融合REMD-NHITS与KDE的锂离子电池健康状态退化预测

姬钰培; 吴康佳; 袁泽宇; 闫俊义; 崔方舒; 乔钢柱

doi:10.19799/j.cnki.2095-4239.2026.0340

您当前的位置：

首页 >

文章列表页 >

融合REMD-NHITS与KDE的锂离子电池健康状态退化预测

储能测试与评价 | 更新时间：2026-08-26

- 融合REMD-NHITS与KDE的锂离子电池健康状态退化预测
- Prediction of lithium-ion battery state of health degradation by integrating REMD-NHITS and KDE
- 储能科学与技术 2026年15卷第8期页码：3324-3340
- 作者机构：
  
  1.中北大学计算机科学与技术学院，山西太原 030051
  2.中北大学极限环境光电动态测试技术与仪器全国重点实验室，山西太原 030051
- 作者简介：
  
  姬钰培（2002—），女，硕士研究生，研究方向为储能电池健康管理，E-mail：jyupeia@163.com；
  乔钢柱，教授，研究方向为物联网技术及应用、大数据处理技术等，E-mail：qiaogz@nuc.edu.cn。
- 基金信息：
  
  山西省基础研究计划(202303021222084;202303021222120)
- DOI：10.19799/j.cnki.2095-4239.2026.0340
  中图分类号： TM 912
- 收稿：2026-04-17，
  
  修回：2026-06-19，
  
  网络首发：2026-08-26，
  
  纸质出版：2026-08-28
- 稿件说明：
移动端阅览
姬钰培, 吴康佳, 袁泽宇, 等. 融合REMD-NHITS与KDE的锂离子电池健康状态退化预测[J]. 储能科学与技术, 2026, 15(8): 3324-3340.

JI Yupei, WU Kangjia, YUAN Zeyu, et al. Prediction of lithium-ion battery state of health degradation by integrating REMD-NHITS and KDE[J]. Energy Storage Science and Technology, 2026, 15(8): 3324-3340.
姬钰培, 吴康佳, 袁泽宇, 等. 融合REMD-NHITS与KDE的锂离子电池健康状态退化预测[J]. 储能科学与技术, 2026, 15(8): 3324-3340. DOI： 10.19799/j.cnki.2095-4239.2026.0340.

JI Yupei, WU Kangjia, YUAN Zeyu, et al. Prediction of lithium-ion battery state of health degradation by integrating REMD-NHITS and KDE[J]. Energy Storage Science and Technology, 2026, 15(8): 3324-3340. DOI： 10.19799/j.cnki.2095-4239.2026.0340.

摘要

锂离子电池健康状态（state of health，SOH）的准确预测对于保障储能系统安全稳定运行具有重要意义。为突破现有锂离子电池SOH预测方法在精度上的局限，提出一种融合鲁棒经验模态分解（robust empirical mode decomposition，REMD）、时间序列神经层次插值（neural hierarchical interpolation for time series，NHITS）模型与核密度估计（kernel density estimation，KDE）的锂离子电池SOH退化预测模型。该方法从锂离子电池退化数据中提取可有效表征电池SOH退化规律的健康因子（health indicator，HI），采用REMD对HI实施自适应分解以获取多个分量；在各分量上分别构建NHITS预测模型进行独立预测，最终通过分量重构得到电池SOH点预测结果。在此基础上，引入KDE对点预测误差分布进行概率建模，建立REMD-NHITS-KDE区间预测框架，实现SOH预测不确定性的量化表征与区间估计。以CALCE数据集电池和NASA数据集电池为实验对象，实验结果表明所提方法在点预测任务中优于NHITS、NBEATS和REMD-NBEATS等模型，平均绝对误差低至0.0047，决定系数提高至99.98%；在区间预测中兼顾了覆盖率与区间宽度，表现出较好的可靠性与实用性。

Abstract

To overcome the accuracy limitations of existing state-of-health (SOH) prediction methods for lithium-ion batteries

a lithium-ion battery SOH degradation prediction model integrating robust empirical mode decomposition (REMD)

neural hierarchical interpolation for time series (NHITS)

and kernel density estimation (KDE) is proposed. The method extracts health indicators (HIs) that effectively represent SOH degradation from lithium-ion battery data and uses REMD to adaptively decompose these HIs into multiple components. NHITS models are constructed for independent prediction on each component

and SOH point predictions are obtained through component reconstruction. Based on this

KDE probabilistically models the distribution of point prediction errors

establishing an REMD-NHITS-KDE interval-prediction framework that quantitatively characterizes and estimates SOH prediction uncertainty. Using batteries from the CALCE and NASA datasets

the experimental results show that the proposed method outperforms models such as NHITS

NBEATS

and REMD-NBEATS in point prediction

achieving a mean absolute error of 0.0047 and a coefficient of determination of 99.98%. In interval prediction

it considers coverage and interval width

demonstrating strong reliability and practicality.

关键词

Keywords

references

李炳金, 韩晓霞, 张文杰, 等. 锂离子电池剩余使用寿命预测方法综述[J]. 储能科学与技术, 2024, 13(4): 1266-1276.

LI B J, HAN X X, ZHANG W J, et al. Review of the remaining useful life prediction methods for lithium-ion batteries[J]. Energy Storage Science and Technology, 2024, 13(4): 1266-1276.

丁赛. 新能源汽车电池盒电泳膜厚仿真研究[J]. 电镀与涂饰, 2024, 43(9): 135-142. DOI:10.19289/j.1004-227x.2024.09.015.

DING S. Simulative study on electrophoretic coating thickness of battery enclosure of new energy car[J]. Electroplating & Finishing, 2024, 43(9): 135-142. DOI:10.19289/j.1004-227x.2024.09.015.

LI Y, SHI H T, WANG S L, et al. A comprehensive review of remaining useful life prediction methods for lithium-ion batteries: Models, trends, and engineering applications[J]. Journal of Energy Chemistry, 2026, 112: 384-414. DOI:10.1016/j.jechem. 2025.08.056.

李卓昊, 石琼林, 王康丽, 等. 锂离子电池健康状态估计方法研究现状与展望[J]. 电力系统自动化, 2024, 48(20): 109-129. DOI:10.7500/AEPS20231221006.

LI Z H, SHI Q L, WANG K L, et al. Research status and prospects of state of health estimation methods for lithium-ion batteries[J]. Automation of Electric Power Systems, 2024, 48(20): 109-129. DOI:10.7500/AEPS20231221006.

AMIR S, GULZAR M, TARAR M O, et al. Dynamic equivalent circuit model to estimate state-of-health of lithium-ion batteries[J]. IEEE Access, 2022, 10: 18279-18288. DOI:10.1109/ACCESS. 2022.3148528.

段慧云, 夏威, 邵杰, 等. 基于充电健康因子优化和数据驱动的锂电池剩余使用寿命预测[J]. 汽车技术, 2024(1): 20-26.

DUAN H Y, XIA W, SHAO J, et al. A data-driven remaining useful life prediction approach for lithium-ion batteries based on charging health feature optimization[J]. Automobile Technology, 2024(1): 20-26.

董作林, 宋金岩, 孟子迪. 基于模态分解和深度学习的锂离子电池寿命预测[J]. 储能科学与技术, 2025, 14(4): 1645-1653. DOI:10.19799/j.cnki.2095-4239.2024.0983.

DONG Z L, SONG J Y, MENG Z D. Lithium-ion battery life prediction based on mode decomposition and deep learning[J]. Energy Storage Science and Technology, 2025, 14(4): 1645-1653. DOI:10.19799/j.cnki.2095-4239.2024.0983.

ZHU G Q, HE C, CHEN Y L, et al. A hybrid parallel informer-LSTM framework based on two-stage decomposition for lithium battery remaining useful life prediction[J]. Energies, 2026, 19(3): 612. DOI:10.3390/en19030612.

衣思彤, 刘雅浓, 马耀浥, 等. 基于贝叶斯优化-卷积神经网络-双向长短期记忆神经网络的锂电池健康状态评估[J]. 电气技术, 2024, 25(5): 1-10, 21. DOI:10.3969/j.issn.1673-3800.2024.05.001.

YI S T, LIU Y N, MA Y Y, et al. State of health assessment of lithium battery based on Bayesian optimization-convolution neural network-bi-directional long short term memory neural network[J]. Electrical Engineering, 2024, 25(5): 1-10, 21. DOI:10.3969/j.issn.1673-3800.2024.05.001.

程思涵, 刘思懿, 郭子旭, 等. 基于CNN-GRU-注意力的锂离子电池SOC估计[J]. 电池, 2026, 56(1): 131-135. DOI:10.19535/j.1001-1579.2026.01.019.

CHENG S H, LIU S Y, GUO Z X, et al. SOC estimation for Li-ion battery based on the CNN-GRU-Attention[J]. Battery Bimonthly, 2026, 56(1): 131-135. DOI:10.19535/j.1001-1579.2026.01.019.

ORESHKIN B N, DUDEK G, PEŁKA P, et al. N-BEATS neural network for mid-term electricity load forecasting[J]. Applied Energy, 2021, 293: 116918. DOI:10.1016/j.apenergy.2021.116918.

LEE M, LEE Y, KIM G. Transfer learning based on N-BEATS in forecasting univariate time series[J]. IEEE Access, 2026, 14: 45191-45212. DOI:10.1109/ACCESS.2026.3675559.

CHALLU C, OLIVARES K G, ORESHKIN B N, et al. NHITS: Neural hierarchical interpolation for time series forecasting[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2023, 37(6): 6989-6997. DOI:10.1609/aaai.v37i6.25854.

姚芳, 韩永康, 李谦, 等. 基于多尺度健康因子-BEAST分解和SARIMA模型结合的锂离子电池剩余使用寿命预测[J]. 天津大学学报(自然科学与工程技术版), 2026, 59(1): 77-89.

YAO F, HAN Y K, LI Q, et al. Remaining useful life prediction of lithium-ion batteries based on combination of multiscale health indicators-BEAST decomposition and SARIMA modeling[J]. Journal of Tianjin University (Science and Technology), 2026, 59(1): 77-89.

张俊贤, 周英超, 李波, 等. 融合EMD与GAIPSO-LSTM算法的锂离子电池RUL预测方法研究[J]. 重庆理工大学学报(自然科学), 2025, 39(6): 28-36.

ZHANG J X, ZHOU Y C, LI B, et al. Research on RUL prediction method of lithium-ion batteries based on EMD and GAIPSO-LSTM algorithm[J]. Journal of Chongqing University of Technology (Natural Science), 2025, 39(6): 28-36.

依沙克·司马义, 陈昊, 张正强, 等. 基于NWP信息的SSA优化EEMD-LSTM风电超短期功率预测[J]. 太阳能学报, 2025, 46(8): 176-183. DOI:10.19912/j.0254-0096.tynxb.2024-0534.

SIMAYI Y, CHEN H, ZHANG Z Q, et al. Ultra-short-term wind power prediction based on nwp information using ssa optimized eemd-lstm[J]. Acta Energiae Solaris Sinica, 2025, 46(8): 176-183. DOI:10.19912/j.0254-0096.tynxb.2024-0534.

ZHANG L Y, WANG Z Y, LI L. Bearing fault identification based on CEEMDAN-SVD and CNN[J]. Mechanical Science and Engineering, 2024: DOI:10.57237/j.mse.2024.01.002.

乔松博, 孙瑜, 胡海, 等. 基于REMD-CNN-Transformer-LSTM组合模型的碳排放交易价格预测[J]. 西安理工大学学报, 2025, 41(2): 186-196. DOI:10.19322/j.cnki.issn.1006-4710.2025.02.004.

QIAO S B, SUN Y, HU H, et al. Carbon emission trading price prediction based on REMD-CNN-Transformer-LSTM combined model[J]. Journal of Xi'an University of Technology, 2025, 41(2): 186-196. DOI:10.19322/j.cnki.issn.1006-4710.2025.02.004.

CHEN C, TAO G Y, SHI J T, et al. A lithium-ion battery degradation prediction model with uncertainty quantification for its predictive maintenance[J]. IEEE Transactions on Industrial Electronics, 2024, 71(4): 3650-3659. DOI:10.1109/TIE.2023.3274874.

ZHANG Y, ZHANG M, LIU C, et al. Reliability enhancement of state of health assessment model of lithium-ion battery considering the uncertainty with quantile distribution of deep features[J]. Reliability Engineering & System Safety, 2024, 245: 110002. DOI:10.1016/j.ress.2024.110002.

汤霜霜, 白建波, 侯天才, 等. 基于蒙特卡洛法的光伏系统老化评估及发电量预测方法研究[J]. 太阳能学报, 2025, 46(5): 311-319. DOI:10.19912/j.0254-0096.tynxb.2024-0089.

TANG S S, BAI J B, HOU T C, et al. Research on aging assessment and power generation prediction of photovoltaic system based on Monte Carlo method[J]. Acta Energiae Solaris Sinica, 2025, 46(5): 311-319. DOI:10.19912/j.0254-0096.tynxb. 2024-0089.

LIU Z F, HUANG Y H, ZHANG S R, et al. A collaborative interaction gate-based deep learning model with optimal bandwidth adjustment strategies for lithium-ion battery capacity point-interval forecasting[J]. Applied Energy, 2025, 377: 124741. DOI:10.1016/j.apenergy.2024.124741.

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

基于超参数自适应优化的Transformer锂离子电池SOH预测算法

基于短时随机充电数据和优化卷积神经网络的锂电池健康状态估计

基于多时间尺度建模自动特征提取和通道注意力机制的锂离子电池健康状态估计

基于多特征量分析和LSTM-XGBoost模型的锂离子电池SOH估计方法

水系锌碘电池聚合物电解液制备及其电化学性能研究