Abstract: Accurately grasping the state of health(SOH) of lithium-ion batteries is crucial for the safe and stable operation of energy storage systems. However, due to the fact that battery SOH cannot be directly measured and its attenuation is influenced by various factors, the degradation process throughout the entire life cycle presents a non-linear nature, making it difficult to estimate battery SOH. Therefore, a method for estimating the SOH of lithium-ion batteries based on constant voltage charging data and stacking models is proposed. By analyzing the current data of the constant voltage charging stage under different cycle cycles, the variation pattern is revealed, and a health feature combination is proposed based on the charging time, current information entropy, current curve skewness, and maximum curvature of the constant voltage charging stage. In order to improve the generalization ability of SOH estimation models, a stacked model containing four different principles machine learning estimators is designed based on the combination of health features. The accuracy of SOH estimation results is improved through double-layer multi model fusion. The experimental results verify that the proposed health feature combination and model can achieve accurate estimation of battery SOH.