Intelligent Control of Ultrasonic Motor Using PSO Type Neural Network

Ultrasonic motor (USM) is a novel kind of electronic motor which employs friction as its drive source. Owing to attractive features of USMs, they are expected to be applied widely in industrial world. However, in the applications of USMs, control performances are limited by characteristic variation and nonlinearity of USMs. To overcome the problems, many control methods have been proposed in previous research. In this paper, a scheme of internal model control (IMC)-PID control combined with particle swarm optimization (PSO) type neural network (NN) is studied for control of USM. In the method, an NN controller is designed for adjusting the control parameter in IMC-PID controller. Meanwhile, the weights of NN are updated by PSO algorithm on-line. By employing the proposed method, the characteristic changes and nonlinearity of USM can be ompensated. On the other hand, the application of PSO algorithm in NN´s learning avoids the problem of Jacobian estimation in conventional NNs which apply back-propagation (BP) method. The effectiveness of the method is confirmed by experiments.