NARMAX based nonlinear predictive control

This paper proposes a novel NARMAX based nonlinear predictive controller synthesis methodology. In this approach the nonlinear predictors are identified directly from experimental data so that the model of the system does not need to be known in advance. The main advantage of the proposed approach is that it avoids time consuming numerical optimisation algorithms that are usually associated with conventional nonlinear predictive control strategies. The proposed design method can deal effectively with noisy measurements. In this context, a procedure for deriving the observer polynomial T is included. Numerical simulations are given to demonstrate the effectiveness and robustness of the proposed approach.