IMC-PID Control of Ultra-Sonic Motor Servo System Based on Neural Network

Aimed at ultra-sonic motor (USM) has nonlinear input-output characteristics and the control parameters of conventional internal model control (IMC) cannot adjust automatically, an neural network (NN) is employed to tune the parameter of the IMC-PID control. The parameter is tuned online, and this will compensate the characteristics variation and uncertain non-linearity of the USM. Let the parameter be the output of the 3-layer NN. NN can get the suitable control parameters of a real-time control system after on-line learning. The weights of the NN are updated to minimize the positioning error of the USM servo system. The experiment results are shown that the control strategy is effective