基于时序模式注意力机制和孤立森林的电池过充热失控预警方法研究

doi:10.19799/j.cnki.2095-4239.2025.0301

储能科学与技术 ›› 2025, Vol. 14 ›› Issue (10): 3742-3754.doi: 10.19799/j.cnki.2095-4239.2025.0301

基于时序模式注意力机制和孤立森林的电池过充热失控预警方法研究

李昌豪¹(), 曹志成², 汪书苹¹, 谢恒³, 张炜鑫²(), 曹元成²

^1.国网安徽省电力有限公司电力科学研究院，安徽省新型电力系统火灾安全与应急技术重点实验室(国家电网公司输变电设施火灾防护实验室)，安徽合肥 230601
^2.国网安徽省电力有限公司，安徽合肥 230601
^3.华中科技大学电气与电子工程学院，湖北武汉 430070

收稿日期:2025-03-26 修回日期:2025-05-11 出版日期:2025-10-28 发布日期:2025-10-20
通讯作者: 张炜鑫 E-mail:346550617@qq.com;weixinzhang@hust.edu.cn
作者简介:李昌豪（1995—），男，硕士研究生，专责/工程师，研究方向为储能安全防护，E-mail：346550617@qq.com；
基金资助:
国家电网有限公司总部管理科技项目资助(5500-202220118A-1-1-ZN)

Warning method for battery-overcharge thermal runaway based on a temporal pattern attention mechanism and an isolated forest algorithm

Changhao LI¹(), Zhicheng CAO², Shuping WANG¹, Heng XIE³, Weixin ZHANG²(), Yuancheng CAO²

^1.Electric Power Science Research Institute of State Grid Anhui Electric Power Co. , Ltd. , Key Laboratory of Fire Safety and Emergency Technology for New Power Systems in Anhui Province (Fire Protection Laboratory for Transmission and Transformation Facilities of State Grid Corporation of China), Hefei 230601, Anhui, China
^2.State Grid Anhui Electric Power Co. , Ltd. , Hefei 230601, Anhui, China
^3.School of Electrical and Electronic Engineering, Huazhong University of Science and Technology, Wuhan 430070, Hubei, China

Received:2025-03-26 Revised:2025-05-11 Online:2025-10-28 Published:2025-10-20
Contact: Weixin ZHANG E-mail:346550617@qq.com;weixinzhang@hust.edu.cn

摘要/Abstract

摘要：

锂离子电池热失控反应迅速且剧烈，能够在储能系统内形成热蔓延，造成巨大损失，成为储能安全领域亟须解决的问题。为满足锂离子电池过充热失控预警的时效性需求，本工作提出了一种基于时序模式注意力机制(TPA)和双向长短期记忆网络(BiLSTM)联合孤立森林的电池过充热失控预警方法。该方法首先通过TPA从多时间步捕获电池状态特征的变化规律，并通过差异化加权关注重要信息；然后利用BiLSTM的双向神经网络结构对电池特征数据进行双向信息提取，提高模型预测精度；最后结合孤立森林算法，利用电池真实数据集建立孤立森林模型并计算电池状态特征的异常分数，通过选取最佳异常分数阈值对电池状态进行分类，实现对电池异常状态的预警。实验结果表明，该方法的F1-score达到0.9509，能够在电池过充前7 s对电池异常状态进行预警，相比温度阈值划分方法提前了252 s。本工作所述方法在提高锂离子电池过充热失控预警的准确性和时效性方面具有重要意义。

关键词: 锂离子电池, 热失控, 预警方法, 机制, 孤立森林

Abstract:

Lithium-ion battery (LIB) thermal runaway proceeds rapidly and violently representing a significant challenging in the field of energy-storage safety, as it can initiate thermal propagation within energy storage systems and lead to losses. To meet the timeliness requirements for early warning of overcharge-induced thermal runaway in LIBs, an early warning method integrating the temporal pattern attention (TPA) mechanism, bidirectional long short-term memory (BiLSTM) network, and isolation forest algorithm is proposed. In this method, the changing patterns of battery-state characteristics across multiple time steps are first captured using TPA, which applies differentiated weighting to focus on the mostvaluable information. Subsequently, the bidirectional neural network structure of BiLSTM is used to extract bidirectional information from battery-characteristic data, thus enhancing the prediction accuracy of the model. Finally, by integrating the isolation forest algorithm, the real battery dataset is used to establish an isolation forest model, which calculates the anomaly scores of battery-state characteristics. Thereafter, the battery states are classified by selecting the optimal anomaly score threshold, enabling early warning of abnormal battery conditions. The experimental results reveal that the proposed method achieves an F1 score of 0.9509 and can provide early warning of abnormal battery conditions 7 s before overcharge-induced thermal runaway; this time is 252 s earlier than that of the temperature threshold method. Overall, the method demonstrated here offers insights into enhancing the accuracy and timeliness of early warning for overcharge-induced thermal runaway in LIBs.

Key words: lithium-ion battery, thermal runaway, warning methods, mechanism, isolation forest

中图分类号:

TM 912

李昌豪, 曹志成, 汪书苹, 谢恒, 张炜鑫, 曹元成. 基于时序模式注意力机制和孤立森林的电池过充热失控预警方法研究[J]. 储能科学与技术, 2025, 14(10): 3742-3754.

Changhao LI, Zhicheng CAO, Shuping WANG, Heng XIE, Weixin ZHANG, Yuancheng CAO. Warning method for battery-overcharge thermal runaway based on a temporal pattern attention mechanism and an isolated forest algorithm[J]. Energy Storage Science and Technology, 2025, 14(10): 3742-3754.

图/表 20

图1

图2

图3

图4

表1

异常分数结果判定"

$g (x)$	$f (x, n)$	结论
→ $C (n)$	→ $T$	无明显异常
→ $0$	→1	异常
→ $n - 1$	→0	正常

表1

表2

表3

图5

表4

图6

图7

表5

图8

图9

表6

图10

图11

图12

图13

表7

参考文献 25

[1]	喻航, 张英, 徐超航, 等. 锂电储能系统热失控防控技术研究进展[J]. 储能科学与技术, 2022, 11(8): 2653-2663. DOI: 10.19799/j.cnki.2095-4239.2022.0116.
	YU H, ZHANG Y, XU C H, et al. Research progress of thermal runaway prevention and control technology for lithium battery energy storage systems[J]. Energy Storage Science and Technology, 2022, 11(8): 2653-2663. DOI: 10.19799/j.cnki.2095-4239.2022.0116.
[2]	管敏渊, 沈建良, 徐国华, 等. 锂离子电池储能系统靶向消防装备设计与性能[J]. 储能科学与技术, 2023, 12(4): 1131-1138. DOI: 10.19799/j.cnki.2095-4239.2022.0719.
	GUAN M Y, SHEN J L, XU G H, et al. Design and performance research of targeted-fire fighting equipment for lithium-ion battery energy storage system[J]. Energy Storage Science and Technology, 2023, 12(4): 1131-1138. DOI: 10.19799/j.cnki.2095-4239.2022.0719.
[3]	FENG X N, ZHENG S Q, REN D S, et al. Investigating the thermal runaway mechanisms of lithium-ion batteries based on thermal analysis database[J]. Applied Energy, 2019, 246: 53-64. DOI: 10.1016/j.apenergy.2019.04.009.
[4]	周剑文. 锂离子电池热失控建模与热蔓延抑制研究[D]. 济南: 山东大学, 2022. DOI: 10.27272/d.cnki.gshdu.2022.004758.
	ZHOU J W. Thermal runaway modeling and thermal spread suppression of lithium-ion batteries[D]. Jinan: Shandong University, 2022. DOI: 10.27272/d.cnki.gshdu.2022.004758.
[5]	李钊, 陈才星, 牛慧昌, 等. 锂离子电池热失控早期预警特征参数分析[J]. 消防科学与技术, 2020, 39(2): 146-149.
	LI Z, CHEN C X, NIU H C, et al. Characteristic parameter analysis of thermal runaway early warning of lithium-ion battery[J]. Fire Science and Technology, 2020, 39(2): 146-149.
[6]	OMAER FARUQ GONI M, MONDAL M N I, RIAZUL ISLAM S M, et al. Diagnosis of malaria using double hidden layer extreme learning machine algorithm with CNN feature extraction and parasite inflator[J]. IEEE Access, 2023, 11: 4117-4130.
[7]	王娜, 李强. 大数据分析管理系统在新能源汽车事故分析中的应用[J]. 时代汽车, 2024(2): 192-194. DOI: 10.3969/j.issn.1672-9668. 2024.02.061.
	WANG N, LI Q. Application of big data analysis management system in new energy vehicle accident analysis[J]. Auto Time, 2024(2): 192-194. DOI: 10.3969/j.issn.1672-9668.2024.02.061.
[8]	周炜航, 叶青, 叶蕾, 等. 锂离子电池内温度场健康状态分布式光纤原位监测技术研究[J]. 中国激光, 2020, 47(12): 1204002. DOI: 10.3788/CJL202047.1204002.
	ZHOU W H, YE Q, YE L, et al. Distributed optical fiber in situ monitoring technology for a healthy temperature field in lithium ion batteries[J]. Chinese Journal of Lasers, 2020, 47(12): 1204002. DOI: 10.3788/CJL202047.1204002.
[9]	蒋建杰, 楼平, 徐国华, 等. 基于预测误差的锂离子电池热失控预警方法研究[J]. 储能科学与技术, 2024, 13(11): 4187-4197. DOI: 10.19799/j.cnki.2095-4239.2024.0539.
	JIANG J J, LOU P, XU G H, et al. Research on lithium-ion battery thermal runaway early warning method based on prediction error[J]. Energy Storage Science and Technology, 2024, 13(11): 4187-4197. DOI: 10.19799/j.cnki.2095-4239.2024.0539.
[10]	WANG Q S, MAO B B, STOLIAROV S I, et al. A review of lithium ion battery failure mechanisms and fire prevention strategies[J]. Progress in Energy and Combustion Science, 2019, 73: 95-131. DOI: 10.1016/j.pecs.2019.03.002.
[11]	JIN Y, ZHENG Z K, WEI D H, et al. Detection of micro-scale Li dendrite via H₂ gas capture for early safety warning[J]. Joule, 2020, 4(8): 1714-1729. DOI: 10.1016/j.joule.2020.05.016.
[12]	杨启帆, 马宏忠, 段大卫, 等. 基于气体特性的锂离子电池热失控在线预警方法[J]. 高电压技术, 2022, 48(3): 1202-1211. DOI: 10.13336/j.1003-6520.hve.20210261.
	YANG Q F, MA H Z, DUAN D W, et al. Thermal runaway online warning method for lithium-ion battery based on gas characteristics[J]. High Voltage Engineering, 2022, 48(3): 1202-1211. DOI: 10.13336/j.1003-6520.hve.20210261.
[13]	梅文昕, 段强领, 王青山, 等. 大型磷酸铁锂电池高温热失控模拟研究[J]. 储能科学与技术, 2021, 10(1): 202-209. DOI: 10.19799/j.cnki.2095-4239.2020.0249.
	MEI W X, DUAN Q L, WANG Q S, et al. Thermal runaway simulation of large-scale lithium iron phosphate battery at elevated temperatures[J]. Energy Storage Science and Technology, 2021, 10(1): 202-209. DOI: 10.19799/j.cnki.2095-4239.2020.0249.
[14]	PAN Y, FENG X N, ZHANG M X, et al. Internal short circuit detection for lithium-ion battery pack with parallel-series hybrid connections[J]. Journal of Cleaner Production, 2020, 255: 120277. DOI: 10.1016/j.jclepro.2020.120277.
[15]	GAO W K, ZHENG Y J, OUYANG M G, et al. Micro-short-circuit diagnosis for series-connected lithium-ion battery packs using mean-difference model[J]. IEEE Transactions on Industrial Electronics, 2019, 66(3): 2132-2142. DOI: 10.1109/TIE.2018.2838109.
[16]	FENG X N, WENG C H, OUYANG M G, et al. Online internal short circuit detection for a large format lithium ion battery[J]. Applied Energy, 2016, 161: 168-180. DOI: 10.1016/j.apenergy. 2015.10.019.
[17]	黄彧, 王占国, 张言茹, 等. 基于离群点检测的动力电池一致性快速辨识方法[J]. 电测与仪表, 2023, 60(10): 66-72. DOI: 10.19753/j.issn1001-1390.2023.10.011.
	HUANG Y, WANG Z G, ZHANG Y R, et al. A fast identification method of power battery consistency based on outlier detection[J]. Electrical Measurement & Instrumentation, 2023, 60(10): 66-72. DOI: 10.19753/j.issn1001-1390.2023.10.011.
[18]	刘鹏, 吴志强, 张照生, 等. 基于电压频域特征和异常系数的动力电池故障诊断方法[J]. 中国公路学报, 2022, 35(8): 89-104. DOI: 10.19721/j.cnki.1001-7372.2022.08.009.
	LIU P, WU Z Q, ZHANG Z S, et al. Fault diagnosis for battery systems based on voltage frequency-domain indicator and abnormal coefficient[J]. China Journal of Highway and Transport, 2022, 35(8): 89-104. DOI: 10.19721/j.cnki.1001-7372.2022.08.009.
[19]	HONG J C, WANG Z P, CHEN W, et al. Synchronous multi-parameter prediction of battery systems on electric vehicles using long short-term memory networks[J]. Applied Energy, 2019, 254: 113648. DOI: 10.1016/j.apenergy.2019.113648.
[20]	ZHOU Z X, HUBER N R, INOUE A, et al. Multislice input for 2D and 3D residual convolutional neural network noise reduction in CT[J]. Journal of Medical Imaging, 2023, 10(1): 014003. DOI: 10.1117/1.JMI.10.1.014003.
[21]	姚越, 刘达. 基于注意力机制的卷积神经网络-长短期记忆网络的短期风电功率预测[J]. 现代电力, 2022, 39(2): 212-218. DOI: 10. 19725/j.cnki.1007-2322.2021.0108.
	YAO Y, LIU D. Short-term wind power forecasting based on attention mechanism of CNN-LSTM[J]. Modern Electric Power, 2022, 39(2): 212-218. DOI: 10.19725/j.cnki.1007-2322.2021.0108.
[22]	CHANG S Z, DU B, ZHANG L P. A subspace selection-based discriminative forest method for hyperspectral anomaly detection[J]. IEEE Transactions on Geoscience and Remote Sensing, 2020, 58(6): 4033-4046. DOI: 10.1109/TGRS.2019.2960391.
[23]	OJO O, LANG H X, KIM Y, et al. A neural network based method for thermal fault detection in lithium-ion batteries[J]. IEEE Transactions on Industrial Electronics, 2021, 68(5): 4068-4078. DOI: 10.1109/TIE.2020.2984980.
[24]	LI D, LIU P, ZHANG Z S, et al. Battery thermal runaway fault prognosis in electric vehicles based on abnormal heat generation and deep learning algorithms[J]. IEEE Transactions on Power Electronics, 2022, 37(7): 8513-8525. DOI: 10.1109/TPEL.2022. 3150026.
[25]	CHEN Y, XU M, WU Z, et al. Enhanced CNN-LSTM model with self-attention mechanism for time series forecasting[J]. IEEE Access, 2023, 11: 34567-34578.

名称	配置信息
操作系统	Ubuntu 22.04
开发语言	Python 3.8.0
框架	Pytorch 1.10.+cuda 11.1
CPU	I9-11900K 11^th
GPU	RTX 3090ti 12GB
内存	200GB

窗口大小	压力预测RMSE	温度预测RMSE
16	2.3358	0.0535
32	3.0370	0.0475
48	2.2097	0.0432
64	2.8472	0.0478

网络结构	MAPE/%	RMSE	MAE
LSTM^[21]	0.1873	0.0594	0.0435
CNN-LSTM^[25]	0.1530	0.0503	0.0415
TPA-LSTM	0.1237	0.0496	0.0367
Attention-BiLSTM	0.1385	0.0503	0.0398
TPA-BiLSTM	0.1144	0.0432	0.0341

网络结构	MAPE/%	RMSE	MAE
LSTM^[21]	0.1472	3.2776	14.227
CNN-LSTM^[25]	0.1582	3.4049	14.475
TPA-LSTM	0.1406	2.7023	13.609
Attention-BiLSTM	0.1234	2.6178	13.526
TPA-BiLSTM	0.1108	2.2097	10.343

预警方法	F1-score	Precision	提前报警时间/s
方法1	0.7443	1	214
方法2	0.8742	0.9276	306
方法3	0.9161	0.9816	306
本工作方法1	0.9688	1	300
本工作方法2	0.9509	1	319

基于时序模式注意力机制和孤立森林的电池过充热失控预警方法研究

Warning method for battery-overcharge thermal runaway based on a temporal pattern attention mechanism and an isolated forest algorithm

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 20

参考文献 25

相关文章 15

编辑推荐

Metrics

本文评价

[1]	陈晔, 李晋, 赵瑞兰, 张少禹, 储玉喜, 杨康, 廖晓雪, 蒋波, 卓萍. 不同荷电状态下电池模组热失控传播对比试验[J]. 储能科学与技术, 2025, 14(9): 3402-3413.
[2]	谈秀雯, 李凌. 局部过热下锂电池热失控特性及其热管理研究[J]. 储能科学与技术, 2025, 14(9): 3521-3529.
[3]	陈峥, 胡竞元, 赵志刚, 申江卫, 夏雪磊, 魏福星. 双体系混装电池组热特性研究及风冷散热结构优化[J]. 储能科学与技术, 2025, 14(9): 3463-3475.
[4]	陈文艳, 贺瑞璘, 常建, 邓永红. 不同形态液态金属电极的储锂机制研究[J]. 储能科学与技术, 2025, 14(9): 3290-3300.
[5]	邓拓, 周海平, 刘煜, 刘畅, 李梓恺, 吴孟强. 化学气相沉积法制备硅碳负极的研究进展[J]. 储能科学与技术, 2025, 14(9): 3354-3372.
[6]	赵岩, 刘浩, 易宗琳, 李莉, 谢莉婧, 苏方远. FEC与VC在锂离子电池石墨负极界面行为研究[J]. 储能科学与技术, 2025, 14(9): 3249-3258.
[7]	钱艺华, 赵耀洪, 王青, 郭鹏, 裴大婷, 曾以柔. 钠离子卤化物固态电解质研究进展与展望[J]. 储能科学与技术, 2025, 14(9): 3389-3401.
[8]	林季锦, 刘倩, 曲涛, 李京鲲, 黄东永, 朱晓庆, 巨星. 锂离子电池储能系统浸没液冷的技术经济性分析[J]. 储能科学与技术, 2025, 14(9): 3622-3635.
[9]	储玉喜, 马畅, 陈红光, 张少禹, 卓萍. 180 Ah钠离子电池热失控与产气特性分析[J]. 储能科学与技术, 2025, 14(9): 3611-3618.
[10]	徐成善, 李涵, 王炎, 卢兰光, 冯旭宁, 欧阳明高. 双层储能电池火蔓延特性及触发过程能量传递机制研究[J]. 储能科学与技术, 2025, 14(9): 3552-3563.
[11]	刘明轩, 陈文涛, 申韶鹏, 张世杰, 韦振, 马彪, 李丹华, 刘仕强, 王芳. 储能用锂离子电池加速老化及老化后安全特性研究[J]. 储能科学与技术, 2025, 14(9): 3530-3537.
[12]	封居强, 张成知, 陈雨杭. 基于数字孪生的高精度SOC和温度联合估计方法[J]. 储能科学与技术, 2025, 14(9): 3567-3580.
[13]	白晓宇, 筵亚静, 张志荣, 孔令丽. 复合石墨锂离子电池性能研究[J]. 储能科学与技术, 2025, 14(9): 3259-3268.
[14]	张磊. 锂离子电池储能电站的运行状态监测与评估[J]. 储能科学与技术, 2025, 14(9): 3538-3540.
[15]	包新宇, 孔祥栋, 吕桃林, 朱志成, 韩雪冰, 来鑫, 郑岳久, 孙涛. 基于产线大数据的电池内阻预测及快速分选方法[J]. 储能科学与技术, 2025, 14(9): 3541-3551.