Neural Network Enhanced Physicsbased Surrogate Model Framework for Multiphysics Battery Simulations

Abstract

Physics-based battery models are important for understanding and predicting the behavior of batteries during operation. To fully describe their performance under different conditions high-fidelity multiphysics models are used. While these offer high accuracy, their computational cost is high. Simplified models, such as equivalent circuit models, provide a faster alternative. However, these models are less flexible and need to be recalibrated for different circumstances, often involving extensive testing. This thesis presents a framework for using neural networks to enhance simplified physicsbased models based on high-fidelity simulation data. Battery voltage and swelling force are simulated with a high-fidelity electrochemical-mechanical battery model for different applied currents and external pre-compression loads, while accounting for electrochemically induced swelling. The voltage response is then modeled with a first-order non-linear equivalent circuit model, and the mechanical response with a non-linear spring along with a swelling rate. Based on the high-fidelity data, neural networks are used to approximate the circuit elements, spring stiffness and swelling rate. Finally, symbolic regression is used to discover closed-form expressions for each neural network, revealing insights in dependencies. The result suggests that highly non-linear circuit elements are required to capture the dynamics of battery voltage. Compared to a fully data-driven baseline, embedding neural networks within an equivalent circuit model preserved physical relations when extrapolating beyond the training domain. This work is limited to fully reversible swelling, and the voltage data considers only discharges. Although the thesis project demonstrates this framework on mechanically coupled battery simulations, we believe that the methodology and the overall workflow can be extended to include other processes, such as thermal effects or degradations in future work.