Model-integrated Neural Networks for Control-oriented Battery Modeling

The concept of integrating physics-based and data-driven approaches has become popular for modeling energy systems. However, existing literature mainly focuses on data-driven surrogates generated to replace physics-based models. These models often trade accuracy for speed, but lack generalizability, adaptability, and interpretability inherent in physics-based models, which are qualities crucial for optimisation and control purposes. We propose a novel machine learning architecture, termed model-integrated neural networks, capable of learning the physics-based dynamics of general autonomous or non-autonomous systems consisting of partial differential algebraic equations. This architecture is then applied to the modelling of different batteries and electrode materials.