基于LOF和数据时空特征的异常锂电池实时检测

doi:10.19799/j.cnki.2095-4239.2025.0318

摘要/Abstract

摘要：

储能系统中的电池模组运行状态复杂，准确识别异常锂电池对于系统的安全性和稳定性至关重要。针对传统异常检测方法存在的实时性不足和对异常样本依赖性强等问题，本工作提出一种融合局部离群因子与电池运行数据时空特征的无监督异常检测方法。该方法充分考虑了电池模组内的单体一致性和运行数据的变化，无需预训练即可实现高效、实时且准确的异常锂电池识别。具体包括：设计基于Cornish-Fisher展开式的分布校正方法以计算自适应阈值；采用滑动窗口机制对储能电站采集的连续数据流进行分段处理，构建动态数据片段，以提升模型对突发异常的响应能力；利用LOF算法对窗口内的时序数据进行局部密度分析，识别密度显著偏低的离群点，实现无监督异常检测。数据集3920~3960时间段的异常检测对比实验结果表明，本方法相较于K-means聚类、隔离森林、香农熵、自编码器等方法，准确识别出了异常的电池155和电池364，检测结果与人工标注完全一致，未出现任何误报或漏检，且所需检测时间最短(平均0.0106 s)，展现出优异的通用性与工程适应性。

关键词: 锂离子电池, 异常检测, 局部异常因子, 滑动窗口, 实时监控

Abstract:

Battery modules exhibit a complex status in energy storage systems, and the accurate identification of abnormal lithium-ion batteries is crucial for system safety and stability. To resolve the shortcomings of traditional anomaly detection methods, such as their insufficient real-time performance and strong dependence on abnormal samples, an unsupervised anomaly detection method that integrates the local outlier factor (LOF) algorithm with spatiotemporal features of battery-operation data is proposed. This method comprehensively explores inter-cell consistency and data variation to achieve efficient, real-time, accurate abnormal celldetection without the need for pre-training. Specifically, the method involves: designing a Cornish-Fisher expansion-based distribution correction approach to calculate thresholds; utilizing a sliding-window mechanism to segment the continuous data stream from energy storage stations into dynamic data slices, thereby improving responsiveness to sudden anomalies; and applying the LOF algorithm for the local-density analysis of time series data within each window to detect low-density outliers and enable unsupervised anomaly detection. Experimental results on the dataset from 3920 to 3960 reveal that the proposed method accurately detects abnormal cells No. 155 and No. 364. These results are fully consistent with manual labeling, with zero false positives or negatives. Furthermore, the proposed method achieves the shortest average detection time of 0.0106 seconds, outperforming traditional methods, such as K-means clustering, isolation forest, Shannon entropy, and autoencoders, thus underscoring its excellent generalizability and engineering adaptability.

Key words: lithium-ion batteries, anomaly detection, local outlier factor, sliding-window, real-time monitoring

中图分类号:

TM 911

刘怡青, 王浩, 陆玲霞, 李昊展, 闫旻睿, 于淼. 基于LOF和数据时空特征的异常锂电池实时检测[J]. 储能科学与技术, 2025, 14(10): 3968-3981.

Yiqing LIU, Hao WANG, Lingxia LU, Haozhan LI, Minrui YAN, Miao YU. Real-time abnormal-battery detection based on the local outlier factor algorithm and the spatiotemporal data features of battery cells[J]. Energy Storage Science and Technology, 2025, 14(10): 3968-3981.

图/表 13

图1

图2

图3

表1

图4

图5

图6

表2

电池簇运行数据样本"

时间	簇总电压(HV1-Vbatt)	簇电流	V1	V2	$⋯$	V364
2024/3/3 23:59	786.4	0	3.212	3.212	$⋯$	3.219
2024/3/3 23:59	786.4	0	3.212	3.212	$⋯$	3.219
$⋮$	$⋮$	$⋮$	$⋮$	$⋮$	$⋮$	$⋮$
2024/3/4 18:36	797.8	131.2	3.258	3.258	$⋯$	3.267
2024/3/4 18:36	787.6	131.3	3.227	3.225	$⋯$	3.23
$⋮$	$⋮$	$⋮$	$⋮$	$⋮$	$⋮$	$⋮$

表2

图7

图8

表3

图9

图10

参考文献 41

[1]	李耀华, 孔力. 发展太阳能和风能发电技术加速推进我国能源转型[J]. 中国科学院院刊, 2019, 34(4): 426-433. DOI: 10.16418/j.issn. 1000-3045.2019.04.007.
	LI Y H, KONG L. Developing solar and wind power generation technology to accelerate China's energy transformation[J]. Bulletin of Chinese Academy of Sciences, 2019, 34(4): 426-433. DOI: 10.16418/j.issn.1000-3045.2019.04.007.
[2]	缪平, 姚祯, LEMMON John, 等. 电池储能技术研究进展及展望[J]. 储能科学与技术, 2020, 9(3): 670-678. DOI: 10.19799/j.cnki.2095-4239.2020.0059.
	MIAO P, YAO Z, LEMMON John, et al. Current situations and prospects of energy storage batteries[J]. Energy Storage Science and Technology, 2020, 9(3): 670-678. DOI: 10.19799/j.cnki.2095-4239.2020.0059.
[3]	MAO N, ZHANG T, WANG Z R, et al. A systematic investigation of internal physical and chemical changes of lithium-ion batteries during overcharge[J]. Journal of Power Sources, 2022, 518: 230767. DOI: 10.1016/j.jpowsour.2021.230767.
[4]	GAN N F, SUN Z Y, ZHANG Z S, et al. Data-driven fault diagnosis of lithium-ion battery overdischarge in electric vehicles[J]. IEEE Transactions on Power Electronics, 2022, 37(4): 4575-4588. DOI: 10.1109/TPEL.2021.3121701.
[5]	HABIB A K M A, HASAN M K, ISSA G F, et al. Lithium-ion battery management system for electric vehicles: Constraints, challenges, and recommendations[J]. Batteries, 2023, 9(3): 152. DOI: 10.3390/batteries9030152.
[6]	孟国栋, 李雨珮, 唐佳, 等. 锂离子电池储能电站的热失控状态检测与安全防控技术研究进展[J]. 高电压技术, 2024, 50(7): 3105-3127. DOI: 10.13336/j.1003-6520.hve.20232357.
	MENG G D, LI Y P, TANG J, et al. Research progress of thermal runaway detection and safety control technology for lithium-ion battery energy storage power stations[J]. High Voltage Engineering, 2024, 50(7): 3105-3127. DOI: 10.13336/j.1003-6520. hve.20232357.
[7]	李首顶, 李艳, 田杰, 等. 锂离子电池电力储能系统消防安全现状分析[J]. 储能科学与技术, 2020, 9(5): 1505-1516. DOI: 10.19799/j.cnki.2095-4239.2020.0111.
	LI S D, LI Y, TIAN J, et al. Current status and emerging trends in the safety of Li-ion battery energy storage for power grid applications[J]. Energy Storage Science and Technology, 2020, 9(5): 1505-1516. DOI: 10.19799/j.cnki.2095-4239.2020.0111.
[8]	BRAVO DIAZ L, HE X Z, HU Z W, et al. Review- Meta-review of fire safety of lithium-ion batteries: Industry challenges and research contributions[J]. Journal of the Electrochemical Society, 2020, 167(9): 090559. DOI: 10.1149/1945-7111/aba8b9.
[9]	CHEN Y Q, KANG Y Q, ZHAO Y, et al. A review of lithium-ion battery safety concerns: The issues, strategies, and testing standards[J]. Journal of Energy Chemistry, 2021, 59: 83-99. DOI: 10.1016/j.jechem.2020.10.017.
[10]	XUE Q, LI G, ZHANG Y J, et al. Fault diagnosis and abnormality detection of lithium-ion battery packs based on statistical distribution[J]. Journal of Power Sources, 2021, 482: 228964. DOI: 10.1016/j.jpowsour.2020.228964.
[11]	WU X G, WEI Z X, WEN T, et al. Research on short-circuit fault-diagnosis strategy of lithium-ion battery in an energy-storage system based on voltage cosine similarity[J]. Journal of Energy Storage, 2023, 71: 108012. DOI: 10.1016/j.est.2023.108012.
[12]	BHASKAR K, KUMAR A, BUNCE J, et al. Data-driven thermal anomaly detection in large battery packs[J]. Batteries, 2023, 9(2): 70. DOI: 10.3390/batteries9020070.
[13]	SHANG Y L, LU G P, KANG Y Z, et al. A multi-fault diagnosis method based on modified Sample Entropy for lithium-ion battery strings[J]. Journal of Power Sources, 2020, 446: 227275. DOI: 10.1016/j.jpowsour.2019.227275.
[14]	WANG Z P, HONG J C, LIU P, et al. Voltage fault diagnosis and prognosis of battery systems based on entropy and Z-score for electric vehicles[J]. Applied Energy, 2017, 196: 289-302. DOI: 10.1016/j.apenergy.2016.12.143.
[15]	郭铁峰, 贺建军, 申帅, 等. 基于动态规整与改进变分自编码器的异常电池在线检测方法[J]. 电子与信息学报, 2024, 46(2): 738-747.
	GUO T F, HE J J, SHEN S, et al. Abnormal battery on-line detection method based on dynamic time warping and improved variational auto-encoder[J]. Journal of Electronics & Information Technology, 2024, 46(2): 738-747.
[16]	ZHAO Y, LIU P, WANG Z P, et al. Fault and defect diagnosis of battery for electric vehicles based on big data analysis methods[J]. Applied Energy, 2017, 207: 354-362. DOI: 10.1016/j.apenergy. 2017.05.139.
[17]	ZHANG J Z, WANG Y N, JIANG B B, et al. Realistic fault detection of li-ion battery via dynamical deep learning[J]. Nature Communications, 2023, 14: 5940. DOI: 10.1038/s41467-023-41226-5.
[18]	CHAE S G, BAE S J. An adaptive mixture prior in Bayesian convolutional autoencoder for early detecting anomalous degradation behaviors in lithium-ion batteries[J]. Reliability Engineering & System Safety, 2025, 259: 110926. DOI: 10.1016/j.ress.2025.110926.
[19]	CHEN J C. Degradation self-supervised learning for lithium-ion battery health diagnostics[EB/OL]. [2025-04-23]. https://arxiv.org/abs/2503.08083.
[20]	GHANIM J, AWAD M. An unsupervised anomaly detection in electricity consumption using reinforcement learning and time series forest based framework[J]. Journal of Artificial Intelligence and Soft Computing Research, 2025, 15(1): 5-24. DOI: 10.2478/jaiscr-2025-0001.
[21]	SONG Y, YU J S, ZHOU J H, et al. Detection of voltage fault in lithium-ion battery based on equivalent circuit model-informed neural network[J]. IEEE Transactions on Instrumentation and Measurement, 2024, 73: 3508010. DOI: 10.1109/TIM.2024. 3350153.
[22]	王昌松, 陈辉, 王立成, 等. 考虑测量野值的锂电池异常数据检测与SOC估计算法[J]. 电子科技, 2023, 36(5): 34-40. DOI: 10.16180/j.cnki.issn1007-7820.2023.05.006.
	WANG C S, CHEN H, WANG L C, et al. Abnormal data detection and SOC estimation algorithm for lithium battery considering measured outliers[J]. Electronic Science and Technology, 2023, 36(5): 34-40. DOI: 10.16180/j.cnki.issn1007-7820.2023.05.006.
[23]	XU Y M, GE X H, SHEN W X. Multi-objective nonlinear observer design for multi-fault detection of lithium-ion battery in electric vehicles[J]. Applied Energy, 2024, 362: 122989. DOI: 10.1016/j.apenergy.2024.122989.
[24]	MCKAY M B, GOPALUNI B, WETTON B. Learning the P2D model for lithium-ion batteries with SOH detection[EB/OL]. [2025-04-23]. https://arxiv.org/abs/2502.14147.
[25]	RAHMAN M A, ANWAR S, IZADIAN A. Electrochemical model based fault diagnosis of a lithium ion battery using multiple model adaptive estimation approach[C]//2015 IEEE International Conference on Industrial Technology (ICIT). March 17-19, 2015, Seville, Spain. IEEE, 2015: 210-217. DOI: 10.1109/ICIT.2015. 7125101.
[26]	MARTIN J, MONSALVE-SERRANO J, MICÓ C, et al. Thermal model for the analysis of the thermal runaway in lithium-ion batteries using accelerating rate calorimetry[C]//SAE Technical Paper Series. Capri, Italy. SAE International, 2023: DOI: 10.4271/2023-24-0162.
[27]	WEI J W, DONG G Z, CHEN Z H. Lyapunov-based thermal fault diagnosis of cylindrical lithium-ion batteries[J]. IEEE Transactions on Industrial Electronics, 2020, 67(6): 4670-4679. DOI: 10.1109/TIE.2019.2931275.
[28]	王莉, 谢乐琼, 张干, 等. 锂离子电池一致性筛选研究进展[J]. 储能科学与技术, 2018, 7(2): 194-202.
	WANG L, XIE L Q, ZHANG G, et al. Research progress in the consistency screening of Li-ion batteries[J]. Energy Storage Science and Technology, 2018, 7(2): 194-202.
[29]	BOUKELA L, ZHANG G X, YACOUB M, et al. A modified LOF-based approach for outlier characterization in IoT[J]. Annals of Telecommunications, 2021, 76(3): 145-153. DOI: 10.1007/s12243- 020-00780-5.
[30]	李晓宇, 陈炫锴, 李嘉栩, 等. 基于机器学习的电动汽车续航里程预测[J]. 电器与能效管理技术, 2021(10): 78-82, 91. DOI: 10.16628/j.cnki.2095-8188.2021.10.013.
	LI X Y, CHEN X K, LI J X, et al. Driving range prediction of electric vehicles based on machine learning[J]. Electrical & Energy Management Technology, 2021(10): 78-82, 91. DOI: 10. 16628/j.cnki.2095-8188.2021.10.013.
[31]	刘启全, 马建, 赵轩, 等. 基于值率模型的电动汽车动力电池电压异常检测[J]. 汽车工程, 2023, 45(9): 1728-1739. DOI: 10.19562/j.chinasae.qcgc.2023.09.021.
	LIU Q Q, MA J, ZHAO X, et al. Abnormal voltage detection of battery for electric vehicles based on value rate model[J]. Automotive Engineering, 2023, 45(9): 1728-1739. DOI: 10.19562/j.chinasae.qcgc.2023.09.021.
[32]	CARDOSO D O, GALENO T D. Online evaluation of the Kolmogorov-Smirnov test on arbitrarily large samples[J]. Journal of Computational Science, 2023, 67: 101959. DOI: 10.1016/j.jocs.2023.101959.
[33]	LI M, CHENG F Y, YANG J, et al. A fault detection method for electric vehicle battery system based on Bayesian optimization SVDD considering a few faulty samples[J]. Energy Engineering, 2024, 121(9): 2543-2568. DOI: 10.32604/ee.2024.051231.
[34]	RUIZ-RODRIGUEZ F J, HERNÁNDEZ J C, JURADO F. Probabilistic load flow for photovoltaic distributed generation using the Cornish-Fisher expansion[J]. Electric Power Systems Research, 2012, 89: 129-138. DOI: 10.1016/j.epsr.2012.03.009.
[35]	HASSAN A F, BARAKAT S, REZK A. Towards a deep learning-based outlier detection approach in the context of streaming data[J]. Journal of Big Data, 2022, 9(1): 120. DOI: 10.1186/s40537-022-00670-8.
[36]	BREUNIG M M, KRIEGEL H P, NG R T, et al. LOF: Identifying density-based local outliers[J]. SIGMOD 2000-Proceedings of the 2000 ACM SIGMOD International Conference on Management of Data, 2000: 93-104
[37]	ALGHUSHAIRY O, ALSINI R, SOULE T, et al. A review of local outlier factor algorithms for outlier detection in big data streams[J]. Big Data and Cognitive Computing, 2021, 5(1): 1. DOI: 10. 3390/bdcc5010001.
[38]	WEN S Q, ZHANG W R, SUN Y F, et al. An enhanced principal component analysis method with Savitzky-Golay filter and clustering algorithm for sensor fault detection and diagnosis[J]. Applied Energy, 2023, 337: 120862. DOI: 10.1016/j.apenergy. 2023.120862.
[39]	KRISHNAN S R, SEELAMANTULA C S. On the selection of optimum savitzky-golay filters[J]. IEEE Transactions on Signal Processing, 2013, 61(2): 380-391. DOI: 10.1109/TSP. 2012. 2225055.
[40]	STALLONE A, CICONE A, MATERASSI M. New insights and best practices for the successful use of Empirical Mode Decomposition, Iterative Filtering and derived algorithms[J]. Scientific Reports, 2020, 10: 15161. DOI: 10.1038/s41598-020-72193-2.
[41]	曾进辉, 苏旨音, 肖锋, 等. 基于经验模态分解和ISSA-LSTM的短期电力负荷预测[J/OL]. 电子测量技术, 2024: 1-7. (2024-10-21). https://kns.cnki.net/kcms/detail/11.2175.TN.20241018.1837.033.html.
	ZENG J H, SU Z Y, XIAO F, et al. Short-term power load forecasting based on empirical mode decomposition and ISSA-LSTM[J/OL]. Electronic Measurement Technology, 2024: 1-7. (2024-10-21). https://kns.cnki.net/kcms/detail/11.2175.TN.2024 1018.1837.033.html.

项目	KS统计值	p值
正态分布	0.034	0.189
拉普拉斯分布	0.058	0.002

方法		是否预训练	检测异常点	异常检出率	检测准确率	检测时间/s
本工作所提出的检测框架	K-means聚类	N	[364]	0.5	99.76%	0.0317
	隔离森林	N	[130][155][182][246] [386][387][388][389]	0.5	98.08%	0.1099
	香农熵	N	[155]	0.5	99.76%	0.0418
	余弦相似度	N	[155][364]	1.0	100%	0.5971
	自编码器	Y	[246][386][388][389]	0	98.56%	2.9411
	LOF	N	[155][364]	1.0	100%	0.0106

PCA+CUSUM		Y	[155][35][241]	0.5	99.28%	2.8286
TSF+DQN		Y	共检测到16个：异常分数前3[155][286] [338](无[364])	0.5	96.15%	5.0936
			共检测到16个：异常分数前3[155][286] [338](无[364])