TY  - CONF
AU  - Abdollahi, Farideh
AU  - Malek, Kourosh
AU  - Kadyk, Thomas
AU  - Eikerling, Michael
TI  - Autonomous Data Analytics for Enhanced Performance and Lifetime Prediction in PEM Fuel Cells and Water Electrolyzers
M1  - FZJ-2025-01426
PY  - 2024
AB  - Longevity is a crucial aspect in evaluating the economic viability of polymer electrolyte fuel cells (PEFCs) in a sustainable energy economy. Making reliable predictions on the performance and lifetime of PEFCs remains challenging due to the complex interplay of processes involved in their operation, including those that drive degradation. The prospects of forecasting PEFC performance with physical models hinges on their completeness in terms of processes accounted for and data available for parameterization. Data-driven models, on the other hand, typically lack the mechanical insight necessary for a deep understanding of degradation causes. We, therefore, pursue the development of a hybrid modeling approach that combines the capabilities of physical models with the agility of data-driven techniques. The aim of this approach is to evaluate the effectiveness of physical models in forecasting performance and to assess their ability for making reliable predictions about performance degradation and lifetime. The combined approach is anticipated to surpass separate physical and data-based models in terms of accuracy, robustness, and interpretability, providing a reliable foundation for identifying maintenance needs and extending the lifespan of PEFCs.
T2  - Helmholtz AI Conference 2024
CY  - 12 Jun 2024 - 14 Jun 2024, Dusseldorf (Germany)
Y2  - 12 Jun 2024 - 14 Jun 2024
M2  - Dusseldorf, Germany
LB  - PUB:(DE-HGF)24
UR  - https://juser.fz-juelich.de/record/1038430
ER  -