基于改进TCN模型的动力电池健康状态评估

doi:10.19799/j.cnki.2095-4239.2021.0637

摘要/Abstract

摘要：

电动汽车动力电池的健康状态(state of heath，SOH)是电池管理系统重点监测指标之一，对其进行精确评估对于整车安全可靠运行具有重要意义。但现有动力电池SOH评估方法存在评估精度不理想、计算复杂度高等问题，为此提出一种基于改进时间卷积网络(Temporal convolutional network，TCN)模型的动力电池SOH评估方法。该方法首先从动力电池充电数据中提取等电压上升时间、等电流下降时间、电压回升数值三个健康因子，并采用皮尔森相关系数验证了其与电池容量之间的相关关系；然后根据TCN模型的感受野对编码器-解码器结构进行调整，通过训练后的编码器对输入序列进行特征提取，获取到长时间序列的短时表达；最后利用TCN模型捕获特征时间序列与电池SOH之间的因果关系，实现对动力电池SOH的精确评估。所提方法应用于公开电池数据集，并将实验结果与当前广泛采用的时间序列分析方法及相关文献中所用方法进行对比，结果表明所提方法可使均方根误差和绝对平均误差分别降低0.0012和0.0008，相同训练次数所需计算时间减少3.4%。

关键词: 动力电池, 健康状态, 编码器, TCN

Abstract:

The state of heath (SOH) of the power battery of an electric vehicle is one of the key monitoring indicators of its battery management system. Accurate evaluation of it is of great significance for the safe and reliable operation of the vehicle. However, existing power battery SOH evaluation methods have problems such as unsatisfactory evaluation accuracy and high computational complexity. For this reason, a power battery health status evaluation method based on an improved Temporal Convolutional Network (TCN) model is proposed. This method first extracts three health factors of equal voltage rise time, equal current fall time, and voltage rise value from the power battery charging data, and uses Pearson correlation coefficient to verify the correlation between them and the battery capacity; then according to the TCN model The receptive field adjusts the encoder-decoder structure, and uses the trained encoder to extract features of the input sequence to obtain the short-term expression of the long-term sequence; finally, the TCN model is used to capture the characteristic time sequence and the battery SOH. Causality, to achieve accurate assessment of the health status of the power battery. The proposed method is applied to the public battery data set, and the experimental results are compared with the widely used time series analysis methods and methods used in related literature. The results show that the proposed method can reduce the root mean square error and absolute average error by 0.0012 respectively. And 0.0008, the calculation time required for the same training times is reduced by 3.4%.

Key words: power battery, health status, encoder, TCN

中图分类号:

TQ 912

张孝远, 张金浩, 蒋亚俊. 基于改进TCN模型的动力电池健康状态评估[J]. 储能科学与技术, 2022, 11(5): 1617-1626.

Xiaoyuan ZHANG, Jinhao ZHANG, Yajun JIANG. Power battery health evaluation based on improved TCN model[J]. Energy Storage Science and Technology, 2022, 11(5): 1617-1626.

图/表 18

图1

图2

图3

图4

图5

图6

表1

图7

图8

表2

图9

图10

图11

图12

图13

表3

图14

表4

参考文献 21

1	王震坡, 李晓宇, 袁昌贵, 等. 大数据下电动汽车动力电池故障诊断技术挑战与发展趋势[J]. 机械工程学报, 2021, 57(14): 52-63.
	WANG Z P, LI X Y, YUAN C G, et al. Challenge and prospects for fault diagnosis of power battery system for electrical vehicles based on big-data[J]. Journal of Mechanical Engineering, 2021, 57(14): 52-63.
2	严干贵, 李洪波, 段双明, 等. 基于模型参数辨识的储能电池状态估算[J]. 中国电机工程学报, 2020, 40(24): 8145-8154, 8251.
	YAN G G, LI H B, DUAN S M, et al. Energy storage battery state estimation based on model parameter identification[J]. Proceedings of the CSEE, 2020, 40(24): 8145-8154, 8251.
3	李相俊, 马锐. 考虑电池组健康状态的储能系统能量管理方法[J]. 电网技术, 2020, 44(11): 4210-4217.
	LI X J, MA R. Energy management method of energy storage system considering the SOH of battery pack[J]. Power System Technology, 2020, 44(11): 4210-4217.
4	王萍, 张吉昂, 程泽. 基于最小二乘支持向量机误差补偿模型的锂离子电池健康状态估计方法[J]. 电网技术, 2022, 46(2): 613-623.
	WANG P, ZHANG J, CHENG Z. State of health estimation of Li-ion battery based on least squares support vector machine error compensation model[J]. Power System Technology, 2022, 46(2): 613-623.
5	韦海燕, 陈孝杰, 吕治强, 等. 灰色神经网络模型在线估算锂离子电池SOH[J]. 电网技术, 2017, 41(12): 4038-4044.
	WEI H Y, CHEN X J, LÜ Z Q, et al. Online estimation of lithium-ion battery state of health using grey neural network[J]. Power System Technology, 2017, 41(12): 4038-4044.
6	耿萌萌, 范茂松, 杨凯, 等. 基于EIS和神经网络的退役电池SOH快速估计[J]. 储能科学与技术, 2022, 11(2): 673-678.
	GENG M M, FAN M S, YANG K, et al. Fast estimation method for state-of-health of retired batteries based on electrochemical impedance spectroscopy and neural network[J]. Energy Storage Science and Technology, 2022, 11(2): 673-678.
7	王凡, 史永胜, 刘博亲, 等. 基于注意力改进BiGRU的锂离子电池健康状态估计[J]. 储能科学与技术, 2021, 10(6): 2326-2333.
	WANG F, SHI Y S, LIU B Q, et al. Health state estimation of lithium-ion batteries based on attention augmented BiGRU[J]. Energy Storage Science and Technology, 2021, 10(6): 2326-2333.
8	王琪, 孙玉坤, 倪福银, 等. 混合动力汽车电池内部状态预测的贝叶斯极限学习机方法[J]. 中国机械工程, 2016, 27(22): 3118-3123.
	WANG Q, SUN Y K, NI F Y, et al. Prediction of internal states of battery in HEV by BELM[J]. China Mechanical Engineering, 2016, 27(22): 3118-3123.
9	焦自权, 范兴明, 张鑫, 等. 基于改进粒子滤波算法的锂离子电池状态跟踪与剩余使用寿命预测方法[J]. 电工技术学报, 2020, 35(18): 3979-3993.
	JIAO Z Q, FAN X M, ZHANG X, et al. State tracking and remaining useful life predictive method of Li-ion battery based on improved particle filter algorithm[J]. Transactions of China Electrotechnical Society, 2020, 35(18): 3979-3993.
10	DONG H C, JIN X N, LOU Y B, et al. Lithium-ion battery state of health monitoring and remaining useful life prediction based on support vector regression-particle filter[J]. Journal of Power Sources, 2014, 271: 114-123.
11	ZAREMBA W, SUTSKEVER I, VINYALS O. Recurrent neural network regularization[EB/OL]. 2014: arXiv: 1409.2329[cs.NE]. https://arxiv.org/abs/1409.2329.
12	张少宇, 伍春晖, 熊文渊. 采用门控循环神经网络估计锂离子电池健康状态[J]. 红外与激光工程, 2021, 50(2): 3788/IRLA20200339.
	ZHANG S Y, WU C H, XIONG W Y. State of health estimation for lithium-ion batteries using recurrent neural networks with gated recurrent unit[J]. Infrared and Laser Engineering, 2021, 50(2): 3788/IRLA20200339.
13	张岸, 杨春德. 基于GAN-CNN-LSTM的锂电池SOH估计[J]. 电源技术, 2021, 45(7): 902-906.
	ZHANG A, YANG C D. SOH estimation of lithium batteries based on GAN-CNN-LSTM[J]. Chinese Journal of Power Sources, 2021, 45(7): 902-906.
14	李超然, 肖飞, 樊亚翔, 等. 基于深度学习的锂离子电池SOC和SOH联合估算[J]. 中国电机工程学报, 2021, 41(2): 681-692.
	LI C R, XIAO F, FAN Y X, et al. Joint estimation of the state of charge and the state of health based on deep learning for lithium-ion batteries[J]. Proceedings of the CSEE, 2021, 41(2): 681-692.
15	LI P H, ZHANG Z J, XIONG Q Y, et al. State-of-health estimation and remaining useful life prediction for the lithium-ion battery based on a variant long short term memory neural network[J]. Journal of Power Sources, 2020, 459: 228069.
16	QU J T, LIU F, MA Y X, et al. A neural-network-based method for RUL prediction and SOH monitoring of lithium-ion battery[J]. IEEE Access, 2019, 7: 87178-87191.
17	刘伟霞, 田勋, 肖家勇, 等. 基于混合模型及LSTM的锂电池SOH与剩余寿命预测[J]. 储能科学与技术, 2021, 10(2): 689-694.
	LIU W X, TIAN X, XIAO J Y, et al. Estimation of SOH and remaining life of lithium batteries based on a combination model and long short-term memory[J]. Energy Storage Science and Technology, 2021, 10(2): 689-694.
18	NAUTA M, BUCUR D, SEIFERT C. Causal discovery with attention-based convolutional neural networks[J]. Machine Learning and Knowledge Extraction, 2019, 1(1): 312-340.
19	CHUNG J, GULCEHRE C, CHO K, et al. Empirical evaluation of gated recurrent neural networks on sequence modeling[EB/OL]. 2014: arXiv: 1412.3555[cs.NE]. https://arxiv.org/abs/1412.3555
20	RUMELHART D E, HINTON G E, WILLIAMS R J. Learning representations by back-propagating errors[J]. Nature, 1986, 323(6088): 533-536.
21	HINTON G E, SALAKHUTDINOV R R. Reducing the dimensionality of data with neural networks[J]. Science, 2006, 313(5786): 504-507.

电池序号	容量/Ah	放电电流	总循环次数	有效循环次数
CS33	1.1	0.5 C/0.55 A	632	592
CS34	1.1	0.5 C/0.55 A	700	621
CS35	1.1	1 C/1.1 A	716	640
CS36	1.1	1 C/1.1 A	791	636

电池序号	DV_UT	DA_DT	DV_RU
CS33	0.987768	-0.843369	-0.968624
CS34	0.993926	-0.876500	-0.966657
CS35	0.990361	-0.870651	-0.592425
CS36	0.993409	-0.904253	-0.938893

评价指标	电池序号	Encorder-TCN	TCN	GRU	LSTM
MSE/%	CS33	0.0167738	0.0412891	0.1702569	0.0437183
	CS34	0.0256044	0.0277419	0.0290808	0.2298263
	CS35	0.0157313	0.0218908	0.0389273	0.0289350
	CS36	0.0096664	0.0133836	0.2969346	0.0362662
RMSE/%	CS33	1.2951373	2.0319711	4.1262198	2.0908928
	CS34	1.6001361	1.6655911	1.7053086	4.7940201
	CS35	1.2542454	1.4795535	1.9729991	1.7010299
	CS36	0.9831776	1.1568763	5.4491701	1.9043677
MAE/%	CS33	1.0180461	1.7207226	2.7546348	1.6967513
	CS34	0.9839263	1.3964472	1.2849554	3.2334339
	CS35	0.9726970	1.1667483	1.5380482	1.3619463
	CS36	0.7271382	0.8386590	4.3778951	1.3251187

Item	Encorder-TCN	TCN	GRU	LSTM
MSE/%	0.0036	0.0090	0.0366	0.0887
RMSE/%	0.5973	0.9504	1.9127	2.9776
MAE/%	0.3650	0.8336	1.7663	2.6325
平均运行时间/s	24.1708	23.0095	25.0214	25.4493

[1]	常泽宇, 张之琦, 张晓东, 李丽, 郁亚娟. 基于数据驱动的动力电池健康状态评估平台[J]. 储能科学与技术, 2022, 11(6): 1847-1853.
[2]	周伟, 符冬菊, 刘伟峰, 陈建军, 胡照, 曾燮榕. 废旧磷酸铁锂动力电池回收利用研究进展[J]. 储能科学与技术, 2022, 11(6): 1854-1864.
[3]	王苏杭, 李建林, 李雅欣, 熊俊杰, 曾伟. 锂离子电池系统低温充电策略[J]. 储能科学与技术, 2022, 11(5): 1537-1542.
[4]	戴彦文, 于艾清. 基于健康特征参数的CNN-LSTM&GRU组合锂电池SOH估计[J]. 储能科学与技术, 2022, 11(5): 1641-1649.
[5]	王芳, 王峥, 林春景, 张国振, 张贵平, 马天翼, 刘磊, 刘仕强. 新能源汽车动力电池安全失效潜在原因分析[J]. 储能科学与技术, 2022, 11(5): 1411-1418.
[6]	刘春辉, 任宏斌. 基于SOC的动力电池组主动均衡研究[J]. 储能科学与技术, 2022, 11(2): 667-672.
[7]	耿萌萌, 范茂松, 杨凯, 赵光金, 谭震, 高飞, 张明杰. 基于EIS和神经网络的退役电池SOH快速估计[J]. 储能科学与技术, 2022, 11(2): 673-678.
[8]	马天翼, 王伟哲, 林春景, 白广利, 韦振, 刘仕强. 磷酸铁锂电池多阶段利用可行性[J]. 储能科学与技术, 2022, 11(1): 119-126.
[9]	胡建, 林春景, 郝维健, 郑天雷. 动力电池标准体系建设现状及建议[J]. 储能科学与技术, 2022, 11(1): 313-320.
[10]	王凡, 史永胜, 刘博亲, 左玉洁, 符政, ALI Jamsher. 基于注意力改进BiGRU的锂离子电池健康状态估计[J]. 储能科学与技术, 2021, 10(6): 2326-2333.
[11]	郑永强, 吴越, 张盼盼, 雷博, 郑耀东. 基于多分支拓扑的梯次利用储能系统电池同期退役协同控制策略[J]. 储能科学与技术, 2021, 10(6): 2283-2292.
[12]	刘霏霏, 鲍荣清, 程贤福, 李骏, 秦武, 杨超峰. 服役工况下车用锂离子动力电池散热方法综述[J]. 储能科学与技术, 2021, 10(6): 2269-2282.
[13]	冯一峰, 沈佳妮, 车海英, 马紫峰, 贺益君, 谈文, 杨庆亨. 钠离子电池健康状态预测[J]. 储能科学与技术, 2021, 10(4): 1407-1415.
[14]	任璞, 王顺利, 何明芳, 范永存, 曹文, 谢伟. 基于内阻增加和容量衰减双重标定的锂电池健康状态评估[J]. 储能科学与技术, 2021, 10(2): 738-743.
[15]	李鹏, 李立伟, 杨玉新. 基于IBOA-PF的锂电池健康状态预测[J]. 储能科学与技术, 2021, 10(2): 705-713.