Molten pool dynamics

This research from Shanghai Jiao Tong University proposes a novel Fusiform Attention Network (FANet) for the real-time visual monitoring of penetration status in laser welding. By integrating Ternary Multi-head Linear Attention (TMLA), the model effectively aligns its receptive field with the physical geometry of the molten pool, balancing local detail extraction (spatter, plasma) with long-range morphological perception (molten pool length). The network achieves a high recognition accuracy of 93.24% with an end-to-end latency of only 5.41ms, meeting the stringent requirements for high-speed online industrial monitoring (100 FPS) and providing a robust solution for closed-loop quality control in automotive, aerospace, and shipbuilding industries.

The article reviews the latest developments and future prospects of beam shaping technology in laser welding. It discusses how different beam morphologies—such as Gaussian, elliptical, flat-top, and dual-mode beams—control molten pool dynamics to reduce defects like porosity and spatter, especially in high-reflectivity materials like aluminum and copper.

This study introduces an in-situ monitoring method for laser processing based on specular reflection of the molten pool surface, enabling high-resolution observation and feature extraction of molten pool dynamics. By analyzing the molten pool surface, keyhole opening, and specular micro-region under varying laser powers, it identifies the specular micro-region as the most sensitive and representative parameter due to its strong periodic behavior linked to laser–keyhole interactions. The findings provide a novel approach for understanding energy transfer mechanisms and offer a reliable foundation for real-time monitoring and intelligent control in laser manufacturing processes.

This study demonstrates that combining nitrogen shielding gas with a reduced laser spot size can effectively suppress humping and root sagging defects in high-speed laser welding of stainless steel foil. By stabilizing molten pool dynamics, reducing oxygen content, and optimizing surface tension gradients, the approach significantly improves weld quality and mechanical performance, achieving defect-free welding at higher speeds while providing a predictive model for process optimization.

This study explores the application of circular laser beam oscillation in narrow-gap welding of 5A06 aluminum alloy thick plates, demonstrating that oscillation significantly improves molten pool stability, modifies flow behavior, and enhances sidewall fusion. By combining experimental validation with numerical modeling, the research confirms that circular oscillation effectively suppresses lack-of-fusion defects, reduces plasma plume energy loss, and improves weld quality, providing a reliable solution for high-performance thick plate laser welding applications.

This study systematically reviews the principles, modeling methods, and solidification mechanisms of laser wobble welding (LWW). Through analysis of heat transfer, molten pool dynamics, and multi-physics simulations, it reveals how beam oscillation stabilizes welding and refines microstructure. The paper highlights current research gaps—particularly in solidification modeling—and proposes directions for developing high-fidelity, data-driven models to enhance weld quality and industrial application.