Robust Smith Predictor (RSP)

As is well known, control of industrial processes with significant dead time by using the conventional Smith Predictor (SP) results better performance in comparison to PI controllers. The SP has a main disadvantage which is a lack of robustness to variations in process parameters. To eliminate this deficiency, this paper presents the Robust Smith Predictor which is based on a modified disturbance observer to estimate disturbance and a filter instead of the Smith controller. The conducted simulations show that the proposed Robust Smith Predictor can eliminate the Smith predictor deficiency in robustness and provide an elegant trade-off between disturbance rejection and robustness.