SRMANet: Toward an Interpretable Neural Network with Multi-Attention Mechanism for Gearbox Fault Diagnosis
Source abstract
Abstract
Deep neural network (DNN), with the capacity for feature inference and nonlinear mapping, has demonstrated its effectiveness in end-to-end fault diagnosis. However, the intermediate learning process of the DNN architecture is invisible, making it an uninterpretable black-box model. In this paper, a stacked residual multi-attention network (SRMANet) is proposed as a means of feature extraction of vibration signals, and visualizing the model training process, designing Squeeze-excitation residual (SE-Res) blocks to obtain additive features with minimal redundancy and sparsity. This study recommends the use of the attention fusion unit to ensure the interpretability of the model and ultimately to obtain representative features. By feeding the output gradient of the attention layer back to the original signal, the key feature components in the time domain signal can be effectively captured. Finally, the interpretability, identification accuracy and adaptability of the model under different operating conditions are verified on 12 different fault tasks in the planetary gearbox.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Machine Fault Diagnosis Techniques · Fault Detection and Control Systems · Industrial Vision Systems and Defect Detection
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
School of Data Science and Technology, North University of China, Taiyuan 030051, China
School of Mechanical Engineering, North University of China, Taiyuan 030051, China
School of Energy and Power Engineering, North University of China, Taiyuan 030051, China
Funding
Young Science Foundation of Shanxi province, China · Innovation Project of Postgraduate Education in Shanxi Province in 2020, China · Key R&D program of Shanxi Province (International Cooperation) · Natural Science Foundation of Shanxi Province · Research Project Supported by Shanxi Scholarship Council of China
References (43)
Show source-provided references
1. Multicomponent signal analysis: Interwoven Fourier decomposition method10.1016/j.dsp.2020.102771 ↗
2. Enhanced periodic mode decomposition and its application to composite fault diagnosis of rolling bearings10.1016/j.isatra.2021.07.014 ↗
3. Zhou, W., Feng, Z., Xu, Y.F., Wang, X., and Lv, H. (2020). Empirical Fourier Decomposition: An Accurate Adaptive Signal Decomposition Method. arXiv.
4. Study on the Dynamic Problems of Double-Disk Rotor System Supported by Deep Groove Ball Bearing10.1155/2019/8120569 ↗
5. Research on vibration performance of the nonlinear combined support-flexible rotor system10.1007/s11071-019-05176-2 ↗
6. The Fourier decomposition method for nonlinear and non-stationary time series analysis10.1098/rspa.2016.0871 ↗
7. Deep separable convolutional network for remaining useful life prediction of machinery10.1016/j.ymssp.2019.106330 ↗
8. A Case study of conditional deep convolutional generative adversarial networks in machine fault diagnosis10.1007/s10845-020-01579-w ↗
9. Ensemble deep contractive auto-encoders for intelligent fault diagnosis of machines under noisy environment10.1016/j.knosys.2020.105764 ↗
10. Long short-term memory neural network with weight amplification and its application into gear remaining useful life prediction10.1016/j.engappai.2020.103587 ↗
11. Actual bearing compound fault diagnosis based on active learning and decoupling attentional residual network10.1016/j.measurement.2020.108500 ↗
12. A diagnosis framework based on domain adaptation for bearing fault diagnosis across diverse domains10.1016/j.isatra.2019.08.040 ↗
13. He, K., Zhang, X., Ren, S., and Sun, J. (2016, January 27–30). Deep Residual Learning for Image Recognition. Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA.10.1109/CVPR.2016.90 ↗
14. ACDIN: Bridging the gap between artificial and real bearing damages for bearing fault diagnosis10.1016/j.neucom.2018.03.014 ↗
15. Deep residual learning with demodulated time-frequency features for fault diagnosis of planetary gearbox under nonstationary running conditions10.1016/j.ymssp.2019.02.055 ↗
16. Deep residual learning-based fault diagnosis method for rotating machinery10.1016/j.isatra.2018.12.025 ↗
17. Deep Residual Networks With Adaptively Parametric Rectifier Linear Units for Fault Diagnosis10.1109/TIE.2020.2972458 ↗
18. multiple wavelet regularized deep residual networks for fault diagnosis10.1016/j.measurement.2019.107331 ↗
19. A new hybrid deep signal processing approach for bearing fault diagnosis using vibration signals10.1016/j.neucom.2018.12.088 ↗
20. Enhanced sparse filtering with strong noise adaptability and its application on rotating machinery fault diagnosis10.1016/j.neucom.2020.02.042 ↗
21. A New Interpretable Learning Method for Fault Diagnosis of Rolling Bearings
22. Collaborative deep learning framework for fault diagnosis in distributed complex systems10.1016/j.ymssp.2021.107650 ↗
23. Interpretable Convolutional Neural Network Through Layer-wise Relevance Propagation for Machine Fault Diagnosis10.1109/JSEN.2019.2958787 ↗
24. One-dimensional fully decoupled networks for fault diagnosis of planetary gearboxes10.1016/j.ymssp.2019.106482 ↗
25. Robust Interpretable Deep Learning for Intelligent Fault Diagnosis of Induction Motors10.1109/TIM.2019.2932162 ↗
26. TScatNet: An Interpretable Cross-Domain Intelligent Diagnosis Model with Antinoise and Few-Shot Learning Capability
27. WaveletKernelNet: An Interpretable Deep Neural Network for Industrial Intelligent Diagnosis10.1109/TSMC.2020.3048950 ↗
28. Stacked sparse autoencoders monitoring model based on fault-related variable selection10.1007/s00500-020-05384-8 ↗
29. Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A.N., Kaiser, Ł., and Polosukhin, I. (2017, January 4–7). Attention Is All You Need. Proceedings of the Advances in Neural Information Processing Systems 30 (NIPS 2017), Long Beach, CA, USA.
30. Squeeze-and-Excitation Networks10.1109/TPAMI.2019.2913372 ↗
31. Adaptive Densely Connected Convolutional Auto-Encoder-Based Feature Learning of Gearbox Vibration Signals10.1109/TIM.2020.3039612 ↗
32. Fault detection and identification of rolling element bearings with attentive dense CNN10.1016/j.neucom.2020.04.143 ↗
33. AKRNet: A novel convolutional neural network with attentive kernel residual learning for feature learning of gearbox vibration signals10.1016/j.neucom.2021.02.055 ↗
34. Fault diagnosis of rolling bearings using an improved multi-scale convolutional neural network with feature attention mechanism10.1016/j.isatra.2020.10.054 ↗
35. Intelligent Bearing Fault Diagnosis Using Multi-Head Attention-Based CNN10.1016/j.promfg.2020.07.005 ↗
36. LEFE-Net: A Lightweight Efficient Feature Extraction Network With Strong Robustness for Bearing Fault Diagnosis10.1109/TIM.2021.3067187 ↗
37. Interpreting network knowledge with attention mechanism for bearing fault diagnosis10.1016/j.asoc.2020.106829 ↗
38. Understanding and improving deep learning-based rolling bearing fault diagnosis with attention mechanism10.1016/j.sigpro.2019.03.019 ↗
39. Learning the mean: A neural network approach10.1016/j.neucom.2011.08.017 ↗
40. Reviews of bearing vibration measurement using fast fourier transform and enhanced fast fourier transform algorithms10.1177/1687814018816751 ↗
41. Dropout: A Simple Way to Prevent Neural Networks from Overfitting
42. Guo, M.-H., Liu, Z.-N., Mu, T.-J., and Hu, S.-M. (2021). Beyond Self-Attention: External Attention Using Two Linear Layers for Visual Tasks. arXiv.10.1109/TPAMI.2022.3211006 ↗
43. Deep learning algorithms for rotating machinery intelligent diagnosis: An open source benchmark study10.1016/j.isatra.2020.08.010 ↗