Compound quality assessment in laser beam melting processes using layer images

Laser beam melting (LBM) is an additive manufacturing (AM) technology that allows the layer-based production of geometrically complex parts from metal powder. Non-destructive evaluation of part quality is a pre-requisite for widespread application of this promising technology and an ongoing research topic. In-process measurement of part layer compound quality will be beneficial during optimization of process parameters for best part quality. We inspect the part surface in high resolution layer images acquired after laser exposure. Based on a fast Fourier transform (FFT) we decompose surface images into laser scan lines (oriented component) and underlying powder structure (not oriented component). By computing the signal energy ratio we obtain a measure for the prominence of laser scan lines which is correlated with the compound quality. For 25 parts with varied laser power and scan velocity the signal energy ratio is compared to the computed energy input. A linear model achieves a root-mean-square error (RMSE) of 9.76Jmm
⁻³
. Additionally we automatically classify part regions as good/not good and compare our results to a manual selection based on light microscopy.