Design of a Backstepping Controller based on an Adaptive Elman Neural Network for a Two-Link Robot System

Abstract

This paper presents a backstepping controller based on an adaptive Elman neural network (BSAENN) to solve the mismatched uncertainty problem of underactuated robotic systems to compensate for the perturbations of nonlinear system. First, the nonlinear dynamical equations of the robot system are transformed to a cascade form. Second, an adaptive backstepping controller has been established. This controller is adopted using the combination of the adaptive Elman neural network (AENN) and the traditional backstepping control (TBS) approach. The AENN is used to approximate the uncertainties and enhance the control behavior against uncertainties. The adaptation laws of the AENN are deduced using Lyapunove stability. Computer simulations, compared to traditional controllers (PID and TBS), show that the adopted control algorithm results in robustness for trajectory tracking performance under the occurrence of uncertainties.