Early Warning of Cylindrical Lithium-Ion Battery Overcharge Failure Characteristic Based on Informer Neural Network

Lithium-ion batteries often experience overcharge due to battery management system failure or battery pack inconsistencies, which lead to serious safety accidents. Therefore, an effective overcharge warning method is of great importance to guarantee the safe performance of batteries. This paper proposes an overcharge warning method for lithium-ion batteries based on the Informer neural network, which realized the warning by internal resistance prediction during the battery charge process in real-time. In this paper, overcharge experiments under different temperatures and charge rates were conducted, and the experimental data were organized for parameter training of the Informer model. The prediction performance of the model under different prediction lengths was verified. The results show that this method can effectively provide warning information for battery overcharge.