Defect Detection Based on Analysis of Optical Images Taken during Additive Printing

A solution to the problem of defect detection in selective laser melting (SLM) is considered; it is based on an analysis of optical images of each layer and subsequent additional verification using computed tomography (CT) data. For calibrating informative features, a 10 × 10 × 10-mm sample with the following artificially formed defects of two types is prepared: pores (spherical regions) and cracks (elongated cylindrical regions) located at different heights and having different spatial orientations. The defects are formed by setting different laser power levels in the simulated defect volume, in particular 0 and 50% of the nominal power. In addition, natural defects, i.e., the defects that appear during SLM, are analyzed. The local zones artificially formed at 0% laser power are shown to reproduce the initial geometry of the defect simulators (unmelted regions), and 50% power radiation forms partially melted regions with blurred boundaries and a distorted shape. Based on CT results, spherical defects 1 and 2 mm in diameter and cylindrical crack simulators can be reliably identified. A procedure for juxtaposing optical and CT data and a method for estimating defect localization and detection, which is suitable for adjusting printing parameters and for implementing in the practice of operational diagnostics of SLM processes, are proposed.