Internal Short Circuit Diagnosis of Lithium-Ion Battery Based on Mechanism Model and Deep Learning

Abstract

Most safety problems of the lithium-ion battery are attributed to internal short circuits in the battery. There are many factors leading to the internal short circuit of Li-ion battery, and this paper makes the experimental repeatability and controllability of the internal short circuit of the battery better by establishing the mechanism model simulation. In the process of internal short-circuit heat generation in the battery, the battery thermal effect affects the electrochemical reaction of the battery, generating a larger short-circuit current, which releases more short-circuit heat. Therefore, a ternary battery electrochemical-thermal-internal short-circuit coupling mechanism model is established based on the characteristic connection between the three factors. Finally, a lithium-ion battery internal short-circuit diagnosis model is established by combining deep learning algorithms. Six evaluation parameters, including model training time, convergence speed, accuracy, precision, recall and F-value, are also compared, and the effectiveness of convolutional neural network (CNN) and long short-term memory neural network (LSTM) for classification and diagnosis of the severity of internal short circuit in batteries is compared: the CNN model gets better results for classification and diagnosis of internal short circuit in batteries. And lays the foundation for online diagnosis of battery internal short circuit.