Hybrid systems Diagnosis using modified particle Petri nets

International audienceThis paper presents an approach of model-based diagnosis for the health monitoring of hybrid systems. These systems have both continuous and discrete dynamics. Modified Particle Petri Nets, initially defined in the context of hybrid systems mission monitoring, are extended to estimate the health state of hybrid systems. This formalism takes into account both uncertainties about the system knowledge and about diagnosis results. A generation of a diagnoser is proposed to track on-line the system health state under uncertainties by using particle filter. An academic example is used all along the article to illustrate the major concepts