Answers are grounded in this paper and cite their sources.
Data-driven process monitoring is an area of growing interest in additive manufacturing processes. This work presents a comprehensive study on Laser Powder Bed Fusion (PBF-LB) process data gathered from a melt pool emissions monitoring system. The first approach consists of aggregating statistical metrics for each layer, creating time series that reveal the evolving patterns and dynamics throughout the printing sequence. Machine Learning (ML) techniques are employed to identify deviations within manufactured parts. Subsequently, a stripe-level approach is carried out, analyzing the influence of different variables on the final quality of produced parts. A regression model based on these variables is developed to forecast the emission level of layers yet to be produced. This model presents a Mean Absolute Error (MAE) close to 0.05, which demonstrates its suitability to the identification of unusual patterns in the typical emissions performance, including the possibility to better understand the process by performing feature importance extraction.
No takes yet. Share an insight, caveat, or question.
Barrutia et al. (2024) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: