Physics-informed cross-attention operator network with hard-constrained Fourier features for heat map prediction of large-scale battery packs

Accurate prediction of temperature distributions is essential for safe and efficient battery pack design and management. In indirect liquid cooling configurations, battery cell layouts strongly influence internal heat transfer, which complicates layout-aware heat map prediction. This paper proposes a physics-informed cross-attention operator network (PI-CAON), a mesh-free neural operator for steady-state heat map prediction in large-scale battery packs. The model integrates Fourier feature encoding, a hard-constrained Fourier embedding to enforce Neumann boundary conditions, and a cross-attention-based feature fusion mechanism to capture layout-dependent thermal interactions. By embedding governing heat transfer physics into the training loss, PI-CAON enables label-free learning while maintaining physical consistency. Numerical experiments on a 20-cell indirect liquid cooling battery pack demonstrate maximum temperature error below 0.03 °C and consistently outperform grid-based and data-driven baselines.