Predictive Models for Voltage Degradation in PEM Fuel Cells Using Gradient Boosting and LSTM Networks

The reliability and operational effectiveness of Pro-ton Exchange Membrane Fuel Cells (PEMFCs) are significantly affected by voltage degradation. To predict voltage degradation in PEMFCs, this study focuses on data-driven approaches that employ machine learning (ML) and deep learning (DL). The research demonstrates how Long-Short Term Memory (LSTM) networks outperform other approaches by training models such as gradient boost regressors, neural networks, and LSTM networks through large PEMFC voltage datasets. Gradient boost regressors and other ML models are not particularly efficient at capturing long-term dependencies, and while LSTM networks are more efficient at controlling temporal patterns, they still require a lot of computational resources and precise hyperparameter tuning. Nevertheless, despite these challenges, integrating these models provides more scalable and accurate prognostics than tra-ditional physics-based techniques, making them highly beneficial for enhancing PEMFC operational efficiency while promoting the greater installation of renewable energy resources.