Event-triggered H-infinity control for unknown continuous-time linear systems using Q-learning
Kyriakos G. Vamvoudakis, Henrique Ferraz · 2016
In this paper we formulate the infinite-horizon game-theoretic problem as an adaptive learning one, so that the optimal performance is guaranteed when the continuous sampling of the state is relaxed by using an event-triggering condition. Then, a Q-learning framework combined with an actor/critic approach approximates the optimal cost, the optimal control, and worst case disturbance without any knowledge of the system model. The overall closed-loop system is modeled as an impulsive system and the asymptotic stability of its equilibrium is proved. A simulation with a numerical example is used to illustrate the results.