Stable hybrid control based on discrete-event automata and receding-horizon neural regulators

A hybrid control scheme for nonlinear discrete-time systems is addressed. S uch a scheme is composed of two control levels: a continuous level characte rized 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 app lied to the plant, depending on the current system conditions and on extern al events that may have occurred. The two-level scheme presents two major i nnovative aspects. first, a new class of hybrid automats, namely discrete-t ime discrete-event automata. is used for the modeling of the proposed hybri d control scheme. Secondly, receding-horizon regulators are used that are b ased on neural approximators, at the continuous level. In the paper, the st ability analysis of the proposed hybrid control system is carried out, and its practical applicability is shown For a case study relevant to traffic c ontrol on freeways. The example is very significant, if one takes into acco unt the complexity of the considered transportation system and also the fac t that the reported simulation results are based on real-traffic data, and hence they well represent critical traffic conditions on freeways. (C) 2001 Elsevier Science Ltd. All rights reserved.