Optimal selection of subsystems for synchronous diagnosis
Keywords: Fault Diagnosis, Diagnosability, Discrete Event Systems, Synchronous Diagnosis, Verifiers
AbstractFault diagnosis of automated systems is a very important task, since faults can alter the expected behavior of systems, damaging equipment and bringing risk to operators. In general, systems are composed of several subsystems or modules, and therefore, the complete system model may grow exponentially with the number of system components. This fact shows that a large amount of memory space may be needed to implement diagnosers computed using traditional methods, since they are based on the complete system model. Recently, a new method for fault diagnosis, called synchronous diagnosis, has been proposed. The diagnoser computed using this method is based on the state observers of the fault-free component models of the system, avoiding the implementation of the state observer of the composed system model. In the synchronous diagnosis strategy it is supposed that all fault-free subsystem models are used to detect the fault occurrence. However, in practice, some subsystems may not add useful information regarding the fault occurrence, or the same information can be obtained from the other modules, which shows that these subsystems are not necessary in the synchronous diagnosis scheme. In this paper, an algorithm for computing all minimal sets of modules that ensure the synchronous diagnosability of a Discrete Event System is proposed. The performance of the proposed algorithm is compared with the performance of the exhaustive search method, and we show that using the proposed method there is a significant reduction in the computational cost of finding all minimal sets of modules that ensure synchronous diagnosability.