A novel quick and robust capacity estimation method for Li-ion battery cell combining information energy and singular value decomposition

Capacity estimation is an important basic function of today's battery management system (BMS) in electric vehicles (EVs). However, because the battery degradation demonstrates high non-linear property and has multiple external factors involved, accurate estimation of capacity still confronts with big challenges, especially under dynamic scenarios for EV application. In this paper, a novel capacity estimation method is proposed combing information energy theory and singular value decomposition (SVD). Firstly, spectral entropy of current and voltage are calculated and arranged to form the information energy matrix according to time series. Then SVD is used to extract two healthy indicators from the energy matrix. It is found that the two healthy indicators have strong linear correlation with capacity degradation if whole discharging data are used, therefore a linear regression model is sufficient to capture the relationship between the two healthy indicators and battery capacity. The robustness of the proposed method is verified against data under different temperatures and different driving cycles. The mean absolute estimation error of the approach is controlled within 2% in most cases. In addition, the proposed method could still be useful if the SoC range shrinks to a certain level.