Online Diagnosis of Systems with Rectangular Hybrid Automata Models

Abstract

We propose an online diagnosis approach for a class of hybrid systems. The normal and the faulty behaviors of the system are modeled with rectangular hybrid automata. Our approach is based on the use of a diagnosis procedure which performs, online, an estimation of the system states, within a given time window, and based on the current record of observable timed events. Each new estimation can be triggered either, by a new event observation, or simply by the elapse of time. We give examples to illustrate the use of our hybrid systems diagnosis approach.