EXTENDED DMC PREDICTIVE CONTROL USING NEURAL NETWORK MODEL

The paper uses feedforward neural networks for modeling a nonlinear system to obtain its nonlinear predictive model and get its linear average model from multi-step responses. We apply the linear average model to rolling optimization, and adjust the average model with the nonlinear predictive model to implement nonlinear predictive control. The simulation results demonstrate the good performance and high real-time property of the proposed control strategy.