Unsupervised domain adaptation transfer learning for the fault diagnosis in rotating machinery
ZHOU Xiangqi，FU Zhongguang，GAO Yucai
Key Laboratory of Power Station Energy Transfer Conversion and System of Ministry of Education, North China Electric Power University, Beijing 102206, China
摘要故障诊断在旋转机械领域具有重要的意义，而深度学习和迁移学习的发展为提高故障诊断的准确性和鲁棒性提供了新的途径。针对旋转机械故障诊断问题，提出了一种基于深度对抗神经网络(Domain-Adversarial Neural Network,DANN) 和多核最大平均差异(Multiple Kernel Maximum Mean Discrepancy,MK-MMD)的无监督域适应迁移学习方法。首先，收集了源工况和目标工况下的振动信号数据并通过快速傅里叶变换转化为频域信号。然后，构建了一个ResNeXt-50特征提取器，并使用DANN和MK-MMD方法进行特征映射和域适应，从而实现源工况到目标工况的迁移学习。实验结果表明，该方法能提高对故障特征的识别精度，且在不同工况下的迁移实验中具有更好的鲁棒性。
Abstract：Fault diagnosis is of great importance in the domain of rotating machinery, and the development of deep learning and transfer learning has provided new avenues in order to enhance the precision and resilience of fault diagnosis. In the context of fault diagnosis in rotating machinery, an unsupervised domain adaptation transfer learning method based on Domain-Adversarial Neural Network (DANN) and Multiple Kernel Maximum Mean Discrepancy (MK-MMD) is proposed. Firstly, vibration signal data from both the source working condition and the target working condition are gathered and converted into frequency domain signals utilizing the Fast Fourier Transform (FFT). Then, a ResNeXt-50 feature extractor is constructed, and DANN and MK-MMD methods are employed for feature mapping and domain adaptation, enabling transfer learning from the source working condition to the target working condition. The experimental findings validate that the proposed method enhances the accuracy of fault feature recognition. and exhibits better robustness in transfer experiments across different working conditions.
周湘淇，付忠广，高玉才. 无监督域适应迁移学习在旋转机械故障诊断中的应用[J]. 振动与冲击, 2024, 43(10): 106-113.
ZHOU Xiangqi，FU Zhongguang，GAO Yucai. Unsupervised domain adaptation transfer learning for the fault diagnosis in rotating machinery. JOURNAL OF VIBRATION AND SHOCK, 2024, 43(10): 106-113.
[1] 侯泽林.旋转机械故障诊断的研究现状及发展前景[J].锻压装备与制造技术,2021,56(05):33-37.DOI:10.16316/j.issn.1672-0121.2021.05.006.
Hou, Z., Research status and development prospects of rotating machinery fault diagnosis. Forging & Stamping Equipment and Manufacturing Technology, 2021, 56(05), 33-37. DOI: 10.16316/j.issn.1672-0121.2021.05.006.
[2] Xu L,Chi Z,Xiang L, et al. Federated transfer learning in fault diagnosis under data privacy with target self-adaptation[J]. Journal of Manufacturing Systems,2023,68.
[3] 杨胜康,孔宪光,王奇斌等.基于多源域深度迁移学习的机械故障诊断[J].振动与冲击,2022,41(09):32-40.DOI:10.13465/j.cnki.jvs.2022.09.005.
YANG Shengkang,KONG Xianguang,WANG Qibin,et al. Mechanical fault diagnosis based on multi-source domain deep transfer learning[J].Journal of Vibration and Shock,2022,41(09):32-40.DOI:10.13465/j.cnki.jvs.2022.09.005.)
[4] Z. Zhao et al., "Applications of Unsupervised Deep Transfer Learning to Intelligent Fault Diagnosis: A Survey and Comparative Study," in IEEE Transactions on Instrumentation and Measurement, vol. 70, pp. 1-28, 2021, Art no. 3525828, doi: 10.1109/TIM.2021.3116309.
[5] 李晶晶,孟利超,张可等.领域自适应研究综述[J].计算机工程,2021,47(06):1-13.DOI:10.19678/j.issn.1000-3428.0060659.
Li, J., Meng, L., Zhang, K., et al. A comprehensive review of domain adaptation research. Computer Engineering, 2021, 47(06), 1-13. DOI: 10.19678/j.issn.1000-3428.0060659.
[6] 邵海东,肖一鸣,颜深.仿真数据驱动的改进无监督域适应轴承故障诊断[J].机械工程学报,2023,59(03):76-85.
Shao, H., Xiao, Y., Yan, S. Improved unsupervised domain adaptation for bearing fault diagnosis driven by simulation data. Journal of Mechanical Engineering, 2023, 59(03), 76-85.
[7] Yahia A,Afendee M M,Saleh A, et al. Speech encryption by multiple chaotic map with fast fourier transform[J]. International Journal of Electrical and Computer Engineering (IJECE),2020,10(6).
[8] Battula B,Ashok B N,Spica G S, et al. Parkinson’s disease detection using modified ResNeXt deep learning model from brain MRI images[J]. Soft Computing,2023,27(16).
[9] 高彤,盛蔚,尹艳召等.基于MMD迁移学习的MEMS惯性传感器故障诊断方法[J].太原科技大学学报,2023,44(02):97-104+110.
Gao Tong, Sheng Wei, Yin Yanzhao, et al. MEMS Inertial Sensor Fault Diagnosis Method Based on MMD Transfer Learning [J]. Journal of Taiyuan University of Science and Technology, 2023, 44(02): 97-104+110.
[10] 徐易芸,马健,陈良等.基于相似性度量迁移学习的轴承故障诊断[J].振动与冲击,2022,41(16):217-223.DOI:10.13465/j.cnki.jvs.2022.16.028.
XU Yiyun,MA Jian,CHEN Liang,et al. Bearing fault diagnosis based on similarity metric transfer learning[J].Journal of Vibration andShock,2022,41(16):217-223.DOI:10.13465/j.cnki.jvs.2022.16.028.)
[11] 王翎,孙涵.结合域混淆与MK-MMD的深度适应网络[J].小型微型计算机系统,2019,40(07):1519-1524.
Wang, L., Sun, H. Deep adaptive network combining domain confusion and MK-MMD. Journal of Chinese Computer Systems, 2019, 40(07), 1519-1524.
[12] Kun C,Zhiyong L,Zhilei L, et al. An improved multi-source domain adaptation network for inter-subject mental fatigue detection based on DANN.[J]. Biomedizinische Technik. Biomedical engineering,2023,68(3).
[13] Rongjun Z,Yun F,Yansong S, et al. A fusion approach to infrared and visible images with Gabor filter and sigmoid function[J]. Infrared Physics and Technology,2023,131.
[14] Chengjie C,Fuhong M,Yunzhen Z, et al. ReLU-type Hopfield neural network with analog hardware implementation[J]. Chaos, Solitons and Fractals: the interdisciplinary journal of Nonlinear Science, and Nonequilibrium and Complex Phenomena,2023,167.
[15] Angel A,Angel Z,Daniel M, et al. Analysis of the use of the Hanning Window for the measurement of interharmonic distortion caused by close tones in IEC standard framework.[J]. Electric Power Systems Research,2022,206.
[16] 司瑾. 基于深度迁移学习的轴承故障诊断方法研究及铁路场景应用[D].北京交通大学,2022.DOI:10.26944/d.cnki.gbfju.2022.000308.
Si, J. Research on bearing fault diagnosis method based on deep transfer learning and its application in railway scenarios [D]. Beijing Jiaotong University, 2022. DOI: 10.26944/d.cnki.gbfju.2022.000308.
[17] 李坤宏,江桂云,朱代兵.数控机床电动主轴WPD-TSNE-SVM模型故障诊断[J/OL].机械科学与技术:1-5[2023-08-26].
Li Kunhong, Jiang Guiyun, Zhu Daibing. Fault Diagnosis of CNC Machine Tool Electric Spindle using WPD-TSNE-SVM Model [J/OL]. Journal of Mechanical Science and Technology: 1-5 [2023-08-26].
[18] Danyang D,Dianchang W,Hongyuan F, et al. An automatic and integrated self-diagnosing system for the silting disease of drainage pipelines based on SSAE-TSNE and MS-LSTM[J]. Tunnelling and Underground Space Technology incorporating Trenchless Technology Research,2023,136.
