Attention Mechanism Empowered Dual Active Bridge Converter Performance Modeling with Enhanced Interpretability and Lighter Data

This paper introduces "DDM-ETS," a novel data-driven modeling approach for enhancing accuracy and interpretability in power converter modeling, specifically tailored for dual active bridge (DAB) converters. This method significantly improves the precision and accuracy of the model by combining data interpretability analysis with TabNet. Furthermore, it improves the transparency and interpretability of the model by analyzing the decision-making process through the integration of Shapley Additive Explanations and mask. The practicality and effectiveness of DDM-ETS are demonstrated through a 1-kW hardware experiment.