Diagnostic reasoning framework combining fuzzy logic and dempster-shafer theory

In this paper we present a method of diagnostic reasoning in the case of imprecise values of diagnostic signals, caused by noise, disturbances and modeling errors. Imprecise values can be handled by fuzzy logic, but fuzzy inference cannot deal with a priori probabilities of faults. Therefore we propose to use Dempster combination rule to join information provided by a fuzzy diagnostic signals and information about the faults from the other sources. Results are exemplified on a three tank system example with known a priori fault probabilities.