Study on dynamic assessment of fast charging safety of electric vehicles based on thermal-electrical-stress coupling modeling and risk warning mechanism

Electric vehicles are prone to induce potential safety hazards such as thermal runaway, lifetime degradation and localized stress concentration during high rate fast charging. This paper proposes a safety assessment method based on thermal-electrical-stress three-field coupling modeling and risk grading dynamic control mechanism. Firstly, a high-precision multi-physics field simulation model is constructed, which includes the thermal conductivity change of the electric core, the expansion of the pole piece and the generation probability of lithium dendrites, and is calibrated based on the experimentally measured parameters. In terms of control strategy, a variable-multiplier constant voltage-constant current (CV-CC) regulation method based on segmented curve fitting is introduced to dynamically adjust the charging power in combination with the real-time temperature and internal resistance trends to avoid exceeding the thermal-stress coupling threshold. Meanwhile, a risk prediction module based on LSTM-Attention network is designed to recognize the high-risk operating condition interval in advance and trigger the cooling subsystem pre-response. The proposed system is validated on an LG 21700 cylindrical cell platform and its fast-charging experimental chamber, and the results show that the thermal peak is reduced by 13.2% and the average pole stress is reduced by 15.7% under the charging multiplication of 4C, and the risk of thermal runaway is successfully predicted in advance up to 6.4 minutes in advance. This study provides a closed-loop safety assurance path integrating modeling-identification-control for fast-charging battery systems.