2009 IEEE International Conference on
Systems, Man, and Cybernetics |
![]() |
Abstract
In classical consistency based approaches used for diagnosis of discrete event systems (DES), an inconsistency between the behaviors described in the reference model and the behavior observed through the sensors detects the occurrence of a failure in the system. If the model is not considered as error free, this traditional interpretation of inconsistencies must be enhanced to include the model errors as a potential responsible for the inconsistency. This paper has two objectives. First, it proposes a mechanism able to discriminate among inconsistencies the ones due to modeling errors an those caused by failures. Second, it describes a method to restore the consistency between the models and the observations.