State of charge (SOC) is one of the most important parameters for the battery management system. Due to the aircraft battery complex working conditions and strong nonlinear, it brings difficulties for the battery online SOC estimation. In order to improve the aircraft battery online SOC estimation accuracy under complicated stress condition, the temperature and discharge rate characteristics of battery performance parameters were researched. Then, this paper puts forward a improved EKF method which combine EKF algorithm with RLS algorithm. This method not only can realize online identification of the equivalent circuit model parameter, but also can realize the online SOC eatimation. This method was physical verified through experimental, and the experimental results show that the improved EKF method can enhance the online SOC estimation obviously, the accuracy can reach above 0.5%.