Finite horizon robust model predictive control with terminal cost constraints

A finite horizon model predictive control (MPC) algorithm that is robust to modelling uncertainties is developed along with the construction of a moving average system matrix to capture modelling uncertainties and facilitate the future output prediction. The authors´ main focus is on the step tracking problem. Using linear matrix inequality techniques, the design is converted into a semi-definite optimisation problem. Closed-loop stability, known to be one of the most challenging topics in finite horizon MPC, is treated by adding extra terminal cost constraints in the semi-definite optimisation. A simulation example demonstrates that the approach can be useful for practical applications.