State of health estimation for lithium batteries based on short-term random charging data and optimized convolutional neural network

doi:10.19799/j.cnki.2095-4239.2024.0964

Abstract

Abstract:

In practical applications, complete charge-discharge curves are often unavailable. To address this issue, this study proposes a lithium-ion battery state of health (SOH) estimation method using short-term stochastic charging data and an optimized convolutional neural network (CNN). The goal is to develop an efficient technique that accommodates the random and disordered nature of charging processes in real-world scenarios. The proposed method segments the original charging voltage time-series data of lithium batteries to generate randomized charging data. A shallow CNN comprising four convolutional layers is then constructed to adaptively extract aging-related features from the data. In addition, the dung beetle optimization algorithm is employed to optimize the model parameters, resulting in a multistage model. The experimental results demonstrate that the proposed method can accurately estimate the SOH of Li-ion batteries during stochastic charging, even when using only five consecutive seconds (100 data points) of raw voltage time-series data. The practicality and accuracy of the proposed model were validated under various charging conditions and rates. The results demonstrate that the model continues to exhibit a low prediction error. The average absolute error of the SOH estimation is less than 2.07% under constant current-voltage charging mode and less than 1.22% under multistage constant current charging mode. In the model comparison, the proposed CNN-based method achieved an average mean absolute error of 1.17%, outperforming integrated prediction models, such as the CNN-Long Short-Term Memory Network and CNN-Gated Recurrent Unit, in terms of estimation accuracy and stability. In addition, the randomized voltage segment used in the CNN accounted for 88.9% of the total charging time, is higher than the other two prediction models. The experimental results demonstrate the effectiveness of the proposed method in addressing the stochastic nature of battery charging data, demonstrating excellent accuracy and high adaptability in the context of charging rates, charging modes, and random charging voltage data over short periods. This results of this study provide a crucial technical reference for future advancement in battery health monitoring and battery management systems.

Key words: state of health, random charging, data segmentation, convolutional neural network, lithium-ion battery

CLC Number:

TM 911

Jiangwei SHEN, Yixin SHE, Xing SHU, Yonggang LIU, Fuxing WEI, Xuelei XIA, Zheng CHEN. State of health estimation for lithium batteries based on short-term random charging data and optimized convolutional neural network[J]. Energy Storage Science and Technology, 2025, 14(4): 1585-1595.

Figures/Tables 14

Table 1

Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Table 2

Fig. 6

Table 3

Table 4

Fig. 7

Fig. 8

Fig. 9

Table 5

References 28

1	HE Y, BAI W Y, WANG L L, et al. SOH estimation for lithium-ion batteries: An improved GPR optimization method based on the developed feature extraction[J]. Journal of Energy Storage, 2024, 83: 110678. DOI: 10.1016/j.est.2024.110678.
2	朱振威, 苗嘉伟, 祝夏雨, 等. 基于机器学习方法的锂电池剩余寿命预测研究进展[J]. 储能科学与技术, 2024, 13(9): 3134-3149. DOI: 10.19799/j.cnki.2095-4239.2024.0713.
	ZHU Z W, MIAO J W, ZHU X Y, et al. Research progress in lithium-ion battery remaining useful life prediction based on machine learning[J]. Energy Storage Science and Technology, 2024, 13(9): 3134-3149. DOI: 10.19799/j.cnki.2095-4239.2024.0713.
3	李英顺, 阚宏达, 郭占男, 等. 基于数据预处理和VMD-LSTM-GPR的锂离子电池剩余寿命预测[J]. 电工技术学报, 2024, 39(10): 3244-3258. DOI: 10.19595/j.cnki.1000-6753.tces.230210.
	LI Y S, KAN H D, GUO Z N, et al. Prediction of remaining useful life of lithium-ion battery based on data preprocessing and VMD-LSTM-GPR[J]. Transactions of China Electrotechnical Society, 2024, 39(10): 3244-3258. DOI: 10.19595/j.cnki.1000-6753.tces.230210.
4	陈峥, 陈洋, 申江卫, 等. 基于优化支持向量回归算法的锂离子电池可用容量估计[J]. 储能科学与技术, 2023, 12(10): 3203-3213. DOI: 10.19799/j.cnki.2095-4239.2023.0387.
	CHEN Z, CHEN Y, SHEN J W, et al. Available capacity estimation of lithium-ion batteriesbased on the optimized support vector regression algorithm[J]. Energy Storage Science and Technology, 2023, 12(10): 3203-3213. DOI: 10.19799/j.cnki.2095-4239. 2023.0387.
5	SUI X, HE S, VILSEN S B, et al. A review of non-probabilistic machine learning-based state of health estimation techniques for Lithium-ion battery[J]. Applied Energy, 2021, 300: 117346. DOI: 10.1016/j.apenergy.2021.117346.
6	申江卫, 高承志, 舒星, 等. 基于迁移模型的锂离子电池宽温度全寿命SOC与可用容量联合估计[J]. 电工技术学报, 2023, 38(11): 3052-3063. DOI: 10.19595/j.cnki.1000-6753.tces.220216.
	SHEN J W, GAO C Z, SHU X, et al. Joint estimation of SOC and usable capacity of lithium-ion battery with wide temperature and full life based on migration model[J]. Transactions of China Electrotechnical Society, 2023, 38(11): 3052-3063. DOI: 10. 19595/j.cnki.1000-6753.tces.220216.
7	XIONG W, MO Y M, YAN C. Online state-of-health estimation for second-use lithium-ion batteries based on weighted least squares support vector machine[J]. IEEE Access, 2020, 9: 1870-1881. DOI: 10.1109/ACCESS.2020.3026552.
8	梁新成, 张勉, 黄国钧. 基于BMS的锂离子电池建模方法综述[J]. 储能科学与技术, 2020, 9(6): 1933-1939. DOI: 10.19799/j.cnki.2095-4239.2020.0166.
	LIANG X C, ZHANG M, HUANG G J. Review on lithium-ion battery modeling methods based on BMS[J]. Energy Storage Science and Technology, 2020, 9(6): 1933-1939. DOI: 10.19799/j.cnki.2095-4239.2020.0166.
9	顾菊平, 蒋凌, 张新松, 等. 基于特征提取的锂离子电池健康状态评估及影响因素分析[J]. 电工技术学报, 2023, 38(19): 5330-5342. DOI: 10.19595/j.cnki.1000-6753.tces.231085.
	GU J P, JIANG L, ZHANG X S, et al. Estimation and influencing factor analysis of lithium-ion batteries state of health based on features extraction[J]. Transactions of China Electrotechnical Society, 2023, 38(19): 5330-5342. DOI: 10.19595/j.cnki.1000-6753.tces.231085.
10	XIONG R, LI Z Y, YANG R X, et al. Fast self-heating battery with anti-aging awareness for freezing climates application[J]. Applied Energy, 2022, 324: 119762. DOI: 10.1016/j.apenergy.2022.119762.
11	ZHAO Y, WANG Z P, SHEN Z M, et al. Assessment of battery utilization and energy consumption in the large-scale development of urban electric vehicles[J]. Proceedings of the National Academy of Sciences of the United States of America, 2021, 118(17): e2017318118. DOI: 10.1073/pnas.2017318118.
12	DENG Z W, HU X S, XIE Y, et al. Battery health evaluation using a short random segment of constant current charging[J]. iScience, 2022, 25(5): 104260. DOI: 10.1016/j.isci.2022.104260.
13	ZHANG C H, KANG Y Z, DUAN B, et al. An adaptive battery capacity estimation method suitable for random charging voltage range in electric vehicles[J]. IEEE Transactions on Industrial Electronics, 2022, 69(9): 9121-9132. DOI: 10.1109/TIE.2021.3111585.
14	DENG Z W, LIN X K, CAI J W, et al. Battery health estimation with degradation pattern recognition and transfer learning[J]. Journal of Power Sources, 2022, 525: 231027. DOI: 10.1016/j.jpowsour.2022.231027.
15	LIU K L, TANG X P, TEODORESCU R, et al. Future ageing trajectory prediction for lithium-ion battery considering the knee point effect[J]. IEEE Transactions on Energy Conversion, 2022, 37(2): 1282-1291. DOI: 10.1109/TEC.2021.3130600.
16	WEN S, LIN N, HUANG S X, et al. Lithium battery health state assessment based on vehicle-to-grid (V2G) real-world data and natural gradient boosting model[J]. Energy, 2023, 284: 129246. DOI: 10.1016/j.energy.2023.129246.
17	RUMPF K, NAUMANN M, JOSSEN A. Experimental investigation of parametric cell-to-cell variation and correlation based on 1100 commercial lithium-ion cells[J]. Journal of Energy Storage, 2017, 14: 224-243. DOI: 10.1016/j.est.2017.09.010.
18	林点, 潘理, 易平. 面向图像识别的卷积神经网络鲁棒性研究进展[J]. 网络与信息安全学报, 2022, 8(3): 111-122. DOI: 10.11959/j.issn.2096-109x.2022037.
	LIN D, PAN L, YI P. Research on the robustness of convolutional neural networks in image recognition[J]. Chinese Journal of Network and Information Security, 2022, 8(3): 111-122. DOI: 10.11959/j.issn.2096-109x.2022037.
19	ZHOU Z Y, LIU Y G, YOU M X, et al. Two-stage aging trajectory prediction of LFP lithium-ion battery based on transfer learning with the cycle life prediction[J]. Green Energy and Intelligent Transportation, 2022, 1(1): 100008. DOI: 10.1016/j.geits.2022.100008.
20	CHOI Y, RYU S, PARK K, et al. Machine learning-based lithium-ion battery capacity estimation exploiting multi-channel charging profiles[J]. IEEE Access, 2019, 7: 75143-75152. DOI: 10.1109/ACCESS.2019.2920932.
21	IOFFE S, SZEGEDY C. Batch normalization: Accelerating deep network training by reducing internal covariate shift[C]// International Conference on Machine Learning, 2015: 448-456.
22	WANG Y Y, LI Y B, SONG Y, et al. The influence of the activation function in a convolution neural network model of facial expression recognition[J]. Applied Sciences, 2020, 10(5): 1897. DOI: 10.3390/app10051897.
23	DUAN Y Z, TIAN J P, LU J H, et al. Deep neural network battery impedance spectra prediction by only using constant-current curve[J]. Energy Storage Materials, 2021, 41: 24-31. DOI: 10.1016/j.ensm.2021.05.047.
24	KINGMA D P, BA J, HAMMAD M M. Adam: A method for stochastic optimization[EB/OL]. https://arxiv.org/abs/1412.6980v9.
25	XUE J K, SHEN B. Dung beetle optimizer: A new meta-heuristic algorithm for global optimization[J]. The Journal of Supercomputing, 2023, 79(7): 7305-7336. DOI: 10.1007/s11227-022-04959-6.
26	WANG Q, YE M, WEI M, et al. Deep convolutional neural network based closed-loop SOC estimation for lithium-ion batteries in hierarchical scenarios[J]. Energy, 2023, 263: 125718. DOI: 10.1016/j.energy.2022.125718.
27	李晓东. Python人工智能[M]. 北京: 电子工业出版社, 2021.
	LI X D. Python artificial intelligence[M]. Beijing: Publishing House of Electronics Industry, 2021.
28	LI Y H, LI K, LIU X, et al. Lithium-ion battery capacity estimation — A pruned convolutional neural network approach assisted with transfer learning[J]. Applied Energy, 2021, 285: 116410. DOI: 10.1016/j.apenergy.2020.116410.

单体电池特性	电池单体参数
正极材料	磷酸铁锂
负极材料	石墨
额定容量	1.1 Ah
充电模式	CC-CV/C1-C2-CV
电压工作范围	2.0～3.6 V
额定电压	3.3 V
实验温度	30 ℃

架构设计	配置
输入层	“Z-score”规范化
卷积层	卷积核大小：[pop 3]，数量：[pop 1]，步长为1
卷积层	卷积核大小：[pop 3]，数量：[pop 2]，步长为1
卷积层	卷积核大小：[pop 5]，数量：[pop 2]，步长为1
卷积层	卷积核大小：[pop 5]，数量：[pop 2]，步长为1
丢弃层	丢弃比例：[pop 4]
平均池化层	步幅为1
全连接层	全连接第32个输出
全连接层	32个输入全连接1个输出
回归层	回归输出

起点电压 /V	窗口长度
	50		75		100
	R²	MAE/%	R²	MAE/%	R²	MAE/%
2.782	0.3642	4.56	0.5575	2.75	0.8719	1.44
3.036	0.8778	1.45	0.8492	1.59	0.8453	1.68
3.268	0.8017	1.79	0.9120	1.20	0.8805	1.43
3.422	0.8855	1.36	0.8554	1.59	0.8608	1.55
3.474	0.7945	1.93	0.7149	2.32	0.7671	2.07
3.500	0.7899	1.91	0.7332	2.20	0.8783	1.40
3.523	0.7917	1.93	0.8255	1.64	0.9611	0.84
3.553	0.8608	1.52	0.9615	0.80	0.9698	0.79
3.594	0.9211	1.29	0.9694	0.84	0.9752	0.73
3.392	0.9557	1.02	0.9787	0.69	0.9697	0.82
3.409	0.9556	1.02	0.9688	0.84	0.9727	0.77
3.422	0.9587	0.91	0.9640	0.84	0.9635	0.82
3.434	0.9023	1.38	0.9207	1.19	0.8834	1.33
平均误差	0.8495	1.71	0.8585	1.39	0.8954	1.17
覆盖率/%	61.1		77.7		88.9

窗口长度	评价	CNN	CNN-LSTM	CNN-GRU
100	R²	0.8954	0.8469	0.8427
	MAE/%	1.17	1.30	1.48
	覆盖率/%	88.9	77.7	77.7

电池16起点电压	R²	MAE/%	电池18起点电压	R²	MAE/%
2.825	0.8975	1.18	2.736	0.8572	1.00
3.062	0.9082	1.06	2.984	0.7186	1.22
3.306	0.9253	0.98	3.227	0.9705	0.57
3.481	0.9105	1.07	3.432	0.9708	0.57
3.535	0.8915	1.07	3.504	0.9748	0.54
3.468	0.9619	0.70	3.465	0.9741	0.52
3.466	0.9364	0.80	3.471	0.8702	0.87
3.482	0.9890	0.36	3.487	0.9813	0.39
3.503	0.9937	0.28	3.504	0.9795	0.45
3.533	0.9905	0.34	3.535	0.9839	0.42
3.583	0.9879	0.35	3.373	0.8924	0.95
			3.417	0.9011	0.81
			3.431	0.9256	0.81
平均误差	0.9448	0.74	平均误差	0.9231	0.70
覆盖率/%	87.5		88.9