A new predictive scheme is proposed for the control of Linear Time Invariant (LTI) systems with a constant and known delay in the input and unknown disturbances. It has been achieved to include disturbances effect in the prediction even though there are completely unknown. The Artstein reduction is then revisited thanks to the computation of this new prediction. An extensive comparison with the standard scheme is presented throughout the article. It is proved that the new scheme leads to feedback controllers that are able to reject perfectly constant disturbances. For time-varying ones, a better attenuation is achieved for a wide range of perturbations and for both linear and nonlinear controllers. A criterion is given to characterize this class of perturbations. Finally, some simulations illustrate the results.

International audienceIt is well-known that standard predictive techniques are not very robust to parameter uncertainties and to external disturbances. Furthermore, they require the exact knowledge of the delay. In practice, these constraints are rarely satisfied. In this paper, solutions are presented to allow the use of predictive control in presence of external disturbances, parameter uncertainties and an unknown input delay. First, a recent predictive control method developed to attenuate the effect of external disturbances is shown to be also robust to parameter uncertainties. In addition, a delay estimator is presented to estimate unknown time-varying delays. Theoretical results are widely illustrated through experimental tests on a DC motor