To address the problem of low force control precision caused by the nonlinear and time-varying characteristics of electro-pneumatic proportional force control systems, a design of an electro-pneumatic proportional force control system based on GA-BP neural network is proposed. The mathematical model of the system was established, and the system control structure and algorithm based on GA-BP neural network were proposed. The random force tracking control precision of the system under varying load capacity and varying load displacement was studied through AMEsim/Simulink joint simulation and experimental research. The results show that compared to conventional PID control, the mean absolute error of random force tracking of this algorithm can be reduced by 48.6%. This algorithm is simple and practical with strong robustness, and can provide guidance for the design of pneumatic force control systems.