Model-based Fault Diagnosis

Paul Frank, University of Duisburg, Germany

The paper gives a survey of the basic concepts, the state of the art, and the latest developments in the theory of fault diagnosis of technical processes using mathematical and knowledge-based process models. Emphasis is placed on the basic approaches using analytical (quantitative) models. Finally we briefly discuss the problem of combining the analytical approach with the knowledge-based approach using fuzzy logic and neural nets for decision making, and we suggest a fault detection and isolation (FDI) expert system integrating both approaches for residual generation with either artificial or natural intelligence for decision making.