Early perception of Lithium-ion battery degradation trajectory with graphical features and deep learning

Haichuan Zhao, Jinhao Meng and Qiao Peng

Applied Energy, 2025, vol. 381, issue C, No S0306261924025984

Abstract: Capturing the degradation path of lithium-ion battery (LIB) at the early stage is critical to managing the whole lifespan of the battery energy storage systems (BESS), while recent research mainly focuses on the short-term battery health diagnosis such as state of health (SOH). This work investigates an innovative concept to perceive the degradation trajectory of the LIBs with few initial cycles, where sufficient tuning space can be left for the sophisticated operation and maintenance of BESS. A novel deep learning framework is proposed to obtain capacity degradation trajectory using graphical features constructed with the early battery usage data. To capture richer capacity decay features, the framework enhances the voltage-capacity data by generating incremental capacity (IC) and capacity difference curves, which are then spliced to construct graphical features. A multi-channel dependent neural network (MCDNet) is developed to extract degradation information from graphical features and predict key trajectory knots using a large-size convolutional kernel and STar Aggregate-Redistribute (STAR) feature fusion method to ensure the advantage of channel independence while facilitating the interaction of channel information. The capacity degradation trajectory will be reconstructed with the key knots using the piecewise cubic Hermite interpolating polynomial (PCHIP). The proposed model is validated against advanced image classification algorithms and its performance is tested under different battery lifetime scenarios, limited cycling data, and different voltage segments. In most cases, the proposed method obtains the capacity degradation trajectories with the mean absolute error of less than 60 cycles.

Keywords: Lithium-ion battery; Degradation trajectory; Feature extraction; Multi-channel dependent neural network; Deep learning (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
http://www.sciencedirect.com/science/article/pii/S0306261924025984
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:381:y:2025:i:c:s0306261924025984

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.2024.125214

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 ().