining useful life prediction based on a double-convolutional neural network architecture," IEEE Trans. Ind. Electron., vol. 66, no. 12, pp. 9521-9530, 2019.
[14]D.-J. Choi, J.-H. Han, S.-U. Park, and S.-K. Hong, "Comparative study of CNN and RNN for motor fault diagnosis using deep learning," in 2020 IEEE 7th International Conference on Industrial Engineering and Applications (ICIEA), 2020: IEEE, pp. 693-696.
[15]D.-T. Hoang and H.-J. Kang, "A survey on deep learning based bearing fault diagnosis," Neurocomputing, vol. 335, pp. 327-335, 2019.
[16]B. U. Deveci, M. Celtikoglu, T. Alp, O. Albayrak, P. Unal, and P. Kirci, "A comparison of deep transfer learning methods on bearing fault detection," in 2021 8th International Conference on Future Internet of Things and Cloud (FiCloud), 2021: IEEE, pp. 285-292.
[17]Z. Zhu et al., "A review of the application of deep learning in intelligent fault diagnosis of rotating machinery," Measurement, p. 112346, 2022.
[18]X. H. Chen, R. Yang, Y. H. Xue, M. J. Huang, R. Ferrero, and Z. D. Wang, "Deep Transfer Learning for Bearing Fault Diagnosis: A Systematic Review Since 2016," (in English), Ieee Transactions on Instrumentation and Measurement, Review vol. 72, p. 21, 2023, Art no. 3508221, doi: 10.1109/tim.2023.3244237.
[19]J. Zhou, M. Xiao, Y. Niu, and G. Ji, "Rolling bearing fault diagnosis based on WGWOA-VMD-SVM," Sensors, vol. 22, no. 16, p. 6281, 2022.
[20]Q. Lu, R. Yang, M. Zhong, and Y. Wang, "An improved fault diagnosis method of rotating machinery using sensitive features and RLS-BP neural network," IEEE Transactions on Instrumentation and Measurement, vol. 69, no. 4, pp. 1585-1593, 2019.
[21]W. Zhang, C. Li, G. Peng, Y. Chen, and Z. Zhang, "A deep convolutional neural network with new training methods for bearing fault diagnosis under noisy environment and different working load," Mechanical systems and signal processing, vol. 100, pp. 439-453, 2018.
[22]Z. Yang, R. Yang, and M. Huang, "Rolling bearing incipient fault diagnosis method based on improved transfer learning with hybrid feature extraction," Sensors, vol. 21, no. 23, p. 7894, 2021.
[23]M. Sun, H. Wang, P. Liu, S. Huang, P. Wang, and J. Meng, "Stack autoencoder transfer learning algorithm for bearing fault diagnosis based on class separation and domain fusion," IEEE Trans. Ind. Electron., vol. 69, no. 3, pp. 3047-3058, 2021.
[24]A. He and X. Jin, "Deep variational autoencoder classifier for intelligent fault diagnosis adaptive to unseen fault categories," IEEE Transactions on Reliability, vol. 70, no. 4, pp. 1581-1595, 2021.
[25]B. Ma, W. Cai, Y. Han, and G. Yu, "A novel probability confidence CNN model and its application in mechanical fault diagnosis," IEEE Transactions on Instrumentation and Measurement, vol. 70, pp. 1-11, 2021.
[26]J. Wu, Z. Zhao, C. Sun, R. Yan, and X. Chen, "Few-shot transfer learning for intelligent fault diagnosis of machine," Measurement, vol. 166, p. 108202, 2020.
[27]M. J. Hasan, M. M. M. Islam, and J. M. Kim, "Bearing Fault Diagnosis Using Multidomain Fusion-Based Vibration Imaging and Multitask Learning," (in English), Sensors, Article vol. 22, no. 1, p. 21, Jan 2022, Art no. 56, doi: 10.3390/s22010056.
[28]L. Wen, X. Li, and L. Gao, "A transfer convolutional neural network for fault diagnosis based on ResNet-50," Neural Computing and Applications, vol. 32, pp. 6111-6124, 2020.
[29]S. Shao, S. McAleer, R. Yan, and P. Baldi, "Highly accurate machine fault diagnosis using deep transfer learning," IEEE Trans. Ind. Inform., vol. 15, no. 4, pp. 2446-2455, 2018.
[30]W. Zhang, D. Chen, and Y. Kong, "Self-supervised joint learning fault diagnosis method based on three-channel vibration images," Sensors, vol. 21, no. 14, p. 4774, 2021.
[31]王鈞弘, 「基於深度遷移學習之卷積神經網路應用於旋轉機械故障診斷,」 碩士論文, 自動化及控制研究所, 國立臺灣科技大學, 台北市, 2023.
[32]Z. Li, F. Liu, W. Yang, S. Peng, and J. Zhou, "A survey of convolutional neural networks: analysis, applications, and prospects," IEEE transactions on neural networks and learning systems, 2021.
[33]F. Zhuang et al., "A comprehensive survey on transfer learning," Proceedings of the IEEE, vol. 109, no. 1, pp. 43-76, 2020.
[34]K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," arXiv preprint arXiv:1409.1556, 2014.
[35]C. W. R. U. B. D. Center, "Case western reserve university bearing data center website," 2018.
[36]H. Huang and N. Baddour, "Bearing vibration data collected under time-varying rotational speed conditions," Data in brief, vol. 21, pp. 1745-1749, 2018.