Aiming at the shortcomings of low estimation accuracy and poor robustness of extended Kalman filter (EKF) state of charge (SOC) under non-Gaussian noise, an adaptive extended Kalman filter algorithm based on dynamic generalized maximum correntropy (DGMCC-AEKF) optimization is proposed. The algorithm improves the robustness by replacing the minimum mean square error criterion with the generalized maximum correntropy criterion, and proposes a strategy of dynamically adjusting the kernel width parameters based on the exponential form combined with the iterative method, which improves the accuracy and stability. Under dynamic stress test (DST) and federal urban driving schedule (FUDS) conditions, the mean absolute error and root mean squared error of the new algorithm are lower than those of the existing adapative extended Kalman filter (AEKF), adapative iterated extended Kalman filler (AIEKF) and maximum correntropy criterion extended Kalman filter (MCC-EKF) algorithms. Compared with other adaptive kernel width methods, it has better performance under four noise schemes.