Research of feedforward +RBF neural network for permanent magnet linear Motor in XY motion platform

Abstract

In recent years, the semiconductor industry has developed rapidly, and more and more industries have applied chips, requiring higher and higher precision of chips. The XY platform of the welding line machine used for semiconductor packaging equipment is composed of two permanent magnet linear motors. The improvement of the control precision of the linear motor can improve the accuracy of the welding line machine in semiconductor processing. In this paper, in order to reduce the vibration of linear motor when starting or stopping, considering the working situation of the motor comprehensively, a third-order S-curve velocity planning scheme is adopted. Then, the common control algorithm is analyzed, and combined with the application of linear motor and precision requirements, the three control algorithms including PID control algorithm, RBF neural network control algorithm and feedforward control algorithm are combined, respectively, the linear motor under PID control, feedforward +PID, Feedforward + RBF-PID for three cases of system tracking error. In order to discuss whether the control algorithm designed in this paper can improve the robustness of linear motor, step input is added at the end to analyze the error variation of the system. Finally, the simulation is carried out on Simulink, the mathematical model of linear motor is built well, and the tracking error in three cases is plotted. It is concluded that the linear motor can effectively improve the robustness of the system, reduce the regulation time, reduce the overshoot and reduce the vibration under the compound control of "feedforward + RBF-PID".