Hybrid control based on discrete-event automata and receding-horizon neural controllers

A hybrid receding-horizon control scheme for nonlinear discrete-time systems is addressed in this work. Such a scheme is composed of two control levels: a continuous level characterized by a finite set of neural receding-horizon feedback control laws, and a discrete-event level aimed at choosing the best control action to be applied to the plant, depending on the current system conditions and on possible occurred external events. The two-level scheme presents two major innovative aspects: first, a new class of hybrid automata, namely the discrete-time discrete-event automata, is used for the modeling of the proposed hybrid control scheme. Moreover, receding-horizon regulators based on neural approximators and off-line determined can be adopted at the continuous level. The stability analysis of the hybrid control system is addressed both in the case in which optimal receding-horizon feedback control functions are used and in the case in which neural approximate regulators are adopted