A Physics-Inspired and Data-Driven Approach for Temperature-Based Condition Monitoring

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

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

Published Jun 27, 2024
Giacomo Garegnani Kai Hencken Frank Kassubek

Abstract

System overheating is a common problem in electric equipment, as degradation of contacts lead to an increase in Ohmic resistance and increased thermal losses. Temperature measurements are widely employed to monitor a device's health status, to estimate its remaining useful life, and to inform maintenance strategies. An issue that is special to electrical distribution networks is the varying heating power, which is in turn due to changes in the current. This results in varying temperatures, which in addition can often be delayed compared to the currents. Simple threshold-based diagnostics approaches are therefore not reliable for detecting faults in contacts. It is common to analyze the thermal behavior of electric devices using thermal networks, for both design and diagnostic purposes. Unfortunately, identifying the parameters of thermal networks from measured temperature data is a challenging problem, mainly due to identifiability issues and to numerical instabilities in parameter estimation. We propose an alternative data-driven strategy to compute the state-of-health of electrical devices which does not resort to thermal networks. Our approach consists in informing physics-based reduced models of the thermal response with sensor data. We show that our method is linked to the thermal network approach but avoids the full identification of the system, leading to better stability as well as less computational effort in the determination of its parameters. Rigorous testing with synthetic and experimental data confirms the efficacy of our methodology.

How to Cite

Garegnani, G., Hencken, K., & Kassubek, F. (2024). A Physics-Inspired and Data-Driven Approach for Temperature-Based Condition Monitoring. PHM Society European Conference, 8(1), 10. https://doi.org/10.36001/phme.2024.v8i1.3977
Abstract 172 | PDF Downloads 169

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

Keywords

Thermal monitoring, reduced-order modeling, electric equipment

References
Boodi, A., Beddiar, K., Amirat, Y., & Benbouzid, M. (2022). Building thermal-network models: a comparative analysis, recommendations, and perspectives. Energies, 15(4), 1328. Brastein, O., Ghaderi, A., Pfeiffer, C., & Skeie, N.-O. (2020). Analysing uncertainty in parameter estimation and prediction for grey-box building thermal behaviour models. Energy and Buildings, 224, 110236. Brastein, O. M., Lie, B., Sharma, R., & Skeie, N.-O. (2019). Parameter estimation for externally simulated thermal network models. Energy and Buildings, 191, 200–210. Hoffmann, M. W., Wildermuth, S., Gitzel, R., Boyaci, A., Gebhardt, J., Kaul, H., . . . Tornede, T. (2020). Integration of novel sensors and machine learning for predictive maintenance in medium voltage switchgear to enable the energy and mobility revolutions. Sensors, 20(7), 2099. Raue, A., Karlsson, J., Saccomani, M. P., Jirstrand, M., & Timmer, J. (2014). Comparison of approaches for parameter identifiability analysis of biological systems. Bioinformatics, 30(10), 1440–1448. Stosur, M., Szewczyk, M., Sowa, K., Dawidowski, P., & Balcerek, P. (2016). Thermal behaviour analyses of gasinsulated switchgear compartment using thermal network method. IET Generation, Transmission & Distribution, 10(12), 2833–2841.
Section
Technical Papers