Design for AC servo position loop based on RBF neural network predictive control

Aiming at wide variations in loads and moment of inertia of the machine tool position servo system, the position loop controller is designed based on RBF neural network control and predictive control. The mathematical model of AC PMSM is established. The predictive model is designed based on controlled autoregressive integral moving average model, obtained the predictive vector and the reference trajectory. The RBF neural network structure is established, tuning PID algorithm is designed. A new control strategy which combined with predictive control and RBF neural network PID control is obtained. Compared with the torque fluctuation, anti-interference ability and tracking properties of the traditional PID control, predictive control ensures that the system tracking performance and RBF neural network control can adjust PID parameters on-line, which guarantees the robustness in the external disturbance and parameter perturbation. The simulation results demonstrate that the RBF neural network predictive controller can guarantee the static and dynamic performance of the system.