A deep learning approach incorporating attention mechanism and transfer learning for lithium-ion battery lifespan prediction

Abstract

Accurate lifespan prediction for lithium-ion batteries is crucial for battery management systems to ensure reliable operation and timely maintenance. However, current lifespan prediction methods often fail to provide accurate predictions during the early stages of battery capacity decline. In this study, a deep learning approach is proposed which incorporates attention mechanism and transfer learning to address the limitations of limited battery early history cycle data. A convolutional neural network is designed to automatically extract features from the input data, which are generated from the capacity change curve of the battery during the early discharge stage. The attention layers are integrated into the convolutional layers, to enhance the extraction of pertinent features by emphasizing critical information through weight recalibration. Moreover, a transfer learning strategy is proposed to selectively transfers parameters from a pre-trained model to a target model, which is then partially retrained with limited target data. The performance of the proposed model is validated using two public datasets. Experimental results demonstrate that the model outperforms existing methods in terms of prediction accuracy and generalization.