数据分布多样性对锂电池SOC预测的泛化影响

doi:10.19799/j.cnki.2095-4239.2024.0003

摘要/Abstract

摘要：

数据驱动模型预测荷电状态(SOC)依赖高质量的实验数据，在应用于实际使用场景下的分布多样的锂电池组数据时会出现预测的准确性不稳定即泛化能力差的情况，限制了模型的实际应用。研究实际场景下的大规模数据的分布多样性对SOC预测模型的泛化性影响具有重要意义。因此，对32个锂电池组的实际运行数据集进行研究，采用经典算法与多输入多输出(MIMO)策略结合来预测多步SOC，对每份数据分别建立模型进行SOC预测，研究了不同算法的应用效果并分析了数据分布多样性对模型的泛化能力的影响规律。结果表明：对大规模的锂电池组数据，LR-MIMO模型训练精度普遍优于RF-MIMO、KNN-MIMO、LSTM-MIMO模型，其预测未来0.5 h的SOC的R²一般在0.98及以上，MAPE基本低于0.05。与其他模型相比，LR-MIMO模型有优秀的预测性能，预测其他数据集的R²基本在0.95以上。而KNN-MIMO模型的预测精度与RF-MIMO模型相当，R²大致在0.7以上，LSTM-MIMO模型的预测性能因数据集不同存在较明显的差异；当数据满足SOC与电压的相关系数≥0.9、SOC和电压分布范围广、核密度曲线呈左偏趋势、分布较均匀时，可使模型训练精度提高。

关键词: 锂离子电池, 荷电状态, 数据驱动, 分布多样性, 泛化性

Abstract:

The prediction of state-of-charge (SOC) in batteries in data-driven models relies on high-quality experimental data. However, when considering lithium battery packs with diverse distributions in real-world scenarios, the accuracy of prediction of SOC in these packs becomes unstable, leading to poor generalization ability and limiting the practical application of the models. Therefore, it is of great significance to investigate the impact of the diverse distribution of large-scale datasets from actual operating scenarios on the generalization ability of SOC prediction models. Consequently, in this study, we conducted research on 32 datasets of operational data associated with lithium battery packs. Classic algorithms were combined with a multi-input multi-output (MIMO) strategy to predict multi-step SOC. Separate models were established for each dataset to analyze the application effects of different algorithms and the influence of data distribution diversity on model generalization ability. The results demonstrated that for large-scale lithium battery pack datasets, the LR-MIMO model generally exhibited higher training accuracy compared to the RF-MIMO, KNN-MIMO, and LSTM-MIMO models. The LR-MIMO model achieved R² values above 0.98 and MAPE values below 0.05 for predicting SOC in the next half hour. Compared to other models, the LR-MIMO model demonstrated excellent predictive performance, with R² values above 0.95 for predicting other datasets. The prediction accuracy of the KNN-MIMO model is comparable to that of the RF-MIMO model, with R² values roughly above 0.7, while the prediction performance of the LSTM-MIMO model differs significantly compared to other models due to the use of different datasets. The accuracy of the model can be improved when the data satisfies certain conditions, such as a correlation coefficient between SOC and voltage exceeding 0.9, a wide range of SOC and voltage distributions, a left-skewed kernel density curve, and a relatively uniform distribution.

Key words: lithium-ion battery, state of charge, data-driven, distribution diversity, generalization

中图分类号:

TM 912

何林, 刘江岩, 刘彬, 李夔宁, 代帅. 数据分布多样性对锂电池SOC预测的泛化影响[J]. 储能科学与技术, 2024, 13(5): 1677-1687.

Lin HE, Jiangyan LIU, Bin LIU, Kuining LI, Shuai DAI. Generalized impact of data distribution diversity on SOC prediction of lithium battery[J]. Energy Storage Science and Technology, 2024, 13(5): 1677-1687.

图/表 11

图1

图2

图3

图4

表1

图5

图6

图7

图8

图9

图10

参考文献 33

1	YANG R X, XIONG R, WEIXIANG S, et al. Extreme learning machine-based thermal model for lithium-ion batteries of electric vehicles under external short circuit[J]. Engineering, 2021, 7(3): 266-289.
2	XIONG R, MA S X, LI H L, et al. Toward a safer battery management system: A critical review on diagnosis and prognosis of battery short circuit[J]. iScience, 2020, 23(4): 101010.
3	CAI Z H, LIU G F, LUO J. Research state of charge estimation tactics of nickel-hydrogen battery[C]//2010 International Symposium on Intelligence Information Processing and Trusted Computing. Huanggang, China. IEEE, 2010: 184-187.
4	XIONG R, CAO J Y, YU Q Q, et al. Critical review on the battery state of charge estimation methods for electric vehicles[J]. IEEE Access, 2018, 6: 1832-1843.
5	BIAN C, HE H, YANG S, et al. State-of-charge sequence estimation of lithium-ion battery based on bidirectional long short-term memory encoder-decoder architecture[J]. Journal of Power Sources, 2020, 449: 227558.
6	CHEMALI E, KOLLMEYER P J, PREINDL M, et al. State-of-charge estimation of Li-ion batteries using deep neural networks: A machine learning approach[J]. Journal of Power Sources, 2018, 400: 242-255.
7	TIAN Y, LAI R, LI X, et al.. A combined method for state-of-charge estimation for lithium-ion batteries using a long short-term memory network and an adaptive cubature Kalman filter[J]. Applied Energy, 2020, 265: 114789.
8	BABAEIYAZDI I, REZAEI-ZARE A, SHOKRZADEH S. State of charge prediction of EV Li-ion batteries using EIS: A machine learning approach[J]. Energy, 2021, 223: 120116.
9	LI Y, ZOU C, BERECIBAR M, et al. Random forest regression for online capacity estimation of lithium-ion batteries[J]. Applied Energy, 2018, 232: 197-210.
10	MENG J H, STROE D I, RICCO M, et al. A simplified model-based state-of-charge estimation approach for lithium-ion battery with dynamic linear model[J]. IEEE Transactions on Industrial Electronics, 2019, 66(10): 7717-7727.
11	HONG J, WANG Z, CHEN W, et al. Online joint-prediction of multi-forward-step battery SOC using LSTM neural networks and multiple linear regression for real-world electric vehicles[J]. Journal of Energy Storage, 2020, 30: 101459.
12	WANG G, LYU Z, LI X. An optimized random forest regression model for Li–ion battery prognostics and health management[J]. Batteries, 2023, 9(6): 332.
13	WANG X, HU B, SU X, et al. State of health estimation for lithium-ion batteries using random forest and gated recurrent unit[J]. Journal of Energy Storage, 2024, 76: 109796.
14	YANG N K, SONG Z Y, HOFMANN H, et al. Robust state of health estimation of lithium-ion batteries using convolutional neural network and random forest[EB/OL]. 2020: arXiv: 2010.10452. http://arxiv.org/abs/2010.10452
15	HU C, JAIN G, ZHANG P Q, et al. Data-driven method based on particle swarm optimization and k-nearest neighbor regression for estimating capacity of lithium-ion battery[J]. Applied Energy, 2014, 129: 49-55.
16	ZHOU Y, HUANG M, PECHT M, et al. Remaining useful life estimation of lithium-ion cells based on knearest neighbor regression with differential evolution optimization[J]. Journal of Cleaner Production, 2020, 249: 119409.
17	CHEN J X, ZHANG Y, WU J, et al. SOC estimation for lithium-ion battery using the LSTM-RNN with extended input and constrained output[J]. Energy, 2023, 262: 125375.
18	LI W, SENGUPTA N, DECHENT P, et al. Online capacity estimation of lithium-ion batteries with deep long short-term memory networks[J]. Journal of Power Sources, 2021, 482: 228863.
19	VIDAL C, KOLLMEYER P, CHEMALI E, et al. Li-ion battery state of charge estimation using long short-term memory recurrent neural network with transfer learning[C]//2019 IEEE Transportation Electrification Conference and Expo (ITEC). Detroit, MI, USA. IEEE, 2019: 1-6.
20	LIU Y, SHU X, YU H Z, et al. State of charge prediction framework for lithium-ion batteries incorporating long short-term memory network and transfer learning[J]. Journal of Energy Storage, 2021, 37: 102494.
21	WANG Y, CHEN Z, ZHANG W. Lithium-ion battery state-of-charge estimation for small target sample sets using the improved GRU-based transfer learning[J]. Energy, 2022, 244: 12317.
22	Pro machine learning algorithms[M]. Apress, Berkeley, CA.
23	TORGO L. Chapman & Hall/CRC data mining and knowledge discovery series[M]. Data Mining with R Volume, 2010.
24	BREIMAN L, CUTLER A. Random forests[J]. Machine Learning, 2001, 45(1): 5-32.
25	BREIMAN L. Bagging predictors[J]. Machine Learning, 1996, 24(2): 123-140.
26	HO T K. The random subspace method for constructing decision forests[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1998, 20(8): 832-844.
27	BREIMAN L, FREIDMANJ H, OLSHEN R A, et al. Classification and regression trees[M]. Chapman & Hau/CRC, 1984.
28	HOCHREITER S, SCHMIDHUBER J. Long short-term memory[J]. Neural Computation, 1997, 9(8): 1735-1780.
29	DR K S, DR C B. Data analytics: Why data normalization[J]. International Journal of Engineering & Technology, 2018, 7(4.6): 209.
30	MURRAY R. Remarks on some nonparametric estimates of a density function[J]. The Annals of Mathematical Statistics, 1956, 27(3): 832-837.
31	PARZEN E. On estimation of a probability density function and mode[J]. Annals of Mathematical Statistics, 1962, 33: 1065-1076.
32	EPANECHNIKOV V A.Nonparametric estimation of a multidimensional proability density[J]. Theory of Proability and Application,1969, 14: 153-158.
33	SILVERMAN B W. Density estimation for statistics and data analysis[M]. London: Chapman and Hall, 1986.

序号	主要参数	量值
1	单电池数量	14
2	电池连接方式	串联
3	标称电压/V	58
4	标称容量/Ah	32
5	充电方式	恒流恒压
6	电池冷却	被动式空气冷却
7	电池加热	无

[1]	唐梓巍, 师玉璞, 张雨禅, 周奕博, 杜慧玲. 基于Informer神经网络的锂离子电池容量退化轨迹预测[J]. 储能科学与技术, 2024, 13(5): 1658-1666.
[2]	缪胤宝, 张文华, 刘伟昊, 王帅, 陈哲, 彭望, 曾杰. 富锂正极材料Li_1.2Ni_0.13Co_0.13Mn_0.54O₂ 的制备及性能[J]. 储能科学与技术, 2024, 13(5): 1427-1434.
[3]	廉高棨, 叶敏, 王桥, 李岩, 麻玉川, 孙乙丁, 杜鹏辉. 基于改进模型与优化自适应CKF的锂离子电池快速变温工况下的SOC估计[J]. 储能科学与技术, 2024, 13(5): 1667-1676.
[4]	吕兆财, 王玉西, 汪智涛, 孙晓辉, 李景康. 热辊压对锂离子电池正极极片性能的影响[J]. 储能科学与技术, 2024, 13(5): 1443-1450.
[5]	李润源, 郭傅傲, 赵钢超. 集装箱式锂离子电池储能系统消防安全早期预警方法[J]. 储能科学与技术, 2024, 13(5): 1595-1602.
[6]	韩亚露, 陈奕戈, 邸会芳, 林杰欢, 王振兵, 张扬, 苏方远, 陈成猛. 锂离子电池不同服役工况下失效研究进展[J]. 储能科学与技术, 2024, 13(4): 1338-1349.
[7]	袁悦博, 王贺武, 孔祥栋, 蒲明伟, 孙玉坤, 韩雪冰, 欧阳明高. 金属异物缺陷演化特性及其对产线 K 值的影响机制[J]. 储能科学与技术, 2024, 13(4): 1197-1204.
[8]	李革, 孔祥栋, 孙跃东, 陈飞, 袁悦博, 韩雪冰, 郑岳久. 基于产线大数据的锂离子电池一致性动态特性分选方法[J]. 储能科学与技术, 2024, 13(4): 1188-1196.
[9]	刘淳正, 来沛霈, 孙卓, 聂耳, 张哲娟. 构造凹陷的硅碳颗粒提高锂离子电池负极电化学性能[J]. 储能科学与技术, 2024, 13(4): 1302-1309.
[10]	李炳金, 韩晓霞, 张文杰, 曾伟国, 武晋德. 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.
[11]	杨家沐, 陈育新, 练成, 徐至, 刘洪来. 电极浆料涂布模头流场分析与结构优化[J]. 储能科学与技术, 2024, 13(4): 1109-1117.
[12]	王玉婷, 李秋桐, 胡一鸣, 郭新. 锂离子电池内部信号监测技术概述[J]. 储能科学与技术, 2024, 13(4): 1253-1265.
[13]	张爱芳, 魏邦达, 李卓昊, 杨洋, 杨添强, 姚俊, 张杰, 刘飞, 李浩秒, 王康丽, 蒋凯. 全钒液流电池建模及SOC在线估计研究进展[J]. 储能科学与技术, 2024, 13(3): 1036-1049.
[14]	申小雨, 尹丛勃. 基于卷积Fastformer的锂离子电池健康状态估计[J]. 储能科学与技术, 2024, 13(3): 990-999.
[15]	胡大林, 任潘利, 张昌明, 杨明阳, 卢周广. Al-Y-Zr原位共掺杂提高4.53 V钴酸锂正极材料的循环性能[J]. 储能科学与技术, 2024, 13(3): 742-748.