Improved identifier to build fuzzy models for qualitative fault diagnosis systems

The work suggests some arguments to develop a general structure for a FDI system. Further a possible identifier to be included in this structure is approached. For this kind of identifier we combine a clustering stage followed by a given technique (the Wang´s approach as a case study). The such obtained models are evaluated on the standard experimental Box-Jenkins´ data set and some practical conclusions able to improve the identification approaches with fuzzy models are formulated.