1: \begin{abstract}

2: We propose a robust adaptive Model Predictive Control (MPC)

3: strategy with online set-based estimation for constrained

4: linear systems with unknown parameters and bounded

5: disturbances. A sample-based test applied to predicted

6: trajectories is used to ensure convergence of parameter

7: estimates by enforcing a persistence of excitation

8: condition on the closed loop system. The control law

9: robustly satisfies constraints and has guarantees of

10: feasibility and input-to-state stability. Convergence of

11: parameter set estimates to the actual system parameter

12: vector is guaranteed under conditions on reachability and

13: tightness of disturbance bounds.

14: %   suitable conditions on the system matrices, linear feedback gain and the additional excitation into the control input are satisfied. 

15:  

16: 

17: %For robust adaptive Model Predictive Control schemes with set-based estimators, we analyse the rate of convergence of parameter estimates and introduce constraints to ensure asymptotic parameter convergence. For constrained linear systems with bounded disturbances we design computationally tractable constraints constraints that enforce persistence of excitation conditions on the solution of a receding horizon optimization problem.

18: %These constraints are constructed so that recursive feasibility of the receding horizon optimization is ensured. 

19: %Methods of ensuring persistence of excitation that either exploit the random nature of disturbance inputs inherent in the model or introduce additional excitation into the control input are compared in terms of parameter convergence and performance.

20: %%

21:   

22: \noindent \textbf{Keywords:} Predictive control, Recursive identification, randomised algorithms.

23: \end{abstract}

24: