A multi-scale state of health prediction framework of lithium-ion batteries considering the temperature variation during battery discharge. (October 2021)

• Record Type: Journal Article
• Authors: Jia, Jianfang; Wang, Keke; Shi, Yuanhao; Wen, Jie; Pang, Xiaoqiong; Zeng, Jianchao
• Is Part Of: Journal of energy storage. Volume 42(2021)
• Publication Date: 2021-10
• Subjects: Lithium-ion battery -- State of health -- Multi-scale prediction model -- Change rate of temperature -- Wavelet packet transform -- Ensemble learning; Energy storage -- Periodicals; Energy storage -- Research -- Periodicals; 621.3126
• DOI: 10.1016/j.est.2021.103076
• Languages: English
• ISSNs: 2352-152X

Abstract

Abstract: State of health (SoH) prediction of Lithium-ion batteries is very essential to monitor and optimize battery behavior and safety in the battery management system (BMS). However, the capacity degradation of the battery is subject to nonlinear phenomena influenced by temperature variation, which is one of the main difficulties in SoH prediction. To solve this difficulty and improve the prediction accuracy, this paper proposes a novel multi-scale model for SoH prediction. Firstly, the temperature characteristic data is dealt with in the frequency domain to overcome the nonlinear fluctuations of the battery capacity degradation. The alternative frequency bands for the temperature characteristics and the capacity degradation of battery are determined by means of wavelet packet transform (WPT) and correlation analysis. Then, a prediction framework based on wavelet neural network and ensemble learning is developed, in which expectation maximization (EM) algorithm is used to optimize the prediction model. Finally, The prediction accuracy and robustness are verified by two data sets of NASA lithium-ion battery. Compared with other data-driven methods, the results show that the root-mean-square error (RMSE) value of the proposed method is less than 1.33% in the SoH prediction. Highlights: Multi-scale SoH prediction model based on WNN and EL is proposed. Temperature characteristics of battery discharge process are extracted. WPT and correlation analysis are applied to the feature data processing. SoH prediction results are optimized using EM algorithm. The effectivity of the proposed method is verified by two data sets.