Position control of ultrasonic motor using PID-IMC combined with neural network based on probability

Abstract

In this paper, a novel control scheme of PID type internal model control (PID-IMC) using neural network (NN) is proposed for the position control of ultrasonic motor (USM). In the proposed scheme, a new NN controller combined with particle swarm optimization (PSO) based on probability is introduced. The NN controller is employed for tuning the control parameter in PID-IMC for compensating the characteristic changes and non-linearity in USM. On one hand, the application of the proposed method avoids the Jacobian estimation in conventional back-propagation (BP) type NN. On the other hand, the PSO based on probability, originated from tribes-PSO, simplifies the PSO setting by reducing the parameters to be set in USM control. For confirming the effectiveness of the proposed method in position control, a group of experiments have been done on an existent USM servo system and the results are studied in this paper.