NEURO ADAPTIVE CONTROL BASED ON MODEL PREDICTIVE CONTROL FOR ROBOT SYSTEMS

Abstract

This article proposes a neural adaptive controller based on a predictive model (MPC) for a single robot system, with an uncertain dynamic model, and considers input constraints in the controller design. In the proposed neuro-based MPC structure, a radial basis function neural networks (RBFNNs) are employed for modeling. The neural network is utilized as a predictive model for the robot system to handle the system uncertainties. Additionally, input constraints are guaranteed by using a non-quadratic cost function for neural network-based MPC. Simulation studies are conducted to verify the effectiveness of the proposed approach.