Robust Neural Network Controller Design for Traveling-Wave Type Ultrasonic Motor

Abstract

The control and identification of a traveling-wave ultrasonic motor (TWUSM) using a neural network control scheme is described in this paper. A recurrent neural network (RNN) control scheme that uses continual online training to simultaneously identify and control the TWUSM is proposed. The control scheme utilizes two three-layer RNNs: one is a recurrent neural network identifier which captures the nonlinear dynamics of the plant system and the other is a recurrent neural network controller to provide the necessary control actions to achieve desired reference trajectory. Some experimental results reveal that the proposed control structure adapts and generalizes its learning to a wide range of operating conditions and provides promising results under system uncertainties and load changes.