Author: Dr. Harsha Kumar M.
Abstract: Real time SLM monitoring usually relies on photodiodes or cameras, yet acoustic signals offer deeper penetration into the melt pool. We deploy piezoelectric arrays to capture ultrasonic bursts, feeding them to a convolutional neural network that predicts porosity with 91 % F1 score. Hardware does not intrude on the build chamber, eliminating optical path contamination issues.
Keywords: Selective laser melting; Acoustic emission; In process monitoring; Convolutional neural network; Porosity prediction
Full Issue
| View or download the full issue | PDF 21-28 |