State of Health estimation of lithium-ion batteries under multistage constant current fast charging

In lithium-ion batteries (LIBs) applications, the widespread use of fast charging technology has dramatically enhanced user convenience. However, different fast charging strategies significantly impact batteries' State of Health (SOH). In particular, under multistage constant current fast charging conditions, accurate estimation of SOH of LIBs faces problems such as poor generalization ability and high computational cost. To solve these problems, this study proposes a battery health feature applicable to multistage constant current fast charging scenarios, which is extracted by charging voltage data, is easy to operate and relevant, and is not affected by the change of cut-off voltage in the later evaluation. Combined with a self-supervised learning approach, this study can accurately estimate the SOH of LIBs by utilizing only a tiny amount of labeling data while significantly reducing the computational cost. The proposed method is validated on two large publicly available fast charging datasets, and the results show that the Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) of SOH estimation for LIBs under multistage constant current fast charging are within 0.5 %, and the coefficient of determination (R2) is above 0.9950, which significantly saves computational cost while improving the model generalization capability.