Statistical Knowledge Integration into Neural Networks: Novel Neuron Units for Bearing Prognostics

##plugins.themes.bootstrap3.article.main##

##plugins.themes.bootstrap3.article.sidebar##

Published Jun 27, 2024
Thomas Pioger Marcia Baptista

Abstract

Prognostics and Health Management (PHM) is a framework that assesses the health condition of complex engineering assets to ensure proper reliability, availability, and maintenance. PHM can be used to determine how long a machine can function before failure by predicting the Remaining Useful Life (RUL). Neural networks have been used for RUL prediction, but these data-driven models rely solely on data to explicitly integrate knowledge. Recently, authors have proposed physics-informed neural networks (PINNs) to address this limitation. PINNs are neural networks that incorporate expert knowledge and physics in different ways (observational,inductive, and learning bias). Despite their significance, these models tend to be case-dependent and challenging to configure. In this work, we propose statistical neuron units that can be integrated into any neural network. The proposed neuron units extract features from raw data using various statistical functions. Importantly, these modules can be located in different parts of the neural network, and they can be optimized automatically by backpropagating the modules’ weights during training. In a study involving bearing degradation behavior, we compare a classical neural network with our modular
version. Our proposed RUL estimation model outperformed the baseline, with a reduction of 13% in the root mean square
error and a reduction of 7% in the mean absolute error. We also observe an increase of 40% and 21% for the α − λ accuracy metric for an α equal to 0.1 and 0.2 respectively. Our code is available publicly on Github.

How to Cite

Pioger, T. ., & Baptista, M. (2024). Statistical Knowledge Integration into Neural Networks: Novel Neuron Units for Bearing Prognostics. PHM Society European Conference, 8(1), 14. https://doi.org/10.36001/phme.2024.v8i1.4088
Abstract 164 | PDF Downloads 142

##plugins.themes.bootstrap3.article.details##

Keywords

Feature extraction, knowledge integration, opotimization of parameters, interpretability, accuracy, modularity, neural network

References
Amer, M., & Maul, T. (2019). A review of modularization techniques in artificial neural networks. Artificial Intelligence Review, 52, 527–561.

Cai, H., Feng, J., Li, W., Hsu, Y.-M., & Lee, J. (2020). Similarity-based particle filter for remaining useful life prediction with enhanced performance. Applied Soft Computing, 94, 106474.

Castillo-Bolado, D., Guerra-Artal, C., & Hern´andez-Tejera, M. (2021). Design and independent training of composable and reusable neural modules. Neural Networks, 139, 294304.

Chao, M. A., Kulkarni, C., Goebel, K., & Fink, O. (2022). Fusing physics-based and deep learning models for prognostics. Reliability Engineering & System Safety, 217, 107961.

Chen, X., Ma, M., Zhao, Z., Zhai, Z., & Mao, Z. (2022). Physics-informed deep neural network for bearing prognosis with multisensory signals. Journal of Dynamics, Monitoring and Diagnostics, 200–207.

Cubillo, A., Perinpanayagam, S., & Esperon-Miguez, M. (2016). A review of physics-based models in prognostics: Application to gears and bearings of rotating machinery. Advances in Mechanical Engineering, 8(8), 1687814016664660.

Cui, L., Wang, X., Xu, Y., Jiang, H., & Zhou, J. (2019). A novel switching unscented kalman filter method for remaining useful life prediction of rolling bearing. Measurement, 135, 678–684.

Dash, T., Chitlangia, S., Ahuja, A., & Srinivasan, A. (2022). A review of some techniques for inclusion of domainknowledge into deep neural networks. Scientific Reports, 12(1), 1040.

Dourado, A. D., & Viana, F. (2020). Physics-informed neural networks for bias compensation in corrosion-fatigue. In Aiaa scitech 2020 forum (p. 1149).

Elfring, J., Torta, E., & Van De Molengraft, R. (2021). Particle filters: A hands-on tutorial. Sensors, 21(2), 438.

Fan, F.-L., Xiong, J., Li, M., & Wang, G. (2021). On interpretability of artificial neural networks: A survey. IEEE Transactions on Radiation and Plasma Medical Sciences, 5(6), 741–760.

Faroughi, S. A., Pawar, N., Fernandes, C., Raissi, M., Das, S., Kalantari, N. K., & Mahjour, S. K. (2022). Physics-guided, physics-informed, and physics-encoded neural networks in scientific computing. arXiv preprint arXiv:2211.07377.

Ferreira, C., & Gonc¸alves, G. (2022). Remaining useful life prediction and challenges: A literature review on the use of machine learning methods. Journal of Manufacturing Systems, 63, 550–562.

Ge, M.-F., Liu, Y., Jiang, X., & Liu, J. (2021). A review on state of health estimations and remaining useful life prognostics of lithium-ion batteries. Measurement, 174, 109057.

Guo, J., Li, Z., & Li, M. (2019). A review on prognostics methods for engineering systems. IEEE Transactions on Reliability, 69(3), 1110–1129.

Hakami, A. (2024). Strategies for overcoming data scarcity, imbalance, and feature selection challenges in machine learning models for predictive maintenance. Scientific Reports, 14(1), 9645.

Hasib, S. A., Islam, S., Chakrabortty, R. K., Ryan, M. J., Saha, D. K., Ahamed, M. H., . . . Badal, F. R. (2021). A comprehensive review of available battery datasets, rul prediction approaches, and advanced battery management. IEEE Access, 9, 86166-86193. doi: 10.1109/ACCESS .2021.3089032

Huang, A. J., & Agarwal, S. (2023). On the limitations of physics-informed deep learning: Illustrations using first order hyperbolic conservation law-based traffic flow models. IEEE Open Journal of Intelligent Transportation Systems.

Jia, C., & Zhang, H. (2019). Rul prediction: Reducing statistical model uncertainty via bayesian model aggregation. In 2019 caa symposium on fault detection, supervision and safety for technical processes (safeprocess) (p. 602-607). doi: 10.1109/SAFEPROCESS45799.2019.9213433

Kang, Z., Catal, C., & Tekinerdogan, B. (2021). Remaining useful life (rul) prediction of equipment in production lines using artificial neural networks. Sensors, 21(3), 932.

Karniadakis, G. E., Kevrekidis, I. G., Lu, L., Perdikaris, P., Wang, S., & Yang, L. (2021). Physics-informed machine learning. Nature Reviews Physics, 3(6), 422–440.
Section
Technical Papers