Robust neuro control for rotating disc vibrations

Abstract

A novel robust-neuro control scheme for rotating disc vibrations is explored. In the approach proposed, a robust controller is first adopted to suppress the regulation error resulting from the unmodelled residuals while avoiding the spillover effect. In addition to the robust controller, a multilayer feedforward neural network is introduced to improve regulation performance. To maintain loop stability during neural network learning, a confinement algorithm for adaptive adjustment of the network weights is investigated. Simulation studies for a flexible rotating disc show that the control scheme proposed provides fast vibration suppression while maintaining robust stability.