Review: Machine Learning Aids Metal Additive Manufacturing Quality Control

A comprehensive review paper has been published, detailing the significant advancements and potential of employing machine learning (ML) techniques to bolster quality control within metal additive manufacturing (MAM). The research synthesizes current findings on how ML algorithms can be integrated to monitor, predict, and mitigate defects during the MAM process.

The paper highlights that ML models are being developed to analyze vast datasets generated during printing, including sensor data, process parameters, and post-print inspection results. By learning from this data, ML can identify subtle anomalies that might indicate potential flaws, such as porosity, cracking, or lack of fusion, often before they become critical issues.

Furthermore, the review discusses various ML approaches being utilized, ranging from supervised learning for defect classification to unsupervised learning for anomaly detection. It also touches upon the challenges, including the need for standardized datasets and the interpretability of ML model predictions in real-time manufacturing environments. The integration aims to move beyond traditional offline inspection methods towards more proactive, in-situ quality assurance.

This research underscores a growing trend in additive manufacturing towards intelligent process control. By leveraging ML, manufacturers can potentially achieve higher part reliability, reduce material waste, and accelerate the adoption of MAM for critical applications where quality assurance is paramount.