A collaborative interaction gate-based deep learning model with optimal bandwidth adjustment strategies for lithium-ion battery capacity point-interval forecasting

Zhi-Feng Liu, Ya-He Huang, Shu-Rui Zhang, Xing-Fu Luo, Xiao-Rui Chen, Jun-Jie Lin, Yu Tang, Liang Guo and Ji-Xiang Li

Applied Energy, 2025, vol. 377, issue PD, No S030626192402124X

Abstract: Lithium-ion batteries (LIBs) are widely employed in electric vehicles due to their environmental friendliness and extended lifespan. However, accurately forecasting the remaining useful life of LIBs presents challenges owing to intricate internal electrochemical reactions and external environmental uncertainties. Therefore, this study proposes a new LSTM-Informer deep learning model based on cooperative interaction gates for lithium-ion battery capacity point-interval prediction. Specifically, building upon the three gating mechanisms of LSTM, a novel cooperative interaction gate is proposed to thoroughly explore the direct correlation between the degraded sequences; and a long and short-term memory weight control strategy based on the capacity regeneration ratio is introduced to dynamically adjust the influence of the four gating mechanisms based on the fluctuation degree of the data, so as to accurately capture the capacity regeneration phenomenon of the degraded sequences, and to increase the prediction point prediction precision. The efficient ProbSparse is introduced to improve the prediction performance at the later stage of sequence data. Based on the interval prediction Kernel Density Estimation model, a dynamic optimal bandwidth optimization strategy is constructed to adaptively adjust the interval width according to the data fluctuation characteristics, which improves the uncertainty expression of the model. The results demonstrate that the proposed model achieves impressive prediction performance for different types of batteries, with the point prediction evaluation index MAPE controlled below 1.5 % and the interval prediction evaluation index CWC controlled below 0.7.

Keywords: Energy application; Lithium-ion battery; Remaining useful life; Deep learning; Point-interval forecasting (search for similar items in EconPapers)
Date: 2025
References: View references in EconPapers View complete reference list from CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S030626192402124X
Full text for ScienceDirect subscribers only

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:eee:appene:v:377:y:2025:i:pd:s030626192402124x

Ordering information: This journal article can be ordered from
http://www.elsevier.com/wps/find/journaldescription.cws_home/405891/bibliographic
http://www.elsevier. ... 405891/bibliographic

DOI: 10.1016/j.apenergy.2024.124741

Access Statistics for this article

Applied Energy is currently edited by J. Yan

More articles in Applied Energy from Elsevier
Bibliographic data for series maintained by Catherine Liu ().