感應馬達在各式系統中扮演著重要的角色，因此感應馬達的維護與保養直接地影響系統經濟效益及人員安全性，希望能利用馬達所得訊號，達到提早發現故障，並且透過訊號所含有之特徵得知故障種類，進而提前安排維修、減少維護時間、減少事故發生。故障診斷系統由三步驟組成：資料蒐集、特徵擷取和故障分類，資料蒐集上受感測器準確度影響，能越少雜訊與誤差越好，特徵擷取在傳統機器學習上須仰賴專家提取故障特徵，在機器學習的演進下，延伸發展出深度學習，透過神經網路自動提取訊號特徵，神經網路可利用所提取之特徵進行分類，而現行趨勢導向轉移學習，希望能藉由不同領域學習的成果應用到其他領域，達到轉移應用之成效。
本論文設計了四種故障狀態感應馬達類型，包含定子故障、轉子故障、軸承故障、不對心故障，並採用了感應馬達在五種負載狀態下之電流與振動訊號，利用原始時域訊號重新排列製圖後分別輸入兩種
傳統深度學習模型（卷積神經網路、循環神經網路）與兩種轉移學習模型（深度域混淆、域對抗神經網路）進行準確度比較與結果分析，選出最佳模型後進行各負載之跨域分析。最後得到重要結果為使用感應馬達之振動訊號，並利用轉移學習之域對抗神經網路方法進行故障診斷可以達到各域都有相當高之準確度。

Induction motors play a crucial role in various systems. Therefore, the maintenance of induction motors directly affects the economic efficiency of the system and personnel safety. Motor signals can be used for early identification of motor fault, with the fault types determined according to the signal features. This allows for arrangement of preemptive maintenance, reduces maintenance time, and reduces accidents. Fault diagnosis involves the following three steps: data collection, feature extraction, and fault classification. The accuracy of the sensor determines the quality of data collection; a small amount of noise and error results in high-quality data collection. Traditional machine learning for feature extraction relies on fault features extracted by experts. Developed upon the advances in machine learning, deep learning enables automatic extraction of signal features through a neural network, which can perform classification according to the extracted features. Transfer learning, which shows growing predominance in the development of machine learning, is aimed at applying the learning results for a field to other ones.
This study designed four types of induction motor fault, namely stator fault, rotor fault, bearing fault, and misalignment fault. The current and vibration signals of the induction motor at five different loads were used to rearrange and map the original time-domain signals. The data were entered into two traditional deep learning models (i.e., convolutional neural network and recurrent neural network) and two transfer learning models (i.e., deep domain confusion and domain-adversarial neural network) for accuracy comparison and result analysis. After the optimal model was determined, cross-domain analysis of each load was performed. According to the study result, high accuracy was achieved by the fault diagnosis method that involved the use of the induction motor’s vibration signals and the transfer-learning-based domain-adversarial neural network in all
domains.

[1] D. Basak, A. Tiwari and S. P. Das, "Fault diagnosis and condition
monitoring of electrical machines-a review," 2006 IEEE
International Conference on Industrial Technology, pp. 3061-3066,
2006.
[2] P. Zhang, Y. Du, T.G. Habetler and B. Lu, "A survey of condition
monitoring and protection methods for medium-voltage induction
motors," IEEE Trans. Ind. Appl., vol. 47, no. 1, pp. 34-46, 2011.
[3] Y. Lei, B. Yang, X. Jiang, F. Jia, N. Li, A. K. Nandi, "Applications of
machine learning to machine fault diagnosis: A review and
roadmap," Mech. Syst. Sig. Process, p. 106587, 2020.
[4] L. Guo, Y. Lei, S. Xing, T. Yan and N. Li, "Deep convolutional
transfer learning network: A new method for intelligent fault
diagnosis of machines with unlabeled data," IEEE Trans. Ind.
Electron., vol. 66, no. 9, pp. 7316-7325, Sep. 2018.
[5] F. Lv, C. Wen, Z. Bao and Meiqin Liu, "Fault diagnosis based on
deep learning," American Control Conference (ACC), 2016.
[6] C. Zhang, L. Xu, X. Li and H. Wang, "A method of fault diagnosis
for rotary equipment based on deep learning," Prognostics and
System Health Management Conference (PHM-Chongqing), pp.
958-962, 2018.
[7] D. Choi, J. Han, S. Park and S. Hong, "Comparative Study of CNN
and RNN for Motor fault Diagnosis Using Deep Learning," 2020
IEEE 7th International Conference on Industrial Engineering and
Applications (ICIEA), pp. 693-696, 2020.
[8] Q. Zhang, J. Lin, H. Song and G. Sheng, "Fault identification based
on PD ultrasonic signal using RNN DNN and CNN," Proc.
Condition Monit. Diagnosis (CMD), pp. 1-6, Sep. 2018.
52
[9] L. Wen, X. Li, L. Gao and Y. Zhang, "A new convolutional neural
network based data-driven fault diagnosis method," IEEE Trans.
Ind. Electron., vol. 65, no. 7, pp. 5990-5998, Jul. 2018.
[10] J. Wang, Z. Mo, H. Zhang and Q. Miao, "A deep learning method for
bearing fault diagnosis based on time-frequency image," IEEE
Access, vol. 7, pp. 42373-42383, 2019.
[11] A. Nandi, S. Biswas, K. Samanta, S. S. Roy and S. Chatterjee,
"Diagnosis of Induction Motor Faults Using Frequency Occurrence
Image Plots—A Deep Learning Approach," 2019 International
Conference on Electrical, Electronics and Computer Engineering
(UPCON), February 2020.
[12] J. Zhu, N. Chen and C. Shen, "A new deep transfer learning method
for bearing fault diagnosis under different working conditions,"
IEEE Sensors J., Aug. 2019.
[13] Q. Liu and C. Huang, "A fault diagnosis method based on transfer
convolutional neural networks," IEEE Access, vol. 7, pp. 171423-
171430, 2019.
[14] Q. Wang, G. Michau and O. Fink, "Domain adaptive transfer
learning for fault diagnosis," Proc. Prognostics Syst. Health
Manage. Conf. (PHM-Paris), pp. 279-285, 2019.
[15] K. Saito, K. Watanabe, Y. Ushiku and T. Harada, "Maximum
classifier discrepancy for unsupervised Domain Adaptation, "
Proceedings of the IEEE Conference on Computer Vision and
Pattern Recognition (CVPR), 2018.
[16] W. Xu, Y. Wan, T.Y. Zuo and X.M. Sha, "Transfer Learning Based
Data Feature Transfer for Fault Diagnosis," IEEE Access, pp. 76120-
76129, 2020.
[17] E. Tzeng, J. Hoffman, N. Zhang, K. Saenko and T. Darrell, "Deep
domain confusion: Maximizing for domain invariance,"
arXiv:1412.3474,2014.
53
[18] Y. Ganin, E. Ustinova, H. Ajakan, P. Germain, H. Larochelle, F.
Laviolette, M. Marchand and V. Lempitsky, "Domain-adversarial
training of neural networks,"Journal of Machine Learning Research,
2016.
[19] A. H. li and S. Li-ping, "Fault diagnosis of induction motor based on
information entropy fusion,"2010 2nd International Conference on
Advanced Computer Control, March 2010.
[20] 機器學習 VS 深度學習，誰在分類問題上獲勝？Retrieved from
https://kknews.cc/tech/g9j6rv9.html.
[21] Hung-yi Lee. "ML Lecture 19: Transfer Learning", Youtube, upload
by Hung-yi Lee, 9 December 2016,
https://www.youtube.com/watch?v=qD6iD4TFsdQ&t=2778s&ab_c
hannel=Hung-yiLee.
[22] G. K. Singh and S. A. S. Al Kazzaz, "Induction machine drive
condition monitoring and diagnostic research—A survey," Elect.
Power Syst. Res., vol. 64, no. 2, pp. 145-158, Feb. 2003.
[23] W.S. Gongora, H.V.D. Silva, A. Goedtel, W.F. Godoy and S.A.O. da
Silva, "Neural approach for bearing fault detection in three phase
induction motors," IEEE International Symposium on Diagnostics
for Electric Machines Power Electronics and Drives (SDEMPED),
pp. 566-572, 2013.
[24] 曾思憲，「感應馬達主動型狀態估測系統之研發」，碩士論文，
國立台灣科技大學，2018 年。
[25] 林芝以，「運用電氣訊號於感應馬達狀態監測與故障診斷之研
究」，碩士論文，台灣科技大學電機工程學系，2016 年。
[26] National Instrument. Retrieved from https://www.ni.com/zh-tw.html.
[27] 人工智慧、機器學習、深度學習介紹 Retrieved from
http://tensorflowkeras.blogspot.com/2017/06/blog-post_15.html.
54
[28] Day01 Transfer Learning 遷移式學習(2020) Retrieved from
https://www.coderbridge.com/series/d4b5a1a1565e4e7a9cd14618ffe
6146f/posts/54584ea6d4c240aeb3b8ae4af3a0531a.
[29] Wikipedia Contributors. "Convolutional Neural Network" Retrieved
from https://en.wikipedia.org/wiki/Convolutional_neural_network.
[30] CH.Tseng. 初探卷積神經網路(2017) Retrieved from
https://chtseng.wordpress.com/2017/09/12/%E5%88%9D%E6%8E
%A2%E5%8D%B7%E7%A9%8D%E7%A5%9E%E7%B6%93%E
7%B6%B2%E8%B7%AF/.
[31] Jimmy Lin 卷積神經網路的運作原理(2016) Retrieved from
https://brohrer.mcknote.com/zhHant/how_machine_learning_works/how_convolutional_neural_net
works_work.html.
[32] Wikipedia Contributors. "Recurrent neural network" Retrieved from
https://en.wikipedia.org/wiki/Recurrent_neural_network.
[33] RNN 介紹 Retrieved from
https://codertw.com/%E7%A8%8B%E5%BC%8F%E8%AA%9E%
E8%A8%80/617324/.
[34] 神奇的深度遷移學習 DDC：Deep Domain Confusion:
Maximizing for Domain Invariance Retrieved from
https://blog.csdn.net/qq_24305433/article/details/80239412.
[35] 基於對抗的遷移學習方法: DANN 域對抗網路 Retrieved from
https://zhuanlan.zhihu.com/p/73947456.
[36] L. van der Maaten and G.E. Hinton, "Visualizing High-Dimensional
Data Using t-SNE," J. Machine Learning Research, vol. 9, pp. 2579-
2605, 2008.
[37] 徐子權，「電氣與振動檢測法於馬達狀態評估之研究」，碩士論
文，國立台灣科技大學，2014 年。
55
[38] 馮建中，「基於卷積神經網路之感應馬達新故障診斷法」，碩士
論文，國立台灣科技大學，2020 年。
[39] 蕭宗閔，「應用數據增強於卷積神經網路之馬達故障診斷之研
究」，碩士論文，國立台灣科技大學，2020 年。