NONLINEAR MODEL PREDICTIVE CONTROL WITH FULL DEAD-TIME COMPENSATION

A nonlinear model predictive control via direct single-shooting, using a full dead-time compensation structure of the filtered Smith predictor in the controller cost function, is presented and applied to the benchmark Van de Vusse reactor modified by different dead-times incorporated in the input and output of the process variables. Simulation results provide insights into improvements in the controller performance and robustness when applied in systems with dominant dead-time.