In Situ X-Ray Imaging and Machine Learning in Ultrasonic Field-Assisted Laser-Based Additive Manufacturing: A Review.

Area of Science:

• Materials Science
• Manufacturing Engineering
• Physics

Background:

• Metal additive manufacturing (AM) faces challenges with defects like cracks and pores due to rapid melting and solidification.
• Ultrasonic field-assisted laser-based additive manufacturing (UF-LBAM) shows promise in mitigating these defects by manipulating melt pool dynamics.
• Understanding the complex physics of UF-LBAM, especially under ultrasonic excitation, is crucial for its practical application.

Purpose of the Study:

• To systematically review X-ray based fundamental studies in UF-LBAM.
• To explore the applications of machine learning (ML) in real-time monitoring and defect prediction for UF-LBAM.
• To identify challenges and future directions for advancing UF-LBAM technology.

Main Methods:

• Systematic literature review focusing on X-ray based studies and ML applications in UF-LBAM.
• Analysis of synchrotron X-ray revealed physical phenomena during the UF-LBAM process.
• Evaluation of ML models for real-time monitoring, defect prediction, and control.

Main Results:

• X-ray studies have elucidated physical phenomena governing melt pool dynamics in UF-LBAM.
• Machine learning models demonstrate potential for real-time defect detection and prediction.
• Advances in understanding and ML application pave the way for industrial implementation of UF-LBAM.

Conclusions:

• Despite progress, fundamental physics gaps and ML model transferability remain challenges.
• Future research should focus on physics-informed models, multimodal diagnostics, and closed-loop control.
• These advancements are key to unlocking UF-LBAM's potential for high-performance metal component fabrication.