In industrial practical applications, the vibration signals of rolling bearings are often overwhelmed by strong background noise, resulting in a significant decline in the performance of traditional deep learning models. Although the Transformer is good at capturing global dependencies, its standard self-attention mechanism is sensitive to noise and has a fixed calculation path, making it difficult to adaptively handle variable features. To address these challenges, this paper proposes a novel fault diagnosis framework called Dynamic Routing Spectrally-Aware Network (DRSAN). This framework integrates two key innovations. First, the Spectrum Enhancement Attention (SEA) mechanism is designed, which incorporates frequency domain weights to focus on key fault frequencies and suppress noise. Second, the Dynamic Feature Routing (AMoE) mechanism routes features adaptively using a Mixture of Experts (MoE) and discards the rigid computational path. Finally, the features extracted by the above modules are fused with fine local features to complete the final fault classification. Validation on an open bearing dataset is given to illustrate the effectiveness of presented framework.
BaJLKirosJRHintonGE (2016) Layer normalization. arXiv:1607.06450, arXiv. Available at. https://arxiv.org/abs/1607.06450accessed 19 January 2026.
2.
CerradaMSanchezRVLiC, et al. (2018) A review on data-driven fault severity assessment in rolling bearings. Mechanical Systems and Signal Processing99: 169–196. https://doi.org/10.1016/j.ymssp.2017.06.012
3.
DosovitskiyABeyerLKolesnikovA, et al. (2021) An image is worth 16x16 words: transformers for image recognition at scale. arXiv:2010.11929.
4.
FedusWZophBShazeerN (2022) Switch transformers: scaling to trillion parameter models with simple and efficient sparsity. Journal of Machine Learning Research23(120): 1–39.
5.
GaoZWangYLiX, et al. (2024) Twins transformer: rolling bearing fault diagnosis based on cross-attention fusion of time and frequency domain features. Measurement Science and Technology35(9): 096113. https://doi.org/10.1088/1361-6501/ad53f1
6.
HendrycksDGimpelK (2023) Gaussian error linear units (GELUs). arXiv:1606.08415, arXiv. Available at. https://arxiv.org/abs/1606.08415accessed 19 January 2026.
7.
HouYWangJChenZ, et al. (2023) Diagnosisformer: an efficient rolling bearing fault diagnosis method based on improved transformer. Engineering Applications of Artificial Intelligence124: 106514. https://doi.org/10.1016/j.engappai.2023.106507
8.
InceTKiranyazSErenL, et al. (2016) Real-time motor fault detection by 1-D convolutional neural networks. IEEE Transactions on Industrial Electronics63(11): 7067–7075. https://doi.org/10.1109/tie.2016.2582729
9.
KingmaDPBaJ (2017) Adam: a method for stochastic optimization. arXiv:1412.6980. arXiv. Available at. https://arxiv.org/abs/1412.6980accessed 19 January 2026.
10.
LeiYYangBJiangX, et al. (2020) Applications of machine learning to machine fault diagnosis: a review and roadmap. Mechanical Systems and Signal Processing138: 106587. https://doi.org/10.1016/j.ymssp.2019.106587
11.
LiCSanchezR-VZuritaG, et al. (2015) Multimodal deep support vector classification with homologous features and its application to gearbox fault diagnosis. Neurocomputing168: 119–127. https://doi.org/10.1016/j.neucom.2015.06.008
12.
LiuZLinYCaoY, et al. (2021) Swin transformer: hierarchical vision transformer using shifted windows. arXiv preprint arXiv:2103.14030.
13.
LiuRWangXSuC, et al. (2024) Bearing fault diagnosis method based on variational mode decomposition optimized by CS-PSO. Journal of Vibration and Control30(5–6): 973–987. https://doi.org/10.1177/10775463231154448
14.
MadryAMakelovASchmidtL, et al. (2019) Towards deep learning models resistant to adversarial attacks. arXiv:1706.06083, arXiv. Available at. https://arxiv.org/abs/1706.06083accessed 30 December 2025.
15.
MenZLiYTangW, et al. (2025) A new multi-modal time series transformation method and multi-scale convolutional attention network for railway wagon bearing fault diagnosis. Journal of Vibration and Control31(17–18): 3675–3690. https://doi.org/10.1177/10775463241276024
16.
NairVHintonGE (n.d.) Rectified linear units improve restricted boltzmann machines.
17.
PeiXZhengXWuJ (2021) Rotating machinery fault diagnosis through a transformer convolution network subjected to transfer learning. IEEE Transactions on Instrumentation and Measurement70: 1–11. https://doi.org/10.1109/tim.2021.3119137
18.
PengGZhengJTongB, et al. (2025) Adaptive fast iterative filter Holo-spectrum analysis and its applications to fault diagnosis of rolling bearing. Journal of Vibration and Control31(17–18): 3918–3930. https://doi.org/10.1177/10775463241281763
19.
RandallRBAntoniJ (2011) Rolling element bearing diagnostics - a tutorial. Mechanical Systems and Signal Processing25(2): 485–520. https://doi.org/10.1016/j.ymssp.2010.07.017
20.
RavanelliMBengioY (2019) Speaker recognition from raw waveform with SincNet. arXiv:1808.00158, arXiv. Available at. https://arxiv.org/abs/1808.00158accessed 19 January 2026.
21.
RenSLouX (2025) Rolling bearing fault diagnosis method based on SWT and improved vision transformer. Sensors25(7): 2090. https://doi.org/10.3390/s25072090
22.
ShaoSMcAleerSYanR, et al. (2019) Highly accurate machine fault diagnosis using deep transfer learning. IEEE Transactions on Industrial Informatics15(4): 2446–2455. https://doi.org/10.1109/tii.2018.2864759
23.
ShazeerNMirhoseiniAMaziarzK, et al. (2017) Outrageously large neural networks: the sparsely-gated mixture-of-experts layer. arXiv:1701.06538. arXiv. Available at. https://arxiv.org/abs/1701.06538accessed 19 January 2026.
24.
SmithWARandallRB (2015) Rolling element bearing diagnostics using the case Western reserve university data: a benchmark study. Mechanical Systems and Signal Processing64–65: 100–131. https://doi.org/10.1016/j.ymssp.2015.04.021
25.
SrivastavaNHintonGKrizhevskyA, et al. (n.d.) Dropout: a simple way to prevent neural networks from overfitting.
26.
VaswaniAShazeerNParmarN, et al. (2017) Attention is all you need. In: Advances in Neural Information Processing Systems. Curran Associates, Inc, Vol. 2017, 5998–6008.
27.
WangBLeiYNaipengL, et al. (2020) A hybrid prognostics approach for estimating remaining useful life of rolling element bearings. IEEE Transactions on Reliability69(1): 401–412. https://doi.org/10.1109/tr.2018.2882682
28.
WenLLiXGaoL, et al. (2018) A new convolutional neural network-based data-driven fault diagnosis method. IEEE Transactions on Industrial Electronics65(7): 5990–5998. https://doi.org/10.1109/tie.2017.2774777
29.
WuWWangQYuS, et al. (2021) Outside box and contactless palm vein recognition based on a wavelet denoising ResNet. IEEE Access9: 82471–82484. https://doi.org/10.1109/access.2021.3086811
30.
XiaoYShaoHFengM, et al. (2023) Towards trustworthy rotating machinery fault diagnosis via attention uncertainty in transformer. Journal of Manufacturing Systems70: 186–201. https://doi.org/10.1016/j.jmsy.2023.07.012
31.
YanJZhuXWangX, et al. (2025) A new fault diagnosis method for rolling bearings with the basis of swin transformer and generalized S transform. Mathematics13(1): 45. https://doi.org/10.3390/math13010045
32.
YeTDongLXiaY, et al. (2025) Differential transformer. arXiv preprint arXiv:2410.05258.
33.
YuanMWuYLinL (2016) Fault diagnosis and remaining useful life estimation of aero engine using LSTM neural network. In: 2016 IEEE International Conference on Aircraft Utility Systems (AUS), Beijing, China, 11–21 June 2016, IEEE, 135–140.
34.
ZhaoRYanRChenZ, et al. (2019) Deep learning and its applications to machine health monitoring. Mechanical Systems and Signal Processing115: 213–237. https://doi.org/10.1016/j.ymssp.2018.05.050
