A multi-step model predictive control method based on recurrent BP neural network

Model Predictive Control is a kind of computer control method which was widely applied in industry process control. The classic model predictive controllers are all based on linear predictive model, and unfit for the objects which have strong nonlinearity and several set-points. This paper used recurrent BP neural network to construct a nonlinear multi-step predictive model, and designed the optimization strategy to form a nonlinear multi-step model predictive controller with constraints. The simulation result show that the model predictive controller proposed in this paper can trace several set-points perfectly.