Dual-aligned Knowledge Distillation for Class-incremental Multi-fault Diagnosis of an Axial Piston Pump
DOI:
https://doi.org/10.13052/ijfp1439-9776.2712Keywords:
Axial piston pump, class-incremental learning, fault diagnosis, knowledge distillationAbstract
Multi-fault diagnosis of the axial piston pump plays a vital role in ensuring the safety and reliability of modern hydraulic transmission and control systems. Current intelligent fault diagnosis methods demonstrate effective performance but fail to generalize if new fault patterns occur. Simply fine-tuning these models only with newly collected data leads to the catastrophic forgetting problem, whereas retraining a new fault diagnosis model with the entire historical data is both resource-intensive and time-consuming. Therefore, a novel class-incremental learning method based on dual-aligned knowledge distillation is proposed for multi-fault diagnosis of the axial piston pump, which can continually learn new fault patterns and preserve fault diagnosis ability on old fault patterns with a limited amount of historical data. On the one hand, the consistency between output-logits of the previous model and that of the current one is enforced in the incremental learning process to mitigate catastrophic forgetting. On the other hand, intermediate feature relationships with different important weights are aligned to further retain fault diagnosis performance on old fault patterns. Both the comparison experiment and the ablation experiment demonstrate the effectiveness of the proposed method.
Downloads
References
S. Mukherjee, L. Shang, A. Vacca, ‘Numerical analysis and experimental validation of the coupled thermal effects in swashplate type axial piston machines’, Mechanical Systems and Signal Processing, vol. 220, p. 111673, 2024.
P. Michael, K. Stelson, D. Williams, H. Malik, ‘Dynamometer testing of hydraulic fluids in an axial piston pump under simulated backhoe loader trenching conditions’, Fluid Power Systems Technology, vol. V001T01A14, 2022.
L.V. Larsson, P. Krus, ‘Modelling of the swash plate control actuator in an axial piston pump for a hardware-in-the-loop simulation test rig’, Fluid Power Systems Technology, vol. V001T01A44, 2016.
I. Baus, R. Rahmfeld, A. Schumacher, H.C. Pedersen, ‘Systematic methodology for reliability analysis of components in axial piston units’, Fluid Power Systems Technology, vol. V001T01A9, 2019.
N. Keller, A. Sciancalepore, A. Vacca, ‘Condition Monitoring of an Axial Piston Pump on a Mini Excavator’, International Journal of Fluid Power, vol. 24, no. 02, pp. 171–206, 2023.
R. Ivantysyn, J. Weber, ‘Advancing Thermal Monitoring in Axial Piston Pumps: Simulation, Measurement, and Boundary Condition Analysis for Efficiency Enhancement’, International Journal of Fluid Power, pp. 547–590, 2024.
S. Wang, J. Xiang, Y. Zhong, H. Tang, ‘A data indicator-based deep belief networks to detect multiple faults in axial piston pumps’, Mechanical Systems and Signal Processing, vol. 112, pp. 154–170, 2018.
Y. He, H. Tang, Y. Ren, A. Kumar, ‘A semi-supervised fault diagnosis method for axial piston pump bearings based on DCGAN’, Measurement Science and Technology, vol. 32, no. 12, p. 125104, 2021.
S. Wang, H. Shuai, J. Hu, et al., ‘Few-shot fault diagnosis of axial piston pump based on prior knowledge-embedded meta learning vision transformer under variable operating conditions’, Expert Systems with Applications, vol. 269, p. 126452, 2025.
S. Liu, J. Zhang, W. Huang, F. Lyu, D. Wang, B. Xu, ‘Temporal–Spatial Attention Network: A Novel Axial Piston Pump Coupled Fault Diagnosis Method’, IEEE Transactions on Instrumentation and Measurement, vol. 73, pp. 1–15, 2024.
E. Belouadah, A. Popescu, I. Kanellos, ‘A comprehensive study of class incremental learning algorithms for visual tasks’, Neural Networks, vol. 135, pp. 38–54, 2021.
A.A. Rusu, N.C. Rabinowitz, G. Desjardins, et al., ‘Progressive neural networks’, arXiv preprint arXiv:1606.04671, 2016.
Y.-X. Wang, D. Ramanan, M. Hebert, ‘Growing a brain: Fine-tuning by increasing model capacity’, Proc. IEEE Conf. on Computer Vision and Pattern Recognition, pp. 2471–2480, 2017.
R. Aljundi, M. Lin, B. Goujaud, Y. Bengio, ‘Gradient based sample selection for online continual learning’, Advances in Neural Information Processing Systems, vol. 32, 2019.
A. Odena, C. Olah, J. Shlens, ‘Conditional image synthesis with auxiliary classifier gans’, Int. Conf. on Machine Learning, pp. 2642–2651, 2017.
J. Kirkpatrick, R. Pascanu, N. Rabinowitz, et. al., ‘Overcoming catastrophic forgetting in neural networks’, Proc. National Academy of Sciences, vol. 114, no. 13, pp. 3521–3526, 2017.
Z. Li, D. Hoiem, ‘Learning without forgetting’, IEEE Trans. on Pattern Analysis and Machine Intelligence, vol. 40, no. 12, pp. 2935–2947, 2017.
S.-A. Rebuffi, A. Kolesnikov, G. Sperl, C.H. Lampert, ‘iCaRL: Incremental classifier and representation learning’, Proc. IEEE Conf. on Computer Vision and Pattern Recognition, pp. 2001–2010, 2017.
G. Hinton, O. Vinyals, J. Dean, ‘Distilling the Knowledge in a Neural Network’, arXiv preprint arXiv:1503.02531, 2015.
G.M. Van de Ven, T. Tuytelaars, A.S. Tolias, ‘Three types of incremental learning’, Nature Machine Intelligence, vol. 4, no. 12, pp. 1185–1197, 2022.
S. Yan, H. Shao, X. Wang, J. Wang, ‘Few-shot class-incremental learning for system-level fault diagnosis of wind turbine’, IEEE/ASME Trans. on Mechatronics, 2024.
M. Kang, J. Park, B. Han, ‘Class-incremental learning by knowledge distillation with adaptive feature consolidation’, Proc. IEEE/CVF Conf. on Computer Vision and Pattern Recognition, pp. 16071–16080, 2022.
M. Ji, G. Peng, S. Li, et al., ‘A neural network compression method based on knowledge-distillation and parameter quantization for the bearing fault diagnosis’, Applied Soft Computing, vol. 127, p. 109331, 2022.
