基于集成学习和分位数回归的锂离子电池寿命预测方法

郑立涵; 林俊楠; 陈平; 施志聪

doi:10.19799/j.cnki.2095-4239.2026.0054

您当前的位置：

首页 >

文章列表页 >

基于集成学习和分位数回归的锂离子电池寿命预测方法

储能测试与评价 | 更新时间：2026-07-13

- 基于集成学习和分位数回归的锂离子电池寿命预测方法
- A framework fusing ensemble learning and quantile regression for lithium-ion battery life prediction
- 储能科学与技术 2026年15卷第6期页码：2380-2391
- 作者机构：
  
  1.广东工业大学材料与能源学院，广东广州 510006
  2.电科电源股份有限公司，广东深圳 518000
  3.衡阳电科电源股份有限公司，湖南衡阳 421000
- 作者简介：
  
  郑立涵（2000—），男，硕士研究生，研究方向为基于机器学习的锂离子电池寿命预测，E-mail：15560520693@163.com；
  施志聪，教授，研究方向为锂离子电池、钠离子电池、金属空气电池等化学电源，E-mail：zhicong@gdut.edu.cn。
- 基金信息：
  
  广东省基础与应用基础研究基金联合基金(2023A1515110792)
- DOI：10.19799/j.cnki.2095-4239.2026.0054
  中图分类号： TM 912
- 收稿：2026-01-19，
  
  修回：2026-01-29，
  
  纸质出版：2026-06-28
- 稿件说明：
移动端阅览
郑立涵, 林俊楠, 陈平, 等. 基于集成学习和分位数回归的锂离子电池寿命预测方法[J]. 储能科学与技术, 2026, 15(6): 2380-2391.

ZHENG Lihan, LIN Junnan, CHEN Ping, et al. A framework fusing ensemble learning and quantile regression for lithium-ion battery life prediction[J]. Energy Storage Science and Technology, 2026, 15(6): 2380-2391.
郑立涵, 林俊楠, 陈平, 等. 基于集成学习和分位数回归的锂离子电池寿命预测方法[J]. 储能科学与技术, 2026, 15(6): 2380-2391. DOI： 10.19799/j.cnki.2095-4239.2026.0054.

ZHENG Lihan, LIN Junnan, CHEN Ping, et al. A framework fusing ensemble learning and quantile regression for lithium-ion battery life prediction[J]. Energy Storage Science and Technology, 2026, 15(6): 2380-2391. DOI： 10.19799/j.cnki.2095-4239.2026.0054.

摘要

锂离子电池寿命的精准预测是保障电池系统安全的关键。针对传统方法依赖复杂特征工程且难以量化预测不确定性的问题，本研究提出一种融合集成学习与分位数回归的锂离子电池寿命预测框架。该框架以搭载了注意力机制的门控循环单元为基学习器，结合引导聚集策略与分位数回归机制，仅需输入经滑动窗口处理的电池容量序列即可完成预测，避免了复杂的特征提取。通过五种不同容量规格电池退化数据验证表明，该框架在不同预测步长下均能保持较高的预测准确性，在300次循环的预测步长下，预测结果的均方根误差、平均绝对百分比误差和平均绝对误差分别低至0.4097%、0.3246%和0.1104%，显著优于传统点预测模型。与单一模型和基础门控循环单元模型相比，该模型在所有测试电池上均取得了最优且最稳定的预测性能，这得益于集成学习策略有效提升了模型的稳定性与泛化能力。在不确定性量化方面，所提框架生成的预测区间对测试电池的真实容量值均实现了有效覆盖，覆盖率显著高于对比模型，证明了其在预测不确定性评估方面的优势。本研究创新性地将集成学习与分位数回归相结合，在实现高精度点预测的同时输出具有95%置信度的预测区间，从而显著提升了锂离子电池寿命预测的准确性与不确定性量化能力。

Abstract

Accurate prediction of lithium-ion battery life is imperative for ensuring the safety of battery systems. Traditional methods rely on complex feature engineering and struggle to quantify prediction uncertainty. To address these limitations

this paper proposes a lithium-ion battery life prediction framework integrating ensemble learning and quantile regression. The proposed framework utilizes a Gated Recurrent Unit (GRU) augmented with an attention mechanism as the fundamental learner. Incorporating a Bootstrap Aggregating strategy and quantile regression

the approach necessitates only battery capacity sequences processed via a sliding window for prediction

thereby eliminating the requirement for complex feature extraction. Validation on degradation data from five lithium-ion batteries with varying capacity specifications demonstrates that the proposed framework maintains high prediction accuracy across different prediction horizons. At a prediction horizon of 300 cycles

the Root Mean Square Error

Mean Absolute Percentage Error

and Mean Absolute Error are as low as 0.4097%

0.3246%

and 0.1104%

respectively

significantly outperforming traditional point prediction models. A comparison of the proposed framework with single models and the standard GRU model reveals that the former attains superior and more stable prediction performance across all tested batteries. This enhancement can be attributed to the ensemble learning strategy

which augments model robustness and generalization. In the context of uncertainty quantification

the prediction intervals generated by the framework effectively encompass the actual capacity values of the test batteries. Furthermore

the coverage rates of the framework are notably higher than those of comparative models

underscoring its superiority in uncertainty assessment. This study innovatively combines ensemble learning with quantile regression

enabling high-precision point predictions while generating 95% confidence prediction intervals. Consequently

this combination significantly improves the accuracy and uncertainty quantification capability of lithium-ion battery life prediction.

关键词

Keywords

references

HUANG J Q, SHEN Z C, ROBERTSON S J, et al. Fluorine grafted gel polymer electrolyte by in situ construction for high-voltage lithium metal batteries[J ] . Chemical Engineering Journal, 2023, 475: 145802. DOI:10.1016/j.cej.2023.145802.

XIE X T, CHEN J B, CHEN X Y, et al. Exploring the effect of lithium halide artificial SEI on the electrochemical performance of lithium metal batteries[J]. Journal of Electroanalytical Chemistry, 2023, 949: 117862. DOI:10.1016/j.jelechem.2023.117862.

LIN Y H, CHEN J H, ZHU J L, et al. In-situ construction of tetraethylene glycol diacrylate based gel polymer electrolyte for long lifespan lithium metal batteries[J ] . Surfaces and Interfaces, 2023, 37: 102737. DOI:10.1016/j.surfin.2023.102737.

SHEN Z C, ZHONG J W, JIANG S Y, et al. Polyacrylonitrile porous membrane-based gel polymer electrolyte by in situ free-radical polymerization for stable Li metal batteries[J ] . ACS Applied Materials & Interfaces, 2022, 14(36): 41022-41036. DOI:10.1021/acsami.2c11397.

CHEN J B, LI D D, LIN K J, et al. Building a stable artificial solid electrolyte interphase on lithium metal anodes toward long-life Li-O 2 batteries[J ] . Journal of Power Sources, 2022, 540: 231603. DOI:10.1016/j.jpowsour.2022.231603.

SHEN Z C, HUANG J Q, XIE Y, et al. Solid electrolyte interphase on lithium metal anodes[J]. ChemSusChem, 2024, 17(11): e202301777. DOI:10.1002/cssc.202301777.

PANG X X, ZHONG S, WANG Y L, et al. A review on the prediction of health state and serving life of lithium-ion batteries[J]. The Chemical Record, 2022, 22(10): e202200131. DOI:10.1002/tcr.202200131.

CHEN D Q, ZHOU X Z. AttMoE: Attention with Mixture of Experts for remaining useful life prediction of lithium-ion batteries[J]. Journal of Energy Storage, 2024, 84: 110780. DOI:10.1016/j.est.2024.110780.

CAI N, QUE X P, ZHANG X, et al. A deep learning framework for the joint prediction of the SOH and RUL of lithium-ion batteries based on bimodal images[J]. Energy, 2024, 302: 131700. DOI:10.1016/j.energy.2024.131700.

EKBERG J, SRIDHARAN N V, KARIM R, et al. State estimation and remaining useful life prediction for lithium-ion batteries[J]. Cell Reports Physical Science, 2025, 6(10): 102854. DOI:10.1016/j.xcrp.2025.102854.

SHI Z L, IBRAHIM A W, XU J Z, et al. Enhanced remaining useful life prediction of lithium-ion battery based on a dual attention hybrid data-driven method[J]. Scientific Reports, 2026, 16: 1303. DOI:10.1038/s41598-025-30849-x.

LUO K, CHEN X, ZHENG H R, et al. A review of deep learning approach to predicting the state of health and state of charge of lithium-ion batteries[J]. Journal of Energy Chemistry, 2022, 74: 159-173. DOI:10.1016/j.jechem.2022.06.049.

XIE Y, LUO K, ZHENG L H, et al. A state of health estimation method for lithium-ion batteries based on initial charging segment and Gated Recurrent Unit neural network[J]. Journal of Power Sources, 2025, 638: 236607. DOI:10.1016/j.jpowsour.2025.236607.

ALSUWIAN T, ANSARI S, AMMIRRUL ATIQI MOHD ZAINURI M, et al. A review of expert hybrid and co-estimation techniques for SOH and RUL estimation in battery management system with electric vehicle application[J]. Expert Systems with Applications, 2024, 246: 123123. DOI:10.1016/j.eswa.2023.123123.

LI X J, YU D, SØREN BYG V, et al. The development of machine learning-based remaining useful life prediction for lithium-ion batteries[J]. Journal of Energy Chemistry, 2023, 82: 103-121. DOI:10.1016/j.jechem.2023.03.026.

LI C F, YANG L, LI Q, et al. SOH estimation method for lithium-ion batteries based on an improved equivalent circuit model via electrochemical impedance spectroscopy[J ] . Journal of Energy Storage, 2024, 86: 111167. DOI:10.1016/j.est.2024.111167.

CHEN L, ZHANG Y, ZHENG Y, et al. Remaining useful life prediction of lithium-ion battery with optimal input sequence selection and error compensation[J]. Neurocomputing, 2020, 414: 245-254. DOI:10.1016/j.neucom.2020.07.081.

JIA J F, LIANG J Y, SHI Y H, et al. SOH and RUL prediction of lithium-ion batteries based on Gaussian process regression with indirect health indicators[J]. Energies, 2020, 13(2): 1-20.

GOMEZ W, WANG F K, CHOU J H. Li-ion battery capacity prediction using improved temporal fusion transformer model[J]. Energy, 2024, 296: 131114. DOI:10.1016/j.energy.2024.131114.

WANG L, QIU Y, YUAN W C, et al. Next-generation battery safety management: Machine learning assisted life-time prediction and performance enhancement[J]. Journal of Energy Chemistry, 2025, 109: 726-739. DOI:10.1016/j.jechem.2025.05.065.

QIU Y, ZHANG X, TIAN Y, et al. Machine learning promotes the development of all-solid-state batteries[J]. Chinese Journal of Structural Chemistry, 2023, 42(9): 100118. DOI:10.1016/j.cjsc. 2023.100118.

CHEN K, LUO Y, LONG Z, et al. Battery state of health estimation using deep transfer learning on short-term charging data[J]. Measurement, 2025, 256: 118233. DOI:10.1016/j.measurement.2025.118233.

GUO J, LI Y Q, MENG J H, et al. Understanding the mechanism of capacity increase during early cycling of commercial NMC/graphite lithium-ion batteries[J]. Journal of Energy Chemistry, 2022, 74: 34-44. DOI:10.1016/j.jechem.2022.07.005.

ANSARI S, AYOB A, HOSSAIN LIPU M S, et al. Optimized data-driven approach for remaining useful life prediction of Lithium-ion batteries based on sliding window and systematic sampling[J]. Journal of Energy Storage, 2023, 73: 109198. DOI:10.1016/j.est.2023.109198.

WANG F K, AMOGNE Z E, CHOU J H, et al. Online remaining useful life prediction of lithium-ion batteries using bidirectional long short-term memory with attention mechanism[J]. Energy, 2022, 254: 124344. DOI:10.1016/j.energy.2022.124344.

LUO K, ZHENG H R, SHI Z C. A simple feature extraction method for estimating the whole life cycle state of health of lithium-ion batteries using transformer-based neural network[J]. Journal of Power Sources, 2023, 576: 233139. DOI:10.1016/j.jpowsour.2023.233139.

HU J M, TANG J W, LIU Z. A novel time series probabilistic prediction approach based on the monotone quantile regression neural network[J]. Information Sciences, 2024, 654: 119844. DOI:10.1016/j.ins.2023.119844.

TIAN Y, DONG Q Y, TIAN J D, et al. Capacity estimation of lithium-ion batteries based on optimized charging voltage section and virtual sample generation[J]. Applied Energy, 2023, 332: 120516. DOI:10.1016/j.apenergy.2022.120516.

TONG Z M, MIAO J Z, MAO J L, et al. Prediction of Li-ion battery capacity degradation considering polarization recovery with a hybrid ensemble learning model[J]. Energy Storage Materials, 2022, 50: 533-542. DOI:10.1016/j.ensm.2022.05.026.

LIN M Q, KE L S, MENG J H, et al. Health status estimation of lithium-ion battery under arbitrary charging voltage information using ensemble learning framework[J]. Reliability Engineering & System Safety, 2025, 256: 110782. DOI:10.1016/j.ress.2024. 110782.

GANAIE M A, HU M H, MALIK A K, et al. Ensemble deep learning: A review[J]. Engineering Applications of Artificial Intelligence, 2022, 115: 105151. DOI:10.1016/j.engappai.2022. 105151.

LUO T T, BAO Z Y, HE Z W, et al. A fusion extended long short-term memory learning architecture for capacity trajectory prediction of Li-ion battery[J]. Journal of Energy Storage, 2025, 128: 116994. DOI:10.1016/j.est.2025.116994.

YAO L, YU C, XIAO Y Q, et al. An improved variational mode decomposition neural network intelligent diagnosis method for battery connection faults based on real vehicle data[J]. Journal of Energy Storage, 2025, 139: 118791. DOI:10.1016/j.est.2025. 118791.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于EEMD-GRU-NN锂离子电池表面温度预测方法研究

基于模态分解的Transformer-GRU联合电池健康状态估计

基于VF-DW-DFN的锂离子电池剩余寿命预测

多特征融合的MVMD-VSN_GRU锂离子电池健康状态预测方法研究

基于超参数自适应优化的Transformer锂离子电池SOH预测算法