Abstract
This paper investigates the problem of diagnosing discrete faults for a class of hybrid dynamical systems, where the evolution of the discrete state is led by a set of discrete modes governed by a partially observed Petri net which means that we can know the activation of some discrete modes and the firing of some switching conditions. The considered class may have several modes evolving simultaneously. The modes can be associated with normal or faulty operating conditions. Therefore, the knowledge of the current mode represents the coeur part of the proposed diagnosis approach. We present at first an asymptotic observer to detect both the activation of discrete modes and the switching conditions. Then, we propose an algorithm to diagnose faults using the estimated discrete states provided by the observer. Simulation results show the effectiveness of the proposed approach.