In order to solve a series of nonlinear factors faced by gun servo system, an active disturbance rejection control (ADRC) strategy based on echo state network (ESN) is designed. The echo state network is used to realize the online tuning of the important parameters of ADRC, and the gradient descent algorithm and the improved gray wolf optimization (GWO) algorithm are introduced to train the echo state network. The simulation results show that the new control method can effectively improve the dynamic response performance, anti-interference performance and tracking accuracy of the gun servo system, and meet the performance requirements of the gun servo system.