2026-10-12T20:05:15Z https://api.figshare.com/v2/oai

oai:figshare.com:article/32641590 2026-10-01T16:22:01Z portal_1172 item_type_8 month_year_10_2026

Physics-informed neural networks for battery response prediction Andrew Forde (19761432) PUREID: 651030469 Battery thermal management system battery electric vehicles machine learning PINN thermal modelling Electric vehicle Battery performance is typically captured by detailed electrochemical models, requiring extensive parameterization of material properties and cell internal geometry, or purely data-driven methods needing large amounts of experimental data. This work introduces methods of physics-informed neural networks (PINNs) lying on a spectrum between these two approaches, enabling prediction of lithium-ion cell response with low computational cost while utilising available data and known behaviours. For electrical behaviour, the equivalent-circuit model (ECM) was replaced with one of two approaches: a Gradient-Informed Neural Network (GINN) incorporating gradient constraints into its loss function, and a Model-Informed Neural Network (MINN) that embeds an ECM within the training loop. Both approaches reduced computation time by up to two magnitudes compared to the ECM (2.34% of runtime). Memory requirements were also lower when simulating more than five cells simultaneously. For thermal predictions, a cloud-point PINN was developed which learns the solution of the heat equation for given boundary conditions. By enforcing the governing PDE and boundary conditions, this model learns to predict cell temperature at any resolution for static and dynamic load profiles. This achieved accurate thermal representation when assuming uniform cell heating but failed to account for heat passing between components for non-uniform heating. A mesh-based timestepping PINN was developed and compared against the Crank-Nicolson method. The resultant model exhibited a runtime scaling linearly with mesh density where conventional solvers scale exponentially. The PINN achieved a sufficient temperature estimation (error <1 °C) while reducing runtime by an order of magnitude for a 50x50 mesh. This work demonstrates that PINNs can provide rapid, accurate battery response predictions across electrical and thermal domains, while tuning models between being physics- and data-driven. This offers potential development of real-time management and efficient design of thermal solutions, leading to more efficient use of batteries in electric vehicles and similar applications. 2026-10-01T16:22:01Z Text Thesis 10.17034/32641590.v1 https://figshare.com/articles/thesis/Physics-informed_neural_networks_for_battery_response_prediction/32641590 All Rights Reserved