基于充电数据的多阶段锂离子电池健康状态估计

State of health estimation of lithium-ion battery is the basis of lithium-ion battery life assessment and health management. A practical multi-stage state of health estimation method was proposed to deal with different charging stages, including the scene of serious lack of charging data. According to the voltage, the constant current-constant voltage charging process was divided into three stages and their target state of health estimation methods were proposed respectively. Especially for the constant current-constant voltage transition stage, being a lack of constant current data and constant voltage data heavily, the relationship between raw voltage/current data and battery state of health was directly established taking the strong data mining capability of convolutional neural network. The proposed method was evaluated by long-term aging experiments on lithium-ion battery. The results show that this method possesses the advantages of high estimation accuracy, strong ability to deal with serious data loss, and strong robustness to battery inconsistency.