Distributed force field heterogeneity for early sensing and hierarchical warning of battery thermal runaway

The "averaging effect" of signals in battery modules remains a significant bottleneck for the early detection of thermal runaway (TR). To address this, this study proposes a novel Micro-Macro dual-layer cascaded warning strategy based on in-situ distributed pressure sensing. Using 52Ah LiFePO₄/graphite (LFP) modules (1P4S) as a testbed, the study analyzes the multi-physical field coupling mechanism during TR induced by overcharge and overheating. Experimental findings indicate that distributed force features possess superior sensitivity to local deformations compared to conventional thermal, electrical, and lumped force signals. By constructing a comprehensive evaluation system and utilizing an automated PCA-based scoring algorithm, three robust precursor features—Correlation Distance, PCA Reconstruction Error, and FFT Energy—are successfully extracted. Validation experiments demonstrate that this multi-level strategy significantly extends the warning window, achieving maximum lead times of 4 min against total force signals and 10 min against surface temperature monitoring. This work provides a theoretical and methodological basis for mechanical-signal-based active safety monitoring and offers a promising approach to mitigating signal lag and masking issues in battery module protection.