Abstract：In low-temperature environments，the electrochemical reaction kinetics of lithium-ion batteries

become hindered，leading to accelerated capacity decay and increased internal resistance，which severely impacts

their lifespan and safety. To achieve non-destructive estimation of internal aging，a battery internal aging state

estimation method based on a spatio-temporal convolutional neural network（ST-CNN）was proposed. Firstly，insitu analysis techniques examined the battery's incremental capacity（IC）and differential voltage（DV）curves to

calculate quantitative parameters for three aging modes：loss of active material（LAM），loss of lithium inventory

（LLI），and loss of conductivity（LC）. Secondly，a quantitative characterization system for internal aging was

established by extracting features from material morphology changes and electrochemical impedance spectroscopy

（EIS）. Thirdly，a temporal-spatial feature modeling framework based on ST-CNN was designed to accurately map

complex internal degradation mechanisms. Finally，the proposed model was validated using experimental data from

low-temperature conditions. Experimental results demonstrate that this method achieves high-precision aging state

estimation across multiple low-temperature conditions：MAE≤1.3%，RMSE≤6.1%，and R2≥0.99. These findings offer novel insights for battery management system lifespan prediction and safety management.

      Key words：lithium-ion battery；low-temperature conditions；degradation mechanism；internal aging estimation；spatio-temporal convolutional neural network（ST-CNN）