Defect Detection and Monitoring in Metal Additive Manufactured Parts through Deep Learning of Spatially Resolved Acoustic Spectroscopy Signals

What is it about?

We devised an artificial intelligence approach to detect defects in 3D printed metal parts using signals obtained from a novel type of non-contact and nondestructive sensor called spatially resolved acoustic spectroscopy (SRAS). We show that defects in 3D printed parts, such as porosity, that are not easy to spot with the naked eye from the raw SRAS signal are clearly detected using our approach. This is practically valuable, because, if defects remain uncorrected, the 3D printed part can fail prematurely. Using the findings of our research, operators can quickly identify defects and hence take the appropriate corrective action during the fabrication process. Hence, this research can potentially accelerate the use of 3D printed parts in strategic industries, such as aerospace where safety is paramount.

Featured Image