In order to achieve the rapid response and robustness of a servo platform load simulator, a BP neural network (BP-GSA) sliding mode control method based on genetic simulated annealing algorithm (GSA) is proposed. According to the hardware composition of the load simulator, the equivalent mathematical model of the system is established. Non-singular terminal sliding mode is used to control the system, and BP neural network is used to approximate the undetermined terms in the state equation, and GSA algorithm is used to adjust the network node weights. The simulation results show that compared with the traditional sliding mode control and PID control, the proposed method has the smallest steady-state error and the fastest tracking speed in the case of disturbance input. The method can effectively improve the response speed and the torque tracking accuracy of the system.