State of health estimation of lithium-ion batteries based on Mixers-bidirectional temporal convolutional neural network

doi:10.1016/j.est.2023.109248

CSpace

	State of health estimation of lithium-ion batteries based on Mixers-bidirectional temporal convolutional neural network
	Gao, Jingyi 1; Yang, Dongfang 2; Wang, Shi 3; Li, Zhaoting 4; Wang, Licheng 5; Wang, Kai 1,6
	2023-12-20
发表期刊	JOURNAL OF ENERGY STORAGE
ISSN	2352-152X
卷号	73 页码:15
摘要	Accurate state of health (SOH) estimation is essential for designing a safe and reliable battery management systems (BMS). Although data-driven methods have achieved great accuracy and satisfied robustness in SOH estimation, for most neural networks, it is a challenge for SOH estimation to learn long dependencies in the training process due to the lack of the scalability for modeling long sequences. In this study, a novel SOH estimation framework combing Mixers and bidirectional temporal convolutional neural network (BTCN) is proposed, which not only takes the greatest advantage of local and global properties of input features to estimate SOH of lithium-ion batterie (LIBs), but also eases the redundancy of temporal and channel information. In the data pre-processing, the voltage change in the equal time interval is extracted from the measured data of the constant current (CC) charging stage, which is easily obtained in the real-world charging scenario. Then, the features that are highly correlated with SOH are selected by Pearson correlation coefficient (PCC), and all the features are normalized by minimum-maximum scaling method to speed up the convergence process and reduce the initialization requirement of learning-rate. After pre-processing, all features are input into the Mixers-BTCN model. We carry out experiments on aging data from two public datasets, NASA and CALCE. The simulations results indicate that the R2 for each dataset is above 0.768. The mean absolute error (MAE) and root mean square error (RMSE) that are both held within 2.34 %, which proves the accuracy and stability of the proposed SOH estimation. In addition, the introduction of transfer learning technology verifies the robustness of the proposed model to different ambient temperatures.
关键词	Lithium-ion battery State of health estimation Bidirectional temporal convolutional neural network Mixers Transfer learning
DOI	10.1016/j.est.2023.109248
收录类别	SCI
语种	英语
资助项目	Guangdong Provincial Key Lab of Green Chemical Product Technology[GC202111] ; Zhejiang Province Natural Science Foundation[LY22E070007] ; National Natural Science Foundation of China[52007170]
WOS研究方向	Energy & Fuels
WOS类目	Energy & Fuels
WOS记录号	WOS:001097429700001
出版者	ELSEVIER
引用统计	被引频次：45[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://119.78.100.204/handle/2XEOYT63/38114
专题	中国科学院计算技术研究所
通讯作者	Yang, Dongfang; Wang, Kai
作者单位	1.Qingdao Univ, Weihai Innovat Res Inst, Sch Elect Engn, Qingdao 266000, Peoples R China 2.Shaanxi Univ Sci & Technol, Haojing Coll, Xian 712046, Peoples R China 3.Chinese Acad Sci, Key Lab Intelligent Informat Proc, Inst Comp Technol, Beijing, Peoples R China 4.Brown Univ, Sch Engn, Providence, RI 02912 USA 5.Zhejiang Univ Technol, Sch Informat Engn, Hangzhou 310014, Hangzhou, Peoples R China 6.Shandong Suoxiang Intelligent Technol Co Ltd, Weifang 261101, Peoples R China
推荐引用方式 GB/T 7714	Gao, Jingyi,Yang, Dongfang,Wang, Shi,et al. State of health estimation of lithium-ion batteries based on Mixers-bidirectional temporal convolutional neural network[J]. JOURNAL OF ENERGY STORAGE,2023,73:15.
APA	Gao, Jingyi,Yang, Dongfang,Wang, Shi,Li, Zhaoting,Wang, Licheng,&Wang, Kai.(2023).State of health estimation of lithium-ion batteries based on Mixers-bidirectional temporal convolutional neural network.JOURNAL OF ENERGY STORAGE,73,15.
MLA	Gao, Jingyi,et al."State of health estimation of lithium-ion batteries based on Mixers-bidirectional temporal convolutional neural network".JOURNAL OF ENERGY STORAGE 73(2023):15.