Robust Composite Saturation Compensation for a Single Flexible Link Using Neural Networks

Abstract

A robust neural network (NN) composite saturation compensation scheme is presented for a trajectory tracking and a vibration suppression of a flexible link robot arm. The scheme is based on a singular-perturbation technique and can accommodate an unknown disturbance and the saturation constraints. The saturation compensator is composed of a robust fast sub-compensator for stabilization of the fast, flexible modes and a slow sub-compensator for a stable tracking control of the slow, rigid modes. The actuator saturation is assumed unknown. A NN-based slow subcompensator is inserted into the feedforward path. The stability analysis is based on the Lyapunov theory. Simulation results indicate that the proposed scheme can effectively compensate the saturation nonlinearity for underactuated systems