Abstract
Two communities are working in parallel in model-based diagnosis: FDI and DX. This paper proposes a new methodology, based on FDI and DX techniques, in order to automate the diagnosis process. According to this approach, only relevant information for identifying the minimal diagnoses is used in order to reduce the computational complexity. All the possible minimal diagnoses are represented as a signature matrix and a set of rules, which are calculated only once for all the observational models. The minimal diagnosis is obtained by evaluating the set of rules and the signature matrix, only this step is on-line, the rest of the process can be done off-line.