Speed regulating valves have good load-sensing characteristics through pressure compensation valves or digital compensators. However, the structure of the speed regulating valve is complex and difficult to manufacture, and the circuit has throttling power loss and low efficiency. Aiming at the above problems, proportional relief valve was used to control the circuit to realize the function of load sensitivity and real-time speed control through two control methods of PID and neural network inverse model. The dynamic characteristics of the circuit were analyzed by building a co-simulation model of AMESim and MATLAB/Simulink. The simulation results show that the both control methods can make the circuit realize the functions of load sensitivity and real-time speed regulation. In terms of load sensitivity characteristics, the neural network inverse model is better than PID, which has a faster response and less overshoot. In terms of real-time speed regulation, PID is better than neural network inverse model, which has a faster response.