Intelligent state of health estimation for lithium-ion battery pack based on big data analysis

• Published in: Journal of energy storage 2020-12, Vol.32, p.101836, Article 101836
• Main Authors: Song, Lingjun, Zhang, Keyao, Liang, Tongyi, Han, Xuebing, Zhang, Yingjie
• Format: Article
• Language: English
• Subjects: Big data analysis; Lithium-ion battery; Machine learning method; State of health
• ISSN: 2352-152X; 2352-1538
• DOI: 10.1016/j.est.2020.101836

•An intelligent SOH estimation framework for EVs big data platform is presented.•The capacity estimation model is established based on neural network algorithm.•The dataset is generated from real-word data of EVs under actual operation.•Big data analysis is used for extraction of health feature parameters and selection of training data.•Accurate SOH evaluation can be obtained by using the historical operating data. State of health (SOH) of in-vehicle lithium-ion batteries not only directly determines the acceleration performance and driving range of electric vehicles (EVs), but also reflects the residual value of the batteries. Especially, with the development of data acquisition and analysis technologies, using big data to realize on-line evaluation of battery SOH shows vital significance. In this paper, we propose an intelligent SOH estimation framework based on the real-world data of EVs collected by the big data platform. Defined by the more accessible detection, the health features are extracted from historical operating data. Then, the deep learning process is implemented in feedforward neural network driven by the degradation index. The estimation method is validated by the one-year monitoring dataset from 700 vehicles with different driving mode. The result shows that the proposed framework can effectively estimate SOH with the maximum relative error of 4.5% and describe the aging trend of battery pack based on big data platform.