State of health estimation based on subtraction average based optimizer and bidirectional long and short term memory networks for lithium-ion batteries

doi:10.19799/j.cnki.2095-4239.2024.0526

Abstract

Abstract:

Accurate estimation of state of health (SOH) is crucial for ensuring the safe and reliable operation of lithium-ion batteries and extending their service life. To address the challenges posed by many health features that fail to characterize the aging mechanism of batteries and the inability to accurately track SOH trends under abnormal working conditions, this study introduces an SOH estimation approach that combines empirical models with data-driven techniques. The lithium-ion battery anode SEI film thickening mechanism is incorporated into Arrhenius' law to develop an empirical model. Parameters are identified using the least-squares method, and Spearman's correlation coefficients are calculated for each parameter and capacity. The results indicate strong correlations between these parameters and capacity decline, confirming their viability as reliable health indicators for estimating SOH. Furthermore, to address the issue of the bidirectional long and short term memory (BiLSTM) network having excessive parameters and being prone to overfitting, this study employs the subtraction average based optimizer (SABO) algorithm to optimize the hyperparameters of the BiLSTM and develop the SOH estimation model. The proposed approach is validated using experimental test data and data from the National Aeronautics and Space Administration data. In addition, the performance of the method is compared with the estimation results from three algorithms: long and short-term memory (LSTM) network, BiLSTM network, and BiLSTM network with Particle Swarm Optimization (PSO). The results reveal that the SABO-BiLSTM model achieves mean absolute percentage errors of 0.043%, 0.053%, 0.259%, and 0.230% for the SOH estimation of four batteries. These values represent reductions of 94.71%, 92.62%, 88.75%, and 90.13% compared to the LSTM model, reductions of 89.11%, 91.60%, 77.90%, and 76.41% compared to the BiLSTM model, and reductions of 58.65%, 58.91%, 65.37%, and 69.29% compared to the PSO-BiLSTM model.

Key words: lithium-ion batteries, Arrhenius law, subtraction average based optimizer, bidirectional long and short-term memory networks

CLC Number:

TM 912

Jianxuan LI, Chen LIN, Zhongkai ZHOU. State of health estimation based on subtraction average based optimizer and bidirectional long and short term memory networks for lithium-ion batteries[J]. Energy Storage Science and Technology, 2025, 14(1): 358-369.

Figures/Tables 14

Fig.1

Table 1

Table 2

Fig.2

Fig.3

Fig.4

Table 3

Fig.5

Fig.6

Fig.7

Table 4

Fig.8

Table 5

References 29

1	JI D, GAN H. Effects of providing total cost of ownership information on below-40 young consumers' intent to purchase an electric vehicle: A case study in China[J]. Energy Policy, 2022, 165: 112954.
2	LIU K. Data-driven health estimation and lifetime prediction of lithium-ion batteries: A review[J]. Renewable and Sustainable Energy Reviews, 2019, 113: 109254.
3	霍丽萍, 栾伟玲, 庄子贤 . 锂离子电池储能安全技术的发展态势——从全球专利数据分析我国的发展现状[J]. 储能科学与技术, 2022, 11(8): 2671-2680.
	HUO L P, LUAN W L, ZHUANG Z X. Development trend of energy storage safety technology for lithium-ion batteries analysis of global patent data and china's development status [J]. Energy Storage Science and Technology, 2022, 11(8): 2671-2680.
4	TENG J H, CHEN R J, LEE P T, et al. Accurate and efficient SOH estimation for retired batteries[J]. Energies, 2023, 16(3): 1240.
5	XU Z, WANG J, LUND P D, et al. Co-estimating the state of charge and health of lithium batteries through combining a minimalist electrochemical model and an equivalent circuit model[J]. Energy, 2022, 240: 122815.
6	NI Y L, XU J N, ZHU C B, et al. Accurate residual capacity estimation of retired LiFePO4 batteries based on mechanism and data-driven model[J]. Applied Energy, 2022, 305: 117922.
7	LI X, JU L, GENG G, et al. Data-driven state-of-health estimation for lithium-ion battery based on aging features[J]. Energy, 2023, 274: 127378.
8	YAN M A, YANG C, FAN Z, et al. Remaining useful life prediction of power battery based on extend h_∞ particle filter algorithm[J]. Journal of Mechanical Engineering, 2019, 55(20): 36-43.
9	CAI Y, CHEN W, SU Y, et al. Review of remaining useful life prediction for lithium-ion batteries[J]. Journal of Power Sources, 2021, 45: 678-682.
10	HU X, CHE Y, LIN X, et al. Battery health prediction using fusion-based feature selection and machine learning[J]. IEEE Transactions on Transportation Electrification, 2020, 7(2): 382 -398.
11	XU Z, WANG J, LUND P D, et al. Estimation and prediction of state of health of electric vehicle batteries using discrete incremental capacity analysis based on real driving data[J]. Energy, 2021, 225: 120160.
12	WENG C, CUI Y, SUN J, et al. On-board state of health monitoring of lithium-ion batteries using incremental capacity analysis with support vector regression[J]. Journal of Power Sources, 2013, 235: 36-44.
13	JENU S, HENTUNEN A, HAAVISTO J, et al. State of health estimation of cycle aged large format lithium-ion cells based on partial charging[J]. Journal of Energy Storage, 2022, 46: 103855.
14	LIN Z, HU H, LIU W, et al. State of health estimation of lithium-ion batteries based on remaining area capacity[J]. Journal of Energy Storage, 2023, 63: 107078.
15	ZHU G R, KONG C, WANG J V, et al. A fractional-order model of lithium-ion battery considering polarization in electrolyte and thermal effect[J]. Electrochimica Acta, 2023, 438: 141461. DOI: 10.1016/j.electacta.2022.141461.
16	LI J, ADEWUYI K, LOTFI N, et al. A single particle model with chemical/mechanical degradation physics for lithium-ion battery State of Health (SOH) estimation[J]. Applied Energy, 2018, 212: 1178-1190.
17	SCHMIDT A P, BITZER M, IMRE Á W, et al. Model-based distinction and quantification of capacity loss and rate capability fade in Li-ion batteries[J]. Journal of Power Sources, 2010, 195(22): 7634-7638.
18	FU H J, LIU Z G, CUI K X, et al. Physics-informed neural network for spacecraft lithium-ion battery modeling and health diagnosis[J]. IEEE/ASME Transactions on Mechatronics, 2024. DOI: 10.1109/TMECH.2023.3348519.
19	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. SOH and remaining life prediction of lithium battery based on hybrid model and LSTM[J]. Energy Storage Science and Technology, 2021, 10(2): 689-694.
20	王义, 刘欣, 高德欣 . 基于BiLSTM神经网络的锂电池SOH估计与RUL预测[J].电子测量技术, 2021, 44(20): 1-5.
	WANG Y, LIU X, GAO D X. SOH estimation and RUL prediction of lithium battery based on BiLSTM neural network[J]. Electronic Measurement Technology, 2021, 44(20): 1-5.
21	ZHANG Y, XIN Y Q, Qian Q Z. PSO-optimised BiLSTM-attention network for lithium battery health state assessment[J]. Control Engineering, 2022, 29(2): 7.
22	TROJOVSKÝ P, DEHGHANI M. Subtraction-average-based optimizer: a new swarm-inspired metaheuristic algorithm for solving optimization problems[J]. Biomimetics, 2023, 8(2): 149.
23	WANG J, LIU P, HICKS-GARNER J, et al. Cycle-life model for graphite-LiFePO4 cells[J]. Journal of Power Sources, 2011, 196(8): 3942-3948.
24	BLOOM I, COLE B W, SOHN J J, et al. An accelerated calendar and cycle life study of Li-ion cells[J]. Journal of Power Sources, 2001, 101(2): 238-247.
25	韩雪冰 . 车用锂离子电池机理模型与状态估计研究[D]. 北京:清华大学, 2016.
	HAN X B. Research on mechanism model and state estimation of automotive lithium-ion batteries [D]. Beijing: Tsinghua University, 2016.
26	吴欢欢, 代娟, 朱振东, 等 . 锂离子电池负极满嵌锂石墨的热稳定性[J]. 电池, 2020, 50(5): 446-449.
	WU H H, DAI J, ZHU Z D, et al. Thermal stability of lithium-ion batteries anode with fully lithiated graphite[J]. Battery Bimonthly, 2020, 50(5): 446-449.
27	米成 . 锂离子电池界面反应活化能应用研究[J]. 湖南有色金属, 2023, 39(1): 55-58.
	MI C. Application research on activation energy of interface reaction in lithium-ion batteries [J]. Hunan Nonferrous Metals, 2023, 39(1): 55-58.
28	毛百海, 覃吴, 肖显斌, 等 . 基于LSTM&GRU-Attention多联合模型的锂离子电池SOH估计[J]. 储能科学与技术, 2023, 12(11): 3519-3527.
	MAO B H, QIN W, XIAO X B, et al. Based on LSTM&GRU-attention multi-modal model, state-of-health estimation of lithium-ion batteries[J]. Energy Storage Science and Technology, 2023, 12(11): 3519-3527.
29	SUN S, SUN J, WANG Z, et al. Prediction of battery soh by cnn-bilstm network fused with attention mechanism[J]. Energies, 2022, 15(12): 4428.

步骤	步骤名称	电流	时间	截止条件
1	恒流充电	1/2 C	—	U ≥4.2 V
2	恒压充电	—	—	I≤1.5 A
3	静置	—	15 min	—
4	恒流放电	1 C	—	U≤3 V
5	静置	—	15 min	—
6	重复1~5步	—	—	SOH≤80%

项目	参数
正极材料	LiNi_0.5M_n0.3Co_0.2
额定容量	2500 mAh
额定电压	3.65 V
最大充电倍率	3 C
生产商	天津力神电池股份有限公司

参数	T21	T22	B005	B007
a	0.999	0.999	0.994	0.997
b	-0.999	-0.999	-0.996	-0.997
c	0.985	0.984	0.841	0.886

Cell	Criteria	LSTM	BiLSTM	PSO-BiLSTM	SABO-BiLSTM
T21	MAE/%	0.699	0.341	0.091	0.036
	RMSE/%	1.549	0.439	0.163	0.054
	MAPE/%	0.794	0.395	0.104	0.043
	Training time/s	1.660	9.753	281.812	103.183

T22	MAE/%	0.659	0.562	0.110	0.045
	RMSE/%	1.465	0.673	0.192	0.069
	MAPE/%	0.741	0.631	0.129	0.053
	Training time/s	1.196	8.061	330.265	109.753

Cell	Criteria	LSTM	BiLSTM	PSO-BiLSTM	SABO-BiLSTM
B005	MAE/%	1.816	0.900	0.636	0.207
	RMSE/%	2.502	1.242	0.938	0.278
	MAPE/%	2.282	1.172	0.748	0.259
	Training time/s	0.690	5.326	168.793	64.482

B007	MAE/%	1.957	0.842	0.663	0.205
	RMSE/%	2.717	1.019	0.965	0.267
	MAPE/%	2.331	0.975	0.749	0.230
	Training time/s	0.741	5.131	182.526	59.685