Aiming at the time variability,low nonlinear control accuracy and poor stability of C32 continuous friction welding machine closed-loop control system,a control algorithm combining BP neural network and incremental PID controller was studied to improve the system performance.Referring to the actual operating parameters of the C32 continuous friction welding machine,a physical model was established through AMESim software;combined with Simulink,the neural network PID adaptive controller was established for joint simulation,and an electro-hydraulic closed-loop control system was established.The results show that the response speed,rise time,control accuracy and stability of C32 welding machine non-linear control system under neural network PID adaptive control are all better than the traditional PID control.The nonlinear control system controlled by BP-PID controller can make the system respond quickly and eliminate the overshoot of the original controller,which greatly improves the stability of the system.