In order to solve the problem that the current Extended Kalman Filter (EFK) algorithm is vulnerable to the interference of system noise and the accumulation of model parameter identification error, which leads to the low estimation accuracy of the SOC of lithium-ion batteries, an adaptive Extended Kalman Filter (AEKF) algorithm is proposed in this paper. Based on the second-order Thevenin equivalent circuit model, the HPPC test of lithium ion battery was carried out. The parameters of the battery model were identified by the test data, and the adaptive Kalman filter algorithm was used to establish the battery state estimation model. Compared with the standard EKF, the adaptive extended Kalman filter algorithm improves the battery state estimation model by automatically correcting errors through iterative cycles, and the prediction accuracy is stabilized within 1%, which improves the stability and accuracy of the battery SOC prediction model.