KAN-Therm: A Lightweight Battery Thermal Model Using Kolmogorov-Arnold Network

A battery management system (BMS) relies on real-time estimation of battery temperature distribution in battery cells to ensure safe and optimal operation of Lithium-ion batteries. However, physical BMS often suffers from memory and computational resource limitations required by high-fidelity models. This study proposes a Kolmogorov-Arnold network (KAN) based thermal model, KAN-therm, which aims to estimate the core temperature of a cylindrical battery more efficiently. Unlike traditional neural network architectures, KAN employs learnable nonlinear activation functions that effectively capture system complexity with relatively simple models. We evaluated KAN-therm's performance by comparing its memory overhead and estimation time against state-of-the-art neural network and tree-based models. The results showcase the applicability and potential scalability of KAN-therm within a physical BMS framework, indicating significant improvements in computational efficiency while maintaining accurate temperature estimates. By addressing the challenges associated with existing models, KAN-therm offers a promising solution for enhancing the thermal management capabilities critical for the safe operation of Lithium-ion batteries.