Abstract: With the growing energy demand, lithium-ion batteries, as the core of energy storage technology, are widely used in new energy fields. However, the problems of fire and explosion that may be caused by batteries under overcharging conditions have attracted great attention to the safety of batteries. Accurate assessment of battery temperature is important to prevent overheating and avoid equipment failure. In this study, an Adam optimization-based long-short time memory network (LSTM) temperature assessment method for commercial lithium-ion batteries with overcharging faults is proposed. The method utilized the advantages of the LSTM model in processing time series data and capturing long and short-term dependencies and constructed a high-precision temperature assessment model by combining the current and voltage characteristics of the battery overcharge fault process. The experimental results indicate that, for the evaluation of battery temperature across five instances, the lowest root mean square error is 0.619 ℃, the lowest mean absolute error is 0.550 ℃, and the coefficient of determination can reach as high as 0.998. This method provides an important data reference for the monitoring and management of overcharge faults of lithium-ion batteries.