Putting more numbers in the qualitative simulator CA-EN

This paper presents the qualitative simulator CA-EN (CAusal ENgine). CA-EN is a causal and constraint-based reasoning system for real-time applications whose first version was devised and implemented for the biotechnology application domain to be part of a model-based supervision system. Having considered to use CA-EN in the framework of the Esprit project TIGER as part of a model-based diagnosis application in the gas turbine domain, we present the limitations of the first version, and the improvement we have made in order to cope with the new application domain too. An application example of prediction in the fuel system of a 28 Mega-Watt GE Frame 5 gas turbine (one of the TIGER project application system) is presented to illustrate the extended formalism which has been proved well-suited for fault detection and the estimation of nonmeasurable variables