SOH monitoring of new energy vehicle lithium batteries based on IPSO-LSTM

doi:10.16265/j.cnki.issn1003-3033.2023.09.1250

Abstract

Abstract:

In order to monitor the SOH of lithium batteries of new energy vehicles and prevent the risk of safety accidents caused by battery failures, a model combining IPSO and LSTM was proposed. First, Spearman correlation analysis method was used to extract the SOH of lithium batteries. Secondly, the linear inertia weight and asymmetric learning factor were used to improve the traditional particle swarm optimization(PSO) algorithm, and the IPSO algorithm was used to optimize the key parameters of the hidden layer neuron number, neuron inactivation rate and batch processing value of the LSTM model, and then the LSTM model was further optimized, and the IPSO-LSTM lithium battery SOH monitoring model was established. Finally, the model was verified with the 18650 lithium battery data set which was the mainstream of new energy vehicles, and compared with BP, LSTM and PSO-LSTM models. The results show that the mean absolute error (MAE) of IPSO-LSTM model is less than 0.02, the root mean square error (RMSE) is less than 0.03, and the monitoring error is less than 15%. Compared with BP, LSTM and PSO-LSTM model, the error index value of IPSO-LSTM model is the smallest, and the model has higher accuracy and stability.

Key words: improved particle swarm optimization (IPSO), long short-term memory(LSTM), new energy vehicle, lithium battery, state of health (SOH)

LIU Dan, WANG Ruihu, LYU Wei, QIN Ling, LIN Shuichun. SOH monitoring of new energy vehicle lithium batteries based on IPSO-LSTM[J]. China Safety Science Journal, 2023, 33(9): 94-102.

Figures/Tables 12

Fig.1

Fig.2

Fig.3

Fig.4

Tab.1

Results of correlation analysis of health characteristics

健康特征	Spearman系数
健康特征	B5	B6	B7	B18
$T 1$	0.954	0.957	0.952	0.917
$T 2$	-0.885	-0.914	-0.852	-0.812
$F 1$	0.953	0.957	0.952	0.917
$F 2$	-0.963	-0.951	-0.890	-0.863
$F 3$	0.954	0.956	0.950	0.920
$F 4$	-0.876	-0.911	-0.852	-0.746
$F 5$	0.949	0.955	0.945	0.851
$F 6$	-0.867	-0.889	-0.830	-0.743
$A 1$	0.037	0.127	0.018	-0.191
$A 2$	-0.337	0.089	-0.056	-0.108

Tab.1

Fig.5

Fig.6

Tab.2

Parameter settings of IPSO

IPSO参数	参数值
粒子数量	20
迭代次数	20
学习因子	$c 1, m a x =$ 1.5, $c 1, m i n =$ 0.7 $c 2, m a x =$ 1.5, $c 2, m i n =$ 0.7
惯性权重	$ω m a x$ =0.9, $ω m i n =$ 0.4

Tab.2

Tab.3

Parameter settings of four models

监测模型	参数设置
IPSO-LSTM	$l 1$ =[60,90], $l 2$ =[40,60],d=[30,50], $α$ =[0.18,0.23], $β$ =[10,30], $G m a x =$ 30
PSO-LSTM	$l 1$ =[60,90], $l 2$ =[40,60],d=[30,50], $α$ =[0.18,0.23], $β$ =[10,30], $G m a x =$ 30 $ω = 0.5, c 1 = 1, c 2 = 1$
LSTM	$l 1$ =75, $l 2$ =50,d=40, $α$ =0.2, $β$ =20, $G m a x =$ 30
BP	lr = 0.01, $G m a x =$ 1 000,输入层节点=5,输出层节点数=1

Tab.3

Fig.7

Tab.4

RUL prediction results of four models

型号	RUL	BP			LSTM			PSO-LSTM			IPSO-LSTM
型号	RUL	$E R$	$E a$	$E r$ /%	$E R$	$E a$	$E r$ /%	$E R$	$E a$	$E r$ /%	$E R$	$E a$	$E r$ /%
B5	26	44	18	69.2	33	7	26.9	28	2	7.7	27	1	3.8
B6	15	21	6	40.0	19	4	26.6	16	1	6.7	15	0	0
B7	62	29	33	53.2	57	5	8.1	66	4	6.5	63	1	1.6
B18	16	—	—	—	19	3	18.8	19	3	18.8	14	2	12.5

Tab.4

Tab.5

References 18

[1]	梁新苗, 肖凌云, 王澎, 等. 数据挖掘与现场调查结合的电动汽车事故分析[J]. 中国安全科学学报, 2022, 32(1):180-187. doi: 10.16265/j.cnki.issn1003-3033.2022.01.024
	LIANG Xinmiao, XIAO Lingyun, WANG Peng, et al. Analysis on electric accidents based on data mining and site investigation[J]. China Safety Science Journal, 2022, 32(1):180-187. doi: 10.16265/j.cnki.issn1003-3033.2022.01.024
[2]	工业和信息化部办公厅. 关于进一步加强新能源汽车企业安全体系建设的指导意见[EB/OL].(2022-04-09). http://www.gov.cn/zhengce/zhengceku/2022-04/09/content_5684250.htm.
[3]	李毅, 于东兴, 张少禹, 等. 锂离子电池火灾危险性研究[J]. 中国安全科学学报, 2012, 22(11):36-41.
	LI Yi, YU Dongxing, ZHANG Shaoyu, et al. Research on fire hazard of lithium-ion battery[J]. China Safety Science Journal, 2012, 22(11):36-41.
[4]	高飞, 杨凯, 李大贺, 等. 锂离子电池组件燃烧性及危险性评价[J]. 中国安全科学学报, 2015, 25(8):62-67.
	GAO Fei, YANG Kai, LI Dahe, et al. Evaluation of combustibility of lithium-ion battery components and dangers they involve[J]. China Safety Science Journal, 2015, 25(8):62-67.
[5]	CHENG Zhang, ALLAFI W, DINH Q, et al. Online estimation of battery equivalent circuit model parameters and state of charge using decoupled least squares technique[J]. Energy, 2018, 142:678-688. doi: 10.1016/j.energy.2017.10.043
[6]	PATIL M A, TAGADE P, HARIHARAN K S, et al. A novel multistage support vector machine based approach for Lithium-ion battery remaining useful life estimation[J]. Applied Energy, 2015, 159(12):285-297. doi: 10.1016/j.apenergy.2015.08.119
[7]	HE Zhiwei, GAO Mingyu, MA Guojin, et al. Online state-of-health estimation of lithium-ion batteries using dynamic Bayesian networks[J]. Journal of Power Sources, 2014, 267(3):576-583. doi: 10.1016/j.jpowsour.2014.05.100
[8]	YANG Duo, ZHANG Xu, PAN Rui, et al. A novel Gaussian process regression model for state-of-health estimation of lithium-ion battery using charging curve[J]. Journal of Power Sources, 2018, 384:387-395. doi: 10.1016/j.jpowsour.2018.03.015
[9]	HARIHARAN K S, KUMAR V S. A nonlinear equivalent circuit model for lithium ion cells[J]. Journal of Power Sources, 2013, 222(1):210-217. doi: 10.1016/j.jpowsour.2012.08.090
[10]	KHALEGHI S, KARIMI D, BEHESHTI S H, et al. Online health diagnosis of lithium-ion batteries based on nonlinear autoregressive neural network[J]. Applied Energy, 2021, 282: DOI: 10.1016/j.apenergy.2020.116159.
[11]	ZHOU Yapeng, HUANG Miaohua, CHEN Yupu, et al. A novel health indicator for on-line lithium-ion batteries remaining useful life prediction[J]. Journal of Power Sources, 2016, 321(7):1-10. doi: 10.1016/j.jpowsour.2016.04.119
[12]	WANG Cunsong, LU Ningyun, WANG Senlin, et al. Dynamic long short-term memory neural-network-based indirect remaining-useful-life prognosis for satellite lithium-ion battery[J]. Applied Sciences, 2018, 8(11):2078-2089. doi: 10.3390/app8112078
[13]	王义, 刘欣, 高德欣. 基于BiLSTM神经网络的锂电池SOH估计与RUL监测[J]. 电子测量技术, 2021, 44(20):1-5.
	WANG Yi, LIU Xin, GAO Dexin. The SOH estimation and RUL prediction of lithium battery based on BiLSTM[J]. Electronic Measurement Technology, 2021, 44(20):1-5.
[14]	MA Guijun, ZHANG Yong, CHENG Cheng, et al. Remaining useful life prediction of lithium-ion batteries based on false nearest neighbors and a hybrid neural network[J]. Applied Energy, 2019, 253: DOI: 10.1016/j.apenergy.2019.113626.
[15]	孙振宇, 王震坡, 刘鹏, 等. 新能源汽车动力电池系统故障诊断研究综述[J]. 机械工程学报, 2021, 57(14):87-104. doi: 10.3901/JME.2021.14.087
	SUN Zhenyu, WANG Zhenpo, LIU Peng, et al. Overview of fault diagnosis in new energy vehicle power battery system[J]. Journal of Mechanical Engineering, 2021, 57(14):87-104. doi: 10.3901/JME.2021.14.087
[16]	ATALAY S, SHEIKH M, MARIANI A, et al. Theory of battery ageing in a lithium-ion battery: capacity fade, nonlinear ageing and lifetime prediction[J]. Journal of Power Sources, 2020, 478: DOI: 10.1016/j.jpowsour.2020.229026.
[17]	李放, 闵永军, 王琛, 等. 基于充电过程的锂电池SOH估计和RUL预测[J]. 储能科学与技术, 2022, 11(10):3316-3327.
	LI Fang, MIN Yongjun, WANG Chen, et al. SOH estimation and RUL prediction of lithium batteries based on charging process[J]. Energy Storage Science and Technology, 2022, 11(10):3316-3327.
[18]	魏腾飞, 潘庭龙. 基于改进PSO优化LSTM网络的短期电力负荷监测[J]. 系统仿真学报, 2021, 33(8):1866-1874. doi: 10.16182/j.issn1004731x.joss.20-0297
	WEI Tengfei, PAN Tinglong. Short-term power load forecasting based on LSTM neural network optimized by improved PSO[J]. Journal of System Simulation, 2021, 33(8):1866-1874. doi: 10.16182/j.issn1004731x.joss.20-0297

型号	BP		LSTM		PSO-LSTM		IPSO-LSTM
型号	MAE	RMSE	MAE	RMSE	MAE	RMSE	MAE	RMSE
B5	0.059 3	0.070 7	0.032 9	0.036 0	0.019 7	0.024 5	0.010 0	0.015 0
B6	0.073 9	0.086 4	0.039 1	0.042 4	0.028 3	0.035 4	0.018 3	0.027 2
B7	0.047 3	0.057 9	0.021 7	0.025 7	0.011 5	0.018 5	0.009 9	0.016 4
B18	0.074 8	0.085 8	0.040 4	0.044 8	0.024 8	0.031 1	0.013 9	0.025 4