Application of the STRAY statistical learning algorithm for the evaluation of in-situ process monitoring data during L-PBF additive manufacturing.

Abstract This study investigates the use of a statistical anomaly detection method to analyse in-situ process monitoring data obtained during the Laser-Powder Bed Fusion of Ti-6Al-4V parts. The printing study was carried out on a Renishaw 500M Laser-Powder Bed Fusion system. A photodiode-based system called InfiniAM was used to monitor the melt-pool emissions along with the operational behaviour of the laser during the build process. The analysis of the in-process data was carried out using an unsupervised machine learning approach called the Search and TRace AnomalY algorithm. The ability to detect defects during the manufacturing of metal alloy parts was demonstrated.