Enhanced transformer encoder long short-term memory hybrid neural network for multiple temperature state of charge estimation of lithium-ion batteries.

Zou, Yuanru; Wang, Shunli; Cao, Wen; Hai, Nan; Fernandez, Carlos

Abstract

Accurate state of charge (SOC) estimation for lithium-ion batteries remains a critical challenge in battery management systems. Existing methods based on machine learning may cause data leakage and inaccuracy during neural network training. Here, this paper proposed a novel sequence-to-point historical data augmentation method that enhances SOC estimation by leveraging only past data, eliminating interference from data leakage. This approach integrates a neural network architecture, combining Transformer Encoder and Long Short-Term Memory (LSTM) models, to predict SOC with high precision. An ensemble learning strategy is employed, where multiple cross-validated models act as weak learners, and their weighted predictions improve generalization and accuracy. Experimental validation across varying temperatures and working conditions demonstrates a mean absolute error (MAE) of less than 0.75%, a root mean square error (RMSE) below 0.93%, and an R-Square (R2) exceeding 0.9988, significantly outperforming baseline models. Our method offers a robust and reliable solution for SOC estimation, with applications in electric vehicles, portable electronics, and energy storage systems.

Citation

ZOU, Y., WANG, S., CAO, W., HAI, N. and FERNANDEZ, C. 2025. Enhanced transformer encoder long short-term memory hybrid neural network for multiple temperature state of charge estimation of lithium-ion batteries. Journal of power sources [online], 632, article number 236411. Available from: https://doi.org/10.1016/j.jpowsour.2025.236411

• Journal Article Type: Article
• Acceptance Date: Jan 29, 2025
• Online Publication Date: Feb 3, 2025
• Publication Date: Mar 15, 2025
• Deposit Date: Feb 7, 2025
• Publicly Available Date: Feb 4, 2026
• Journal: Journal of power sources
• Print ISSN: 0378-7753
• Electronic ISSN: 1873-2755
• Publisher: Elsevier
• Peer Reviewed: Peer Reviewed
• Volume: 632
• Article Number: 236411
• DOI: https://doi.org/10.1016/j.jpowsour.2025.236411
• Keywords: Lithium-ion battery; State of charge estimation; Battery management system; Historical information data enhancement; Transformer encoder LSTM; Ensemble learning
• Public URL: https://rgu-repository.worktribe.com/output/2696105