Capacity prediction of lithium-ion batteries based on a VAE-STAGCNN model

LIAO Weiqiang; ZHANG Mengyuan; WANG Yilong; ZHANG Changkun; WU Yuegao

doi:10.19783/j.cnki.pspc.260016

English Version Journal Login

您当前的位置：

首页 >

文章列表页 >

Capacity prediction of lithium-ion batteries based on a VAE-STAGCNN model

更新时间：2026-07-27

- Capacity prediction of lithium-ion batteries based on a VAE-STAGCNN model
- Power System Protection and Control Vol. 54, Issue 13, (2026)
- 作者机构：
- 作者简介：
- 基金信息：
- DOI：10.19783/j.cnki.pspc.260016
  CLC：
- Published：2026
- 稿件说明：
移动端阅览
LIAO Weiqiang, ZHANG Mengyuan, WANG Yilong, et al. Capacity prediction of lithium-ion batteries based on a VAE-STAGCNN model[J]. Power System Protection and Control, 2026, 54(13).
DOI：

LIAO Weiqiang, ZHANG Mengyuan, WANG Yilong, et al. Capacity prediction of lithium-ion batteries based on a VAE-STAGCNN model[J]. Power System Protection and Control, 2026, 54(13). DOI： 10.19783/j.cnki.pspc.260016.

摘要

锂离子电池容量预测面临工况复杂、测试成本高及衰退过程强非线性等挑战，传统数据驱动方法在数据匮乏场景下准确性与泛化性受限。为此，提出融合变分自编码器 (variational autoencoder

VAE) 与时空注意力图卷积神经网络 (spatio-temporal attention graph convolutional neural network

STAGCNN) 的预测方法。首先，利用 VAE 学习得到电池老化潜在分布，通过隐空间重采样生成多样化样本，增强少样本条件下的数据覆盖与模型鲁棒性。其次，将电池多维度特征构建为图节点，基于特征相关性动态生成邻接矩阵，并引入时间注意力机制，实现空间关联与时间依赖的协同建模。最后，利用实际电池数据集进行实验验证。结果表明，所提模型预测精度显著提升，并具有一定的跨电池泛化潜力。

Abstract

Lithium-ion battery capacity prediction faces challenges such as complex operating conditions

high testing costs

and highly nonlinear degradation processes. Traditional data-driven methods often suffer from limited accuracy and generalization capability in scenarios with scarce data. To address these issues

this paper proposes a prediction method that integrates variational autoencoder (VAE) with spatiotemporal attention graph convolutional neural network (STAGCNN). First

the latent distribution of battery aging is learned using a VAE

and diversified samples are then generated through latent space resampling

thereby enhancing data coverage and model robustness under small-sample conditions. Second

multidimensional battery features are represented as graph nodes

and an adjacency matrix is dynamically generated based on feature correlations. A temporal attention mechanism is further introduced to achieve collaborative modeling of spatial associations and temporal dependencies. Finally

experiments are conducted using real-world battery datasets. The results show that the proposed model significantly improves the prediction accuracy while exhibiting promising cross-battery generalization capability.

关键词

Keywords

references

LIU Yongtao, LIU Chuanpan, LIU Yongjie, et al. Review on degradation mechanism and health state estimation methods of lithium-ion batteries[J]. Journal of Traffic and Transportation Engineering, 2023, 10(4): 578-610.

LIU Yuefeng, HE Yingjie, BIAN Haodong, et al. A review of lithium-ion battery state of charge estimation based on deep learning: directions for improvement and future trends[J]. Journal of Energy Storage, 2022, 52.

WANG Xiaohua, LI Mengting, HU Min, et al. SOH prediction and fault early warning of lithium-ion batteries based on MLSTM-FSL model[J]. Signal, Image and Video Processing, 2025, 19(5).

REN Jiahui, MA Jinkai, WANG Honghong, et al. A comprehensive review on research methods for lithium-ion battery of state of health estimation and end of life prediction: methods, properties, and prospects[J]. Protection and Control of modern Power Systems, 2024, 10(3): 146-165.

WANG Shunli, GAO Haiying, TAKYI-ANINAKWA P, et al. Improved multiple feature-electrochemical thermal coupling modeling of lithium-ion batteries at low-temperature with real-time coefficient correction[J]. Protection and Control of Modern Power Systems, 2024, 9(3): 157-173.

ALI H A A, RAIJMAKERS L H J, CHAYAMBUKA K, et al. A comparison between physics-based Li-ion battery models[J]. Electrochimica Acta, 2024, 493.

WU Chunling, XU Chenfeng, WANG Liding, et al. Lithium-ion battery remaining useful life prediction based on data-driven and particle filter fusion model[J]. Green Energy and Intelligent Transportation, 2025, 4(5).

TIAN Linghu, YUAN Bingxia. Prediction of ion battery remaining life of energy storage system based on data preprocessing and computer VMD-LSTM-GPR[J]. Energy Storage Science and Technology, 2024, 13(1): 336-338.

DURMUS F, KARAGOL S. Lithium-ion battery capacity prediction with GA-optimized CNN, RNN, and BP[J]. Applied Sciences, 2024, 14(13): 56-62.

LI Zihan, BAI Fang, ZUO Hongfu, et al. Remaining useful life prediction for lithium-ion batteries based on iterative transfer learning and mogrifier LSTM[J]. Batteries, 2023, 9(9).

KIM R, KIM Y, LEE D H, et al. Thermal-electrochemical effect on the degradation of lithium-ion batteries during the charging process[J]. International Communications in Heat and Mass Transfer, 2024, 158.

ZHANG Wen, PRANAV R S B, WANG Rui, et al. Lithium-ion battery life prediction using deep transfer learning[J]. Batteries, 2024, 10(12).

SHI Yongsheng, WANG Leicheng, LIAO Na, et al. Lithium-ion battery degradation based on the CNN-transformer model[J]. Energies, 2025, 18(2).

ZHOU Kaile, ZHANG Zhiyue. Remaining useful life prediction of lithium-ion batteries based on data denoising and improved transformer[J]. Journal of Energy Storage, 2024, 100.

NAAZ F, HERLE A, CHANNEGOWDA J, et al. A generative adversarial network-based synthetic data augmentation technique for battery condition evaluation[J]. International Journal of Energy Research, 2021, 45(13): 19120-19135.

JIANG Lidang, HU Changyan, JI Sibei, et al. Generating comprehensive lithium battery charging data with generative AI[J]. Applied Energy, 2025, 377.

JI Shanling, ZHU Jianxiong, LÜ Zhiyang, et al. Deep learning enhanced lithium-ion battery nonlinear fading prognosis[J]. Journal of Energy Chemistry, 2023, 78: 565-573.

TANG Aihua, JIANG Yihan, YU Quanqing, et al. A hybrid neural network model with attention mechanism for state of health estimation of lithium-ion batteries[J]. Journal of Energy Storage, 2023, 68.

YAO Xingyan, CHEN Guolin, PECHT M, et al. A novel graph-based framework for state of health prediction of lithium-ion battery[J]. Journal of Energy Storage, 2023, 58.

KONG Ziqian, JIN Xiaohang, XU Zhengguo, et al. Spatio-temporal fusion attention: a novel approach for remaining useful life prediction based on graph neural network[J]. IEEE Transactions on Instrumentation and Measurement, 2022, 71: 1-12.

SUH S, MITTAL D A, BELLO H, et al. Remaining useful life prediction of lithium-ion batteries using spatio-temporal multimodal attention networks[J]. Heliyon, 2024, 10(16).

WU Zonghan, PAN Shirui, CHEN Fengwen, et al. A comprehensive survey on graph neural networks[J]. IEEE Transactions on Neural Networks and Learning Systems, 2020, 32(1): 4-24.

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

State-of-health estimation method for lithium-ion battery based on a coupled degradation model and CNN-GRU-AE fusion network

数据缺失下基于变递推间隔修正辅助模型的电池荷电状态估计

基于变分自编码器的多源数据融合窃电检测方法

基于自适应噪声完全集合经验模态分解与BiLSTM-Transformer的锂离子电池剩余使用寿命预测

不同驾驶习惯下车用电池电流特征与容量衰退的关联性研究

Related Author

YANG Jian

ZHANG Hua

DING Hongshuai

WANG Weiqing

WU Jiahui

WANG Ziyi

ZHAO Jinbin

MAO Ling

Related Institution

The Fourth Construction Co., Ltd. of China Electronics System Engineering

Chongqing Changan Automobile Company Limited

Offshore Wind Power Research Institute, Shanghai University of Electric Power

Shanghai Power T&T Engineering Company Co., Ltd.

BI Junhui(1. College of Electrical Engineering, Shanghai University of Electric Power

AI问答

⁰