Robust predictive control combined with an adaptive mechanism for constrained uncertain systems subject to disturbances

Abstract: This paper proposes a discrete-time adaptive model predictive control (MPC) algorithm for a class of constrained linear time-invariant systems subject to state-dependent disturbances, which updates the estimate of uncertain system parameters on-line and produces the control input ensuring the constraint fulfillment. This method is based on an adaptive mechanism and a robust MPC algorithm using the comparison model which enables to estimate the future prediction error bound. First, the parameter estimation method for MPC based on the moving horizon estimation is introduced. It allows to predict explicitly the worst-case estimation error bound over the prediction horizon. Second, we propose an adaptive-type MPC strategy developed by combining an on-line parameter estimator with a robust MPC method based on the modified comparison model. The MPC controller designed in this way guarantees constraint fulfillment, closed-loop stability and feasibility in the presence of uncertain system parameters and state-dependent disturbances. A numerical example demonstrates the unique features of the proposed approach.