A multirate adaptive composite controller for flexible-link robots using neural networks

Abstract

A multirate adaptive composite controller using neural networks is presented in this paper for the trajectory tracking control of a flexible-link robot with a partially known nonlinear dynamics. Based on the singular perturbation method and two time-scale decomposition, the flexible-link robot model is decomposed into two subsystems: the slow subsystem and the fast subsystem. Thus, separate slow and fast control laws can be designed for each subsystem and then combined into a composite control. The slow control is implemented by a neural network-based adaptive controller with novel properties and structure, which is used to control the slow subsystem, an equivalent rigid-link manipulator. The fast control is designed to stabilize the fast subsystem around the equilibrium trajectory set up by the slow subsystem under the effect of the slow control.