A fuzzy classifier for fault diagnosis of valve actuators

This paper describes a fuzzy classifier for fault diagnosis of valve actuators. The belonging of a feature vector to the normal class and/or a faulty class of system behavior is described using fuzzy sets. The actual diagnostic task is performed using the fuzzy membership functions for the feature vector to be classified. The theoretical aspects of the classifier are presented. Then the case study-the DAMADICS benchmark flow control valve, is shortly introduced and also the method used to generate the data for designing and testing the classifier. Finally, the simulation results are discussed.