Constrained data-driven RMPC with guaranteed stability

Being able to tackle constraints, the adaptive model predictive control (AMPC) is one of the most frequently used methodologies in coping with model uncertainties. Fundamentally, the core procedure of AMPC is dependent on the model identification to increase the suboptimality of MPC. However, only the type of data which admits the specified requirements is accessible to the identification, which implies the failure of adaptation method to some extent. Aimed at this proposal, this paper proposes a new control technique which adds data-driven representation skills into the robust model predictive controller (RMPC). The advantages over other controllers were explicitly reflected in numerical simulation. Feasibility of the algorithm is able to be proven and the stability of the controlled system is verified.