Observer based on-line fault diagnosis of continuous systems modeled as Petri nets

This paper describes a technique for achieving on-line fault diagnosis in continuous systems that are modeled using Petri nets. The effect of place markings and transition markings are considered and based on the computed error between the initial marking and subsequent markings evolved in time, the faults are categorized assuming that the markings are both observable and unobservable. An algorithm has been suitably proposed for achieving detection of faults for a typical continuous three tank system along with suitable results.