Hybrid manufacturing line supervision and diagnosis by means of fuzzy rules connected with a causal graph

The method proposed here consists in generating reactive knowledge represented by fuzzy rules from a given causal graph model. Causal reasoning forms a practical support for model-based diagnosis. Also the fuzzy logic allows us to generate the corrective actions for a qualitative model-based system (supervision). These two kinds of reasoning have been integrated in a computer system for a real world application. A global model based on a causal graph of the process is used for the diagnostic, but local fuzzy reasoning blocks are used for supervision.