Constraint Databases Technology for Polynomial Models Diagnosis

Abstract

Model-based Diagnosis allows the identification of the parts which fail in a system. The models are based on the knowledge of the system to diagnose, and they can be represented by constraints associated to the components. The variables including in these con straints can be observable or non-observable, depending on the situa tion of sensors. In order to obtain the minimal diagnosis in a system, an important issue is related to find out the minimal possible conflicts in an efficient way. We consider that Constraint Databases represent an excellent approach in order to solve this problem in complex sys tems, where a tuple in a relational database could be replaced by a conjunction of constraints. In this work we have used a novel logical architecture of Con straint Databases which has allowed us to obtain these possible min imal conflicts by means of a standard query language though the in formation is stored in a conventional relational database. Moreover, we have considered Grobner bases as a projection operator to obtain ¨ the minimal possible conflicts of a system.