Robust adaptive control of flexible link manipulators using multilayer perceptron

Abstract

This paper presents a robust adaptive control method based on neural networks for flexible link manipulators which is a nonlinear non-minimum phase system. The development of the control method comprises of two steps First, an appropriate reference signal is designed such that the internal dynamic subsystem become input-to-state practical stable. Then an output feedback control, which does not rely on the state estimation, is designed such that the output of system asymptotically tracks this reference signal. This controller is comprised of a dynamic linear controller, an adaptive neural network and a discontinuous robustifying term. Stability of the overall system is proved using the small gain theorem.