Acoustic Emissions Detection and Ranging of Cracks in Metal Tanks Using Deep Learning

摘要

This work proposes a new method for the estimation of the distance of cracks in pressure metal tanks. This method is obtained coupling the acoustic emissions analysis and the deep learning techniques. Using a 2D CNN we are able to estimate the distance between a crack and an acoustic emission piezoelectric sensor. The CNN is trained on images representing the spectrogram of acoustic emission located at distances of 2, 20,40, 60, 80, 100, 120 and 140 cm. We obtained a RMSE of 2.54 cm.