Coupled physics-informed neural networks for thermal assessment of a phase change material-based battery pack under high discharge rates

Key Points

• This research aims to enhance thermal analysis of phase change material-based battery packs using a novel neural network approach.
• Developed a coupled PINN ensemble for analyzing battery thermal behavior.
• Embedded an enthalpy-based partial differential equation to model latent heat.
• Utilized Fourier features to address sharp thermal gradients at cell surfaces.
• Achieved less than 0.5 °C root mean square error (RMSE) using sparse experimental data.
• Generated continuous 4D/3D/2D thermal and phase change visualizations.