A subspace predictive control method for partially unknown systems with parameter learning event-triggered law

This paper proposes a novel event-triggered subspace predictive control (SPC) method for a class of linear discrete-time partially unknown systems. Without the complete system parameter information, the design parameters of the event-triggered law are first derived via system data by the reinforcement learning method. The proposed event-triggered law depends on the defined input error and the state-dependent threshold. The receding horizon principle in the typical predictive control methods is substituted by the event-triggered law, which can ensure the stability of the predictive input with optimality. The proposed method can considerably reduce the data computation and transmission load of the conventional SPC methods. The simulation results illustrate the effect and the satisfactory performance of the proposed method.