A novel physics-embedded hybrid learning framework for reliable state-of-charge estimation in Lithium-Ion batteries

Accurate state-of-charge (SoC) estimation is crucial for efficient and safe operation of Lithium-Ion batteries in Electric Vehicles (EVs). However, SoC estimation is challenging due to nonlinear battery dynamics and sensitivity to environmental factors. Current hybrid models often lack explicit integration of physical constraints, which leads to error accumulation and performance instability under dynamic operating conditions. To address this, this article introduces Hybrid Coulomb-Gated Network (HCG-Net), a novel hybrid SoC estimation framework that combines data-driven learning with physics-based constraints to improve SoC estimation accuracy and robustness. The framework mediates measurement error and physical fidelity through a tunable parameter, \(\lambda\) . HCG-Net is evaluated on an extensive dataset spanning various temperatures (25°C, 10°C, 0°C, and -10°C) and benchmarked against leading SoC estimation techniques. Numerical results show that HCG-Net achieves a 15% reduction in RMSE and a 36% reduction in MAE over the best-performing baselines. This stable gain in accuracy and robustness across thermal variations demonstrates HCG-Net’s suitability for real-time SoC estimation. This study contributes to the United Nations Sustainable Development Goals (SDGs), particularly SDG 7 (Affordable and Clean Energy) and SDG 13 (Climate Action).