Machine learning-assisted metal additive manufacturing has been widely applied in performance optimization and control, such as process parameter optimization, structural optimization of formed parts, defect identification, and mechanical property prediction. However, the multi-physics phenomena and closed-loop control involved in the forming process have not yet been thoroughly investigated.
This review provides an in-depth analysis of multi-physics fields (temperature field, fluid dynamics, stress/strain field) and systematically examines how machine learning is used to understand the fundamental physical mechanisms underlying quality control in metal additive manufacturing. It elaborates on leveraging insights from machine learning to optimize key quality attributes, including defect suppression, geometric fidelity, and material property tailoring. Furthermore, the review clarifies the relationship between machine learning-based prediction/control of multi-physics fields and the optimization of final quality. It also focuses on machine learning-driven real-time closed-loop control, highlighting how cross-scale coordination, multimodal data fusion, and feedback optimization enable full-process regulation, paving the way for integrating advanced technologies such as digital twins and edge computing, and outlining key future research directions.
The work entitled “ Machine Learning A ssisted Q uality Control in Metal Additive Manufacturing: A Review ” was published in Advanced Powder Materials (Available online on 16 September 2025).
Advanced Powder Materials
Experimental study
Not applicable
Machine learning assisted quality control in metal additive manufacturing: a review
16-Sep-2025