領域自適應是機械故障診斷領域的一個關鍵挑戰，因為傳統故障診斷模型的性能在應用於不同的工作條件或域時會顯著降低。本論文中，我們提出了一種用於機械故障診斷領域自適應的鑑別分類器生成對抗網路 (Critic Classifier Generative Adversarial Network GAN)新方法。我們的方法旨在通過對齊源域和目標域來提高診斷性能，從而實現它們之間的高效知識轉移。我們利用 Critic Classifier GAN 框架的強大功能，該框架結合了生成器對抗網路和鑑別分類器，來學習域不變表示並準確分類故障模式。此外，我們採用差異損失函數，如最大平均差異（MMD）和最大分類器差異（MCD）方法，以進一步增強域對齊和分類對齊。與現有領域自適應技術相比，對各種機械故障數據集進行的實驗評估證實了我們提出的方法的有效性和穩健性。 我們的研究有效地解決了域轉移所帶來的挑戰，並在各種工作條件下的機械故障診斷中取得了最佳性能。

Domain adaptation is a crucial challenge in the field of machinery fault diagnosis, as the performance of traditional fault diagnosis models can significantly degrade when applied to different working conditions or domains. In this thesis, we propose a novel approach for domain adaptation in machinery fault diagnosis based on the Critic Classifier Generative Adversarial Network (GAN). Our method aims to improve diagnostic performance by aligning the source and target domains, enabling effective knowledge transfer between them. We leverage the power of the Critic Classifier GAN framework, which incorporates both a generator adversarial network and a critic classifier, to learn domain-invariant representations and classify fault patterns accurately. Additionally, we employ domain discrepancy loss functions, such as Maximum Mean Discrepancy (MMD) and the Maximum Classifier Discrepancy (MCD) method, to further enhance domain alignment and classifiers to align distributions. Experimental evaluations conducted on various mechanical failure datasets confirm the effectiveness and robustness of our proposed method in comparison to existing domain adaptation techniques. Our proposed solution effectively overcomes the challenges arising from domain shift and achieved state-of-the-art performance in machinery fault diagnosis under various working conditions.

摘要-----------------------------------------I
ABSTRACT-------------------------------------II
誌謝-----------------------------------------III
目錄-----------------------------------------IV
List of Figures------------------------------VI
List of Tables-------------------------------VII
Chapter 1 Introduction-----------------------1
1.1 Motivation-------------------------------1
1.2 Problem Statement------------------------3
1.3 Contributions----------------------------5
1.4 Thesis Organization----------------------6
Chapter 2 Related Works----------------------7
2.1 ML/DL Learning-based Methods ------------7
2.2 Domain Adaptation Methods----------------10
Chapter 3 Proposed Method--------------------14
3.1 Overview---------------------------------14
3.2 Problem Formulation----------------------16
3.3 Network Architecture---------------------17
3.4 Domain Alignment Method:-----------------18
3.6 Objective Function-------------------20
3.6.1 Adversarial loss---------------------20
3.7 Training Procedure-------------------20
Chapter 4 Data Preparation-------------------23
4.1 Descriptions of Datasets-----------------23
4.1.1 CWRU Dataset---------------------------23
4.1.2 JUN Dataset----------------------------24
4.2 Data Preprocessing-----------------------24
4.2.1 Input Types----------------------------25
4.2.2 Data Splitting-------------------------27
4.2.3 Normalization--------------------------27
Chapter 5 Experiments------------------------28
5.1 Implementation and Experiment Settings---28
5.2 Experiment Comparison--------------------29
5.2.1 Comparison with CWRU-------------------29
5.2.2 Comparison with JUN--------------------32
5.3 Ablation Study---------------------------36
Chapter 6 Conclusions------------------------39
References-----------------------------------40

1. Mao, G., et al., Fusion Domain-Adaptation CNN Driven by Images and Vibration Signals for Fault Diagnosis of Gearbox Cross-Working Conditions. Entropy, 2022. 24(1): p. 119.
2. Li, X., et al., Diagnosing rotating machines with weakly supervised data using deep transfer learning. IEEE transactions on industrial informatics, 2019. 16(3): p. 1688-1697.
3. Zhang, W., et al., Universal domain adaptation in fault diagnostics with hybrid weighted deep adversarial learning. IEEE Transactions on Industrial Informatics, 2021. 17(12): p. 7957-7967.
4. Dai, S., et al. Contrastively smoothed class alignment for unsupervised domain adaptation. in Proceedings of the Asian Conference on Computer Vision. 2020.
5. Li, X., et al., Deep learning-based machinery fault diagnostics with domain adaptation across sensors at different places. IEEE Transactions on Industrial Electronics, 2019. 67(8): p. 6785-6794.
6. Pinedo-Sanchez, L.A., D.A. Mercado-Ravell, and C.A. Carballo-Monsivais, Vibration analysis in bearings for failure prevention using CNN. Journal of the Brazilian Society of Mechanical Sciences and Engineering, 2020. 42(12): p. 1-17.
7. Jung, D., et al., Combining model-based diagnosis and data-driven anomaly classifiers for fault isolation. Control Engineering Practice, 2018. 80: p. 146-156.
8. Wu, X., et al., A hybrid classification autoencoder for semi-supervised fault diagnosis in rotating machinery. Mechanical Systems and Signal Processing, 2021. 149: p. 107327.
9. Shao, S., P. Wang, and R. Yan, Generative adversarial networks for data augmentation in machine fault diagnosis. Computers in Industry, 2019. 106: p. 85-93.
10. Wu, J., et al., Learning from Class-imbalanced Data with a Model-Agnostic Framework for Machine Intelligent Diagnosis. Reliability Engineering & System Safety, 2021. 216: p. 107934.
11. Han, T., et al., Learning transferable features in deep convolutional neural networks for diagnosing unseen machine conditions. ISA Trans, 2019. 93: p. 341-353.
12. Saito, K., et al. Maximum classifier discrepancy for unsupervised domain adaptation. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
13. Huo, C., et al., A Class-Level Matching Unsupervised Transfer Learning Network for Rolling Bearing Fault Diagnosis Under Various Working Conditions. Available at SSRN 4273459, 2022.
14. Sun, B., J. Feng, and K. Saenko. Return of frustratingly easy domain adaptation. in Proceedings of the AAAI conference on artificial intelligence. 2016.
15. Ghifary, M., et al. Deep reconstruction-classification networks for unsupervised domain adaptation. in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part IV 14. 2016. Springer.
16. Ganin, Y. and V. Lempitsky. Unsupervised domain adaptation by backpropagation. in International conference on machine learning. 2015. PMLR.
17. Li, Y., et al., Revisiting batch normalization for practical domain adaptation. arXiv preprint arXiv:1603.04779, 2016.
18. Sankaranarayanan, S., et al. Generate to adapt: Aligning domains using generative adversarial networks. in Proceedings of the IEEE conference on computer vision and pattern recognition. 2018.
19. Liu, M.-Y. and O. Tuzel, Coupled generative adversarial networks. Advances in neural information processing systems, 2016. 29.
20. Kang, G., et al. Contrastive adaptation network for unsupervised domain adaptation. in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2019.
21. Lin, Z., et al. Improving maximum classifier discrepancy by considering joint distribution for domain adaptation. in Web Information Systems Engineering–WISE 2018: 19th International Conference, Dubai, United Arab Emirates, November 12-15, 2018, Proceedings, Part II 19. 2018. Springer.
22. Borgwardt, K.M., et al., Integrating structured biological data by kernel maximum mean discrepancy. Bioinformatics, 2006. 22(14): p. e49-e57.
23. Ganin, Y., et al., Domain-adversarial training of neural networks. The journal of machine learning research, 2016. 17(1): p. 2096-2030.
24. Odena, A., C. Olah, and J. Shlens. Conditional image synthesis with auxiliary classifier gans. in International conference on machine learning. 2017. PMLR.
25. Mirza, M. and S. Osindero, Conditional generative adversarial nets. arXiv preprint arXiv:1411.1784, 2014.
26. Odena, A., Semi-supervised learning with generative adversarial networks. arXiv preprint arXiv:1606.01583, 2016.
27. Zhao, Z., et al., Deep learning algorithms for rotating machinery intelligent diagnosis: An open source benchmark study. ISA transactions, 2020. 107: p. 224-255.
28. Loparo, K. CWRU, Case western reserve University bearing data center, Seeded Fault Test Data. 2012; Available from: https://engineering.case.edu/bearingdatacenter/download-data-file.
29. Li, K., et al., Sequential fuzzy diagnosis method for motor roller bearing in variable operating conditions based on vibration analysis. Sensors, 2013. 13(6): p. 8013-8041.
30. Dennis, J., H.D. Tran, and H. Li, Spectrogram image feature for sound event classification in mismatched conditions. IEEE signal processing letters, 2010. 18(2): p. 130-133.
31. Jia, F., et al., Deep neural networks: A promising tool for fault characteristic mining and intelligent diagnosis of rotating machinery with massive data. Mechanical systems and signal processing, 2016. 72: p. 303-315.
32. Zhang, W., et al., A new deep learning model for fault diagnosis with good anti-noise and domain adaptation ability on raw vibration signals. Sensors, 2017. 17(2): p. 425.
33. Zhang, B., et al., Adversarial adaptive 1-D convolutional neural networks for bearing fault diagnosis under varying working condition. arXiv preprint arXiv:1805.00778, 2018.
34. Han, T., et al., Learning transferable features in deep convolutional neural networks for diagnosing unseen machine conditions. ISA transactions, 2019. 93: p. 341-353.
35. Long, M., et al. Learning transferable features with deep adaptation networks. in International conference on machine learning. 2015. PMLR.
36. Long, M., et al. Deep transfer learning with joint adaptation networks. in International conference on machine learning. 2017. PMLR.
37. Sun, B. and K. Saenko. Deep coral: Correlation alignment for deep domain adaptation. in Computer Vision–ECCV 2016 Workshops: Amsterdam, The Netherlands, October 8-10 and 15-16, 2016, Proceedings, Part III 14. 2016. Springer.
38. Long, M., et al., Conditional adversarial domain adaptation. Advances in neural information processing systems, 2018. 31.