Offset-free adaptive nonlinear model predictive control for pneumatic servo system

Abstract

This paper presents an adaptive nonlinear model predictive control method with zero steady-state error (called offset free) in the presence of the plant-model mismatch and external disturbances. A neural network model is trained online to predict the process output recursively over the prediction horizon. The output of the neural network is modified by the current output prediction error to achieve offset-free model predictive control method. The stability of the closed-loop system is shown using the Lyapunov direct method. Simulation results on a pneumatic servo system show effectiveness of the control strategy as compared with the recently reported methods in literature under plant-model mismatches and unmeasured disturbances.