Sensor fusion of pyrometry and acoustic measurements for localized keyhole pore identification in laser powder bed fusion

In-situ process monitoring as an aid to part qualification for laser powder-bed fusion (L-PBF) technology is a topic of increasing interest to the additive manufacturing community. In this work, airborne acoustic and inline py-rometry measurements were recorded simultaneously with the laser position. X-ray radiography imaging was used to spatio-temporally register keyhole pore locations to the pyrometry and acoustic signals, enabling binary labeling of data partitions based on pore formation. These labeled partitions were subsequently featurized using a highly comparative time-series analysis toolbox. Acoustic data was found to be much more effective than the pyrometry data for keyhole pore identification. However, when the information contained in both sensing modalities was combined in a sensor fusion strategy, the error rates of the top performing models were signif-icantly reduced.