SOH Estimation of Lithium-ion Battery Based on Energy Features and WOA-XGBoost Model

Lithium-ion batteries have been widely used in various energy storage systems. To ensure the system's stable operation, it is necessary to accurately estimate the state of health (SOH) of lithium-ion batteries. Aiming at the problems of low practicability and weak generalization ability of traditional methods, we propose a SOH estimation method for lithium-ion batteries based on whale optimization algorithm and extreme gradient boosting (WOA-XGBoost) model. Firstly, we select two energy-based features by studying the relationship between the capacity degradation of lithium-ion battery and the changing trend of voltage and current during charging. On the other hand, we use WOA to optimize the parameters of XGBoost and introduce data incremental update mechanism to improve the generalization ability of the WOA-XGBoost model. The proposed method is verified on NASA data sets. Experiments show that the proposed method has minor estimation errors and strong generalization ability.