DIAGNOSABILITY OF FAIR DISCRETE EVENT SYSTEMS

Failure diagnosability has been widely Studied using discrete event system (DES) models. It is, however, shown in this work by means of a counterexample that the diagnosability condition, which has been shown to be necessary and Sufficient ill the DES context, fails to hold for many real-world hybrid systems. This is because the abstraction employed in formulating the DES models obliterates the Continuous dynamics. In the present work, a new failure diagnosability mechanism has been developed for discrete time hybrid system (DTHS) models to alleviate this problem. A new diagnosability condition is proposed and its necessity and sufficiency with respect to the diagnosability definition are established formally. Finally, the method of A-diagnosability, which call also be used to Circumvent this problem and which needs additional probabilistic information for diagnosability analysis, has been shown to have a higher computational complexity than the DTHS model based method proposed in this paper. Further, it is also highlighted that the DTHS model based diagnosability analysis technique is capable of diagnosing faults that degrade the temporal performance of the system, which cannot be handled by the A-diagnosability analysis mechanism.