Early safety warning method for lithium-ion batteries under mechanical abuse conditions based on online electrochemical parameter identification

Lithium-ion batteries are critical energy storage components yet face significant safety challenges under mechanical abuse conditions. This study proposes an early safety warning method that utilizes online electrochemical parameter identification to detect internal damage triggered by mechanical stress. By integrating an improved multi-particle model (MP+) with a data-mechanism driven identification approach, we establish a cross-scale correlation framework connecting microscopic aging mechanisms to macroscopic performance degradation. Our method monitors multidimensional indicators, including abrupt changes in polarization coefficients and diffusion parameters, fused with capacity degradation gradients to establish dynamic warning thresholds. This strategy enables risk level assessment 100 to 200 cycles prior to critical failure. Validation through multi-scale analysis, including macroscopic observation, X-ray computed tomography (XCT), and scanning electron microscopy (SEM), confirms the method's effectiveness in predicting safety hazards and enhancing battery reliability.