The bogie plays a crucial role in ensuring the safe and comfortable operation of high-speed trains (HST). Fault localization is a valuable approach for scheduling vehicle maintenance plans and ensuring the safe operation of high-speed rail transportation. However, in real train operating conditions, two limitations complicate the accurate fault localization of bogies using deep learning: substantial railway differences and a shortage of labeled data. This paper proposes a novel multi-sensor graph transfer network (3MAGTN), equipped with a multi-attention mechanism, to achieve end-to-end localization of faults in HST bogies across various running railways. The designed multi-attention mechanism exhibits multi-level, multi-dimensional, and multi-focus attention capabilities, while operating at three hierarchical levels: worker, extractor, and optimizer. It enhances fault location accuracy at the local scale by prioritizing essential features and improves cross-railway capability at the global scale. Moreover, an adaptive multi-objective optimization strategy is designed to distinguish perplexing data and utilize unlabeled data efficiently. The effectiveness of 3MAGTN is demonstrated by sophisticated and imbalanced mutual transfer experiments across three vehicle running railways, with the average location and identification accuracy of 30 tests being 86.10% and 92.71%, respectively.
ChenDLinYLiW, et al. (2020a) Measuring and relieving the over-smoothing problem for graph neural networks from the topological view. Proceedings of the AAAI Conference on Artificial Intelligence34: 3438–3445.
2.
ChenHJiangBDingSX, et al. (2020b) Data-driven fault diagnosis for traction systems in high-speed trains: a survey, challenges, and perspectives. IEEE Transactions on Intelligent Transportation Systems23(3): 1700–1716.
3.
ChenMZhaoSLiuH, et al. (2020c) Adversarial-learned loss for domain adaptation. Proceedings of the AAAI Conference on Artificial Intelligence34: 3521–3528.
4.
GaninYLempitskyV (2015) Unsupervised domain adaptation by backpropagation. In: International Conference on Machine Learning, Vienna, Austria, 13 Jul, 2025 – Sat, 19 Jul, 2025, 1180–1189.
5.
GhorveiMKavianpourMBeheshtiMT, et al. (2023) Spatial graph convolutional neural network via structured subdomain adaptation and domain adversarial learning for bearing fault diagnosis. Neurocomputing517: 44–61.
6.
HeDLaoZJinZ, et al. (2023) Train bearing fault diagnosis based on multi-sensor data fusion and dual-scale residual network. Nonlinear Dynamics111(16): 14901–14924.
7.
HeCShiHLiR, et al. (2024a) Interpretable modulated differentiable stft and physics-informed balanced spectrum metric for freight train wheelset bearing cross-machine transfer fault diagnosis under speed fluctuations. Advanced Engineering Informatics62: 102568–102593.
8.
HeCShiHLiuX, et al. (2024b) Interpretable physics-informed domain adaptation paradigm for cross-machine transfer diagnosis. Knowledge-Based Systems288: 111499–111512.
9.
HuangGLiuZVan Der MaatenL, et al. (2017) Densely connected convolutional networks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 4700–4708.
10.
HuangDLiSQinN, et al. (2021) Fault diagnosis of high-speed train bogie based on the improved-CEEMDAN and 1-D CNN algorithms. IEEE Transactions on Instrumentation and Measurement70: 1–11.
11.
JiWLvDLuoS, et al. (2024) Multiple models-based fault tolerant control of levitation module of maglev vehicles against partial actuator failures. IEEE Transactions on Vehicular Technology2024: 1–11.
12.
JiaXQinNHuangD, et al. (2022) A clustered blueprint separable convolutional neural network with high precision for high-speed train bogie fault diagnosis. Neurocomputing500: 422–433.
13.
JiaXQinNHuangD, et al. (2024a) GraphFL: graph federated learning for fault localization of multirailway high-speed train suspension systems. IEEE Transactions on Instrumentation and Measurement73: 1–11.
14.
JiaXYangJLuK, et al. (2024b) Enhanced robust motion control based on unknown system dynamics estimator for robot manipulators. In: 2024 IEEE International Conference on Robotics and Automation (ICRA). Piscataway: IEEE, 3514–3519.
15.
KerivenN (2022) Not too little, not too much: a theoretical analysis of graph (over) smoothing. Advances in Neural Information Processing Systems35: 2268–2281.
16.
LiCZhangSQinY, et al. (2020) A systematic review of deep transfer learning for machinery fault diagnosis. Neurocomputing407: 121–135.
17.
LiTZhaoZSunC, et al. (2021) Domain adversarial graph convolutional network for fault diagnosis under variable working conditions. IEEE Transactions on Instrumentation and Measurement70: 1–10.
18.
LiuRYangBZioE, et al. (2018) Artificial intelligence for fault diagnosis of rotating machinery: a review. Mechanical Systems and Signal Processing108: 33–47.
19.
LiuSJohnsEDavisonAJ (2019) End-to-end multi-task learning with attention. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 1871–1880.
20.
LiuGShenWGaoL, et al. (2022) Knowledge transfer in fault diagnosis of rotary machines. IET Collaborative Intelligent Manufacturing4(1): 17–34.
21.
ManJDongHJiaL, et al. (2022a) AttGGCN model: a novel multi-sensor fault diagnosis method for high-speed train bogie. IEEE Transactions on Intelligent Transportation Systems23(10): 19511–19522.
22.
ManJDongHYangX, et al. (2022b) GCG: graph convolutional network and gated recurrent unit method for high-speed train axle temperature forecasting. Mechanical Systems and Signal Processing163: 108102–108125.
23.
ManJDongHJiaL, et al. (2023) An adaptive multisensor fault diagnosis method for high-speed train bogie. IEEE Transactions on Intelligent Transportation Systems24(6): 6292–6306.
24.
QinNWuBHuangD, et al. (2022) Stepwise adaptive convolutional network for fault diagnosis of high-speed train bogie under variant running speeds. IEEE Transactions on Industrial Informatics18(12): 8389–8398.
25.
TangYZhangXQinG, et al. (2022) Graph cardinality preserved attention network for fault diagnosis of induction motor under varying speed and load condition. IEEE Transactions on Industrial Informatics18(6): 3702–3712.
26.
WangPLiuJZhouJ, et al. (2022) Cross-domain fault diagnosis of rotating machinery based on graph feature extraction. Measurement Science and Technology34(2): 025116–025130.
27.
WuXRakhejaSQuS, et al. (2018) Dynamic responses of a high-speed railway car due to wheel polygonalisation. Vehicle System Dynamics56(12): 1817–1837.
28.
WuZChenHHanJ, et al. (2023) Graph mutual mapping transfer networks for machine fault diagnosis. IEEE Transactions on Industrial Informatics20(1): 995–1006.
29.
XuJKeHChenZ, et al. (2022) Oversmoothing relief graph convolutional network-based fault diagnosis method with application to the rectifier of high-speed trains. IEEE Transactions on Industrial Informatics19(1): 771–779.
30.
YangBWangTXieJ, et al. (2023) Deep adversarial hybrid domain-adaptation network for varying working conditions fault diagnosis of high-speed train bogie. IEEE Transactions on Instrumentation and Measurement72: 1–10.
31.
YangFLiuJHuaC, et al. (2024) Early fault diagnosis strategy for high-speed train suspension systems based on model-agnostic meta-learning. Vehicle System Dynamics62(10): 2510–2532.
32.
YeYHuangPZhangY (2022) Deep learning-based fault diagnostic network of high-speed train secondary suspension systems for immunity to track irregularities and wheel wear. Railway Engineering Science30(1): 96–116.
33.
ZhangYQinNHuangD, et al. (2022) High-accuracy and adaptive fault diagnosis of high-speed train bogie using dense-squeeze network. IEEE Transactions on Vehicular Technology71(3): 2501–2510.
34.
ZhangDXieMYangJ, et al. (2023) Multi-sensor graph transfer network for health assessment of high-speed rail suspension systems. IEEE Transactions on Intelligent Transportation Systems24(9): 9425–9434.
35.
ZhaoXSunYLiY, et al. (2024) Applications of machine learning in real-time control systems: a review. Measurement Science and Technology36(1): 012003–012026.
36.
ZhuYZhuangFWangJ, et al. (2021) Deep subdomain adaptation network for image classification. IEEE Transactions on Neural Networks and Learning Systems32(4): 1713–1722.
37.
ZhuZLeiYQiG, et al. (2023) A review of the application of deep learning in intelligent fault diagnosis of rotating machinery. Measurement (Mahwah, N. J.)206: 112346–112370.
