Robust Model Predictive Control of Systems by Modeling Mismatched Uncertainty

This study addresses to the robustness of model predictive control in the presence of the mismatched uncertainty, e.g. disturbance, noise and parameter variations. Model predictive control is solved online and its control action is fed to the real system with the additional control action that is required to maintain the controlled trajectories in a simple uncertainty tube in practice where the center of the aforementioned tube is the trajectory of the nominal model. For this purpose, a sliding mode controller as variable control structure is designed taking the difference between the real system and nominal system into consideration. The stability of the overall system is proven taking the modeling error on the uncertainty model into account.