TY  - CONF
AU  - Zimmer, Marcel
AU  - De Din, Edoardo
AU  - Carta, Daniele
AU  - Benigni, Andrea
TI  - Power Electronics Parameter Estimation by Physics-Informed Gaussian Processes
PB  - IEEE
M1  - FZJ-2025-05277
SP  - 1-6
PY  - 2025
AB  - Non-invasive parameter estimation of power electronics enables condition and health monitoring, enhances control strategies, and thus, guarantees reliable and stable operation of power converters. In this work, we propose the application of Physics-Informed Gaussian Processes (PIGP) for parameter estimation of power electronic converters. We provide a detailed model-building scheme allowing non-invasive characterisation of power converters. In particular, we show that the proposed approach can be applied independently of the particular noise level without the need for data pre-processing. Tested with a DC-DC buck converter in a simulation scenario, we show that the proposed approach is immune to measurement noise and can be executed with low sampling rates. This allows fast, possibly online, execution and thus tracking of converter parameter changes due to thermal effects or aging during operation.
T2  - IECON 2025 – 51st Annual Conference of the IEEE Industrial Electronics Society
CY  - 14 Oct 2025 - 17 Oct 2025, Madrid (Spain)
Y2  - 14 Oct 2025 - 17 Oct 2025
M2  - Madrid, Spain
LB  - PUB:(DE-HGF)8
DO  - DOI:10.1109/IECON58223.2025.11221137
UR  - https://juser.fz-juelich.de/record/1049195
ER  -