Sistema Fuzzy para Detecção de Novidades em Inspeção Não Destrutiva por Ultrassom

In this paper, Fuzzy decision rules, in the context of novelty detection, aims to increase efficiency and reliability related to defect identification associated with a set of A-Scan signals, obtained from a welded test piece in a non-destructive inspection using the Time-of-Flight Diffraction (ToFD) ultrasound technique. The training process only required samples without defects, and the result consisted in assigning a probability that a given analyzed signal was related to some non-conformity. ToFD ultrasonic signals from five types of defects indicated the method's efficiency.