A Robust Lithium-ion Battery SoH Estimation Method Using Refined RC-Network ECM and SVR

Efficient and accurate battery state-of-health (SoH) estimation is crucial for ensuring the safe and reliable operation of the on-board battery system. The battery SoH estimation methods utilizing charging data under the constant-voltage (CV) scenario have gathered the widespread attention due to their protocol simplicity and insensitivity to initial states. Firstly, a refined equivalent circuit model containing two parallel-connected RC networks is introduced to accurately characterize the CV charging current. Furthermore, considering the incomplete CV charging scenario, the correlation analysis is conducted to select the feature-of-interest (FoI) related to the battery capacity degradation, and determine the corresponding data length interval. Subsequently, combined the support vector regression with the grey wolf optimization, the mapping relationships between the battery SoH and the selected FoI for different data lengths are constructed to obtain the battery SoH estimation model. Lastly, the validation results based on the Tongji University dataset demonstrated that the root-mean-square-error (RMSE) of the current estimation by the proposed refined battery model is generally less than 30 mA throughout the aging process, and RMSEs of the battery SoH estimation results are overall within 3.5% under different charging scenarios, proving the accuracy and the robustness of the proposed method.