Physics-Informed Hybrid Neural Architecture for Coupled Degradation Modeling and Remaining Useful Life Prediction of LiFePO4 Batteries

E. Lixin; Jun Wang; Yue Sun; Weixiang Shen; Rui Xiong

doi:10.1149/1945-7111/addd4e

Physics-Informed Hybrid Neural Architecture for Coupled Degradation Modeling and Remaining Useful Life Prediction of LiFePO₄ Batteries

E. Lixin
, Jun Wang
, Yue Sun
, Weixiang Shen
, Rui Xiong^*

^*Corresponding author for this work

Beijing Institute of Technology
Swinburne University of Technology

Research output: Contribution to journal › Article › peer-review

7 Citations (Scopus)

Abstract

Highlights A multi-factor aging physical model incorporating knee point information is proposed. Achieving end-to-end accurate prediction of knee points and RUL based on CNN. Integration of physical and data-driven models for accurate degradation prediction. Deep coupling of remaining useful life prediction with degradation trajectory prediction. Limited training data shows strong early prediction capability.

Original language	English
Article number	060505
Journal	Journal of the Electrochemical Society
Volume	172
Issue number	6
DOIs	https://doi.org/10.1149/1945-7111/addd4e
Publication status	Published - 1 Jun 2025
Externally published	Yes

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 7 Affordable and Clean Energy

Keywords

degradation trajectory
knee point
physics-informed neural network
remaining useful life

Access to Document

10.1149/1945-7111/addd4e

Cite this

@article{ff8d43d523aa4fbbb31383046a45d265,

title = "Physics-Informed Hybrid Neural Architecture for Coupled Degradation Modeling and Remaining Useful Life Prediction of LiFePO4 Batteries",

abstract = "Highlights A multi-factor aging physical model incorporating knee point information is proposed. Achieving end-to-end accurate prediction of knee points and RUL based on CNN. Integration of physical and data-driven models for accurate degradation prediction. Deep coupling of remaining useful life prediction with degradation trajectory prediction. Limited training data shows strong early prediction capability.",

keywords = "degradation trajectory, knee point, physics-informed neural network, remaining useful life",

author = "E. Lixin and Jun Wang and Yue Sun and Weixiang Shen and Rui Xiong",

note = "Publisher Copyright: {\textcopyright} 2025 The Electrochemical Society (“ECS”). Published on behalf of ECS by IOP Publishing Limited. All rights, including for text and data mining, AI training, and similar technologies, are reserved.",

year = "2025",

month = jun,

day = "1",

doi = "10.1149/1945-7111/addd4e",

language = "English",

volume = "172",

journal = "Journal of the Electrochemical Society",

issn = "0013-4651",

publisher = "Institute of Physics",

number = "6",

}

TY - JOUR

T1 - Physics-Informed Hybrid Neural Architecture for Coupled Degradation Modeling and Remaining Useful Life Prediction of LiFePO4 Batteries

AU - Lixin, E.

AU - Wang, Jun

AU - Sun, Yue

AU - Shen, Weixiang

AU - Xiong, Rui

N1 - Publisher Copyright: © 2025 The Electrochemical Society (“ECS”). Published on behalf of ECS by IOP Publishing Limited. All rights, including for text and data mining, AI training, and similar technologies, are reserved.

PY - 2025/6/1

Y1 - 2025/6/1

N2 - Highlights A multi-factor aging physical model incorporating knee point information is proposed. Achieving end-to-end accurate prediction of knee points and RUL based on CNN. Integration of physical and data-driven models for accurate degradation prediction. Deep coupling of remaining useful life prediction with degradation trajectory prediction. Limited training data shows strong early prediction capability.

AB - Highlights A multi-factor aging physical model incorporating knee point information is proposed. Achieving end-to-end accurate prediction of knee points and RUL based on CNN. Integration of physical and data-driven models for accurate degradation prediction. Deep coupling of remaining useful life prediction with degradation trajectory prediction. Limited training data shows strong early prediction capability.

KW - degradation trajectory

KW - knee point

KW - physics-informed neural network

KW - remaining useful life

UR - https://www.scopus.com/pages/publications/105007927440

U2 - 10.1149/1945-7111/addd4e

DO - 10.1149/1945-7111/addd4e

M3 - Article

AN - SCOPUS:105007927440

SN - 0013-4651

VL - 172

JO - Journal of the Electrochemical Society

JF - Journal of the Electrochemical Society

IS - 6

M1 - 060505

ER -