Temperature Prediction Method of Lithium-Ion Battery Pack Based on Attention Mechanism Optimized Convolutional Neural Network and Gated Recurrent Unit

Lithium-ion battery pack play a critical role in electric vehicles and renewable energy storage systems, with temperature regulation being crucial for their performance and lifespan. Elevated temperatures can expedite battery depletion, curtail cycle life, and heighten safety hazards. Consequently, precise temperature prediction and stringent control are essential. In this paper, we propose a temperature prediction method for lithium-ion battery pack based on convolutional neural network (CNN) and gated recurrent unit (GRU), with an integrated attention mechanism (AM) to augment the model's predictive prowess. Firstly, COMSOL Multiphysics software is used to simulate the operation process of lithium-ion battery pack, and a large amount of parameter data is collected as the training and testing set of the model. Subsequently, the CNN is combined with GRU, and the attention mechanism is introduced to enhance the prediction performance of the model. Empirical results demonstrate that the optimized CNN-GRU model exhibits high accuracy and reliability in temperature prediction of lithium-ion battery pack, with R2 values of more than 0.99, RMSE values of less than 0.08, and MSE values of less than 0.007 for both the training and test sets.