A combined analytical and knowledge based method for fault detection and isolation

Fault detection and isolation (FDI) methods based on analytical and qualitative models play an important task in supervision and modern automatic control. There are two important steps in FDI: residual generation and residual evaluation. In the first step, several analytical methods are used, the process characteristics play an important role in the choice of the method. The second step is a decision making problem. The methods of qualitative reasoning are more and more used. In this paper a combined analytical and knowledge based method for fault detection and isolation is presented. The residuals are generated using a set of adaptive observers. For residuals evaluation behavioural models (under the form of a decision tree) are extracted by means of a classification technique. This method is illustrated by a simulation example of a biotechnological process.