Caracterização de sinais de emissão acústica em ensaios CTOD por análise fractal

Abstract

The research had the objective of correlating different defects and their criticalities in specimens of ASTM A516 grade 60 carbon steel for CTOD tests, whose notches were made in the base metal, in the weld metal and in the thermally affected zone of a welded joint, with the acoustic emission signals obtained during the test. Four methods of fractal analysis were used: rescaled Hurst analysis (R/S), boxcounting dimension, minimal cover dimension an detrended fluctuation analysis (DFA). A minimal number of points necessary for obtaining the exponents associated with each of the different filtering methods on these exponents. It has been verified that none of the methods alone has been able to characterize either the defects, or the loading regimes. However, by using the bi-dimensional representations of all the exponents, together with the standard deviation of the signals, it has been possible to establish satisfactorily a hierarchical characterization of the defects and the loading regimes.