Estimating Spall Severity in Rolling Element Bearings: A Supervised Learning Approach With Naturally Progressing Spalls
##plugins.themes.bootstrap3.article.main##
##plugins.themes.bootstrap3.article.sidebar##
Abstract
Estimating the severity of localized defects in rolling element bearings is critical for accurate Remaining Useful Life (RUL) estimation, yet it remains challenging under non-stationary operating conditions with fluctuating speeds. Existing data-driven methods struggle to generalise due to the lack of high-fidelity, damage-progression data, and susceptibility to machine-specific structural transfer functions. A Siamese Transformer based neural network is utilized to predict continuous spall size directly from concurrent vibration measurements across three selected operating speeds, reducing the need for long-term trending. Using the amplitudes at the ball-pass frequency and its harmonics from spectra, and a data augmentation strategy, the proposed approach aims to decouple the fault signature from the system transfer function. Trained on a single run-to-failure dataset of one N209 ECP bearing with automated ground truth sizing for labels, the network acts on a regression target to learn the mapping between spectral features and the defect size. Preliminary results suggest that this proof-of-concept framework shows promise for generalization to unseen speed combinations and synthetic transfer function profiles within the scope of the studied experiment.
How to Cite
##plugins.themes.bootstrap3.article.details##
Rolling element bearing diagnostics, Spall severity estimation, Siamese neural network, Transfer function augmentation
Bublil, T., Taal, C., Maljaars, B., Klein, R., & Bortman,
J. (2024). Labeling algorithm for outer-race faults in bearings based on load signal. PHM Society European Conference, 8(1), 7–7.
Bublil, T., Taal, C., Maljaars, B., Matania, O., Klein, R., & Bortman, J. (2025). Experimental study on condition indicators for severity estimation of growing spalls in bearings. Mechanical Systems and Signal Processing, 236, 113019.
Chen, Y., Liu, X., Rao, M., Qin, Y., Wang, Z., & Ji,Y. (2025). Explicit speed-integrated lstm network for non-stationary gearbox vibration representation and fault detection under varying speed conditions. Reliability Engineering and System Safety, 254, 110596. doi: 10.1016/j.ress.2024.110596
Chopra, S., Hadsell, R., & LeCun, Y. (2005). Learning a similarity metric discriminatively, with application to face verification. 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR’05), 1, 539–546.
Dong, Y., Li, Y., Zheng, H., Wang, R., & Xu, M. (2022). A new dynamic model and transfer learning based intelligent fault diagnosis framework for rolling element bearings race faults: Solving the small sample problem. ISA Transactions, 121, 327-348. doi:10.1016/j.isatra.2021.03.042
Fan, C., Wang, P., Zhang, Y., Ma, H., Li, X., & Wang, Q. (2025). Digital twin assisted degradation assessment of bearing cage performance. IEEE Transactions on Industrial Informatics, 21(7), 5171-5181. doi:10.1109/TII.2025.3552655
Koch, G., Zemel, R., & Salakhutdinov, R. (2015). Siamese neural networks for one-shot image recognition. ICML deep learning workshop, 2.
Kotzalas, M. N., & Harris, T. A. (2001). Fatigue failure progression in ball bearings. J. Trib., 123(2), 238–242.
Liefstingh, M., Taal, C., Restrepo, S. E., & Azarfar, A. (2021). Interpretation of deep learning models in bearing fault diagnosis. Annual Conference of the PHM Society, 13(1).
Mason, J. K., Trivedi, H. K., & Rosado, L. (2017). Spall propagation characteristics of refurbished vimvar aisi m50 angular contact bearings. Journal of Failure Analysis and Prevention, 17(3), 426–439.
Ngo-The, D., Taal, C., & Antoni, J. (2025). Bearing spall size estimation under varying speed conditions. Annual Conference of the PHM Society, 17(1). doi:10.36001/phmconf.2025.v17i1.4354
Ni, Q., Ji, J., Halkon, B., Feng, K., & Nandi, A. K. (2023). Physics-informed residual network (piresnet) for rolling element bearing fault diagnostics. Mechanical Systems and Signal Processing, 200, 110544. doi: 10.1016/j.ymssp.2023.110544 Rosado, L., Forster, N. H., Thompson, K. L., & Cooke,
J. W. (2009). Rolling contact fatigue life and spall propagation of aisi m50, m50nil, and aisi 52100, part i: Experimental results. Tribology Transactions, 53(1), 29–41.
Shen, S., Lu, H., Sadoughi, M., Hu, C., Nemani, V., Thelen, A., . . . Kenny, S. (2021). A physics-informed deep learning approach for bearing fault detection. Engineering Applications of Artificial Intelligence, 103, 104295. doi: 10.1016j.engappai.2021.104295
Song, Y., & Liu, P. (2024). Federated domain generalization for intelligent fault diagnosis based on pseudo-siamese network and robust global model aggregation. International Journal of Machine Learning and Cybernetics, 15(2), 685–696. doi:10.1007/s13042-023-01934-2
Vu, M.-H., Nguyen, V.-Q., Tran, T.-T., Pham, V.-T., & Lo, M.-T. (2024). Few-shot bearing fault diagnosis via ensembling transformer-based model with mahalanobis distance metric learning from multiscale features. IEEE Transactions on Instrumentation and Measurement, 73, 1-18. doi:10.1109/TIM.2024.3381270
Xu, M., Zhang, H., Miao, H., Hao, J., Li, C., Song, W., . . . Zhang, Y. (2023). Model-based vibration response analysis and experimental verification of lathe spindle-housing-belt system with rubbing. Mechanical Systems and Signal Processing, 186, 109841. doi: 10.1016/j.ymssp.2022.109841
Yang, X., Yang, J., Jin, Y., & Liu, Z. (2024). A new method for bearing fault diagnosis across machines based on envelope spectrum and conditional metric learning. Sensors, 24(9). doi: 10.3390/s24092674
Zhang, H., Borghesani, P., Smith, W. A., Randall, R. B., Shahriar, M. R., & Peng, Z. (2021). Tracking the natural evolution of bearing spall size using cyclic natural frequency perturbations in vibration signals. Mechanical Systems and Signal Processing, 151, 107376. doi: 10.1016/j.ymssp.2020.107376
This work is licensed under a Creative Commons Attribution 3.0 Unported License.
The Prognostic and Health Management Society advocates open-access to scientific data and uses a Creative Commons license for publishing and distributing any papers. A Creative Commons license does not relinquish the author’s copyright; rather it allows them to share some of their rights with any member of the public under certain conditions whilst enjoying full legal protection. By submitting an article to the International Conference of the Prognostics and Health Management Society, the authors agree to be bound by the associated terms and conditions including the following:
As the author, you retain the copyright to your Work. By submitting your Work, you are granting anybody the right to copy, distribute and transmit your Work and to adapt your Work with proper attribution under the terms of the Creative Commons Attribution 3.0 United States license. You assign rights to the Prognostics and Health Management Society to publish and disseminate your Work through electronic and print media if it is accepted for publication. A license note citing the Creative Commons Attribution 3.0 United States License as shown below needs to be placed in the footnote on the first page of the article.
First Author et al. This is an open-access article distributed under the terms of the Creative Commons Attribution 3.0 United States License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.