Visual Detection of TIG Welding Molten Pool Shape Parameters

Literature Overview

Published in Acta Metallurgic Sinica (Vol. 36, No. 12, 2000, pp. 1284-1288), this paper by Gao Jinqiang and Wu Chuansong from the Institute of Joining Technology, Shandong University of Technology, presents a visual detection system for TIG welding molten pool shape parameters. Funded by the National Natural Science Foundation of China (59875053), this early but pioneering work established the foundation for in-situ welding process monitoring through image-based techniques.

Core Technical Findings

The research developed a visual sensing system based on arc spectrum analysis and composite filter lens design. The system consists of a CCD camera equipped with a custom composite filter lens, enabling real-time observation of the molten pool from the front of the workpiece.

System Component	Description	Function
Composite Filter Lens	Designed based on TIG arc spectrum analysis	Blocks arc light while transmitting molten pool image
CCD Camera	Standard imaging device	Captures molten pool images
Brightness Adjustment Software	Current-dependent software adjustment	Ensures clear molten pool images
Image Processing Algorithm	Self-developed algorithm	Detects molten pool shape parameters
Processing Time	Less than 70 ms per frame	Satisfies real-time detection requirements

Key achievements:

Composite filter lens effectively isolates molten pool image from intense arc radiation.
Software-based brightness adjustment compensates for varying welding current levels.
Self-developed image processing algorithm accurately extracts molten pool front shape parameters.
Processing time of less than 70 ms per frame meets real-time monitoring requirements.

Technical Interpretation

The fundamental challenge in TIG welding molten pool imaging is the intense arc radiation that overwhelms the relatively weak thermal emission from the molten pool. The composite filter lens design addresses this by selectively transmitting wavelengths characteristic of the molten pool while blocking the dominant arc emission bands. This spectral filtering approach is based on careful analysis of the TIG arc spectrum, identifying wavelength regions where arc emission is minimal but molten pool emission is significant.

The software-based brightness adjustment is critical because welding current variations directly affect both arc intensity and molten pool temperature. Without dynamic brightness compensation, the captured images would exhibit inconsistent contrast, leading to unreliable shape parameter extraction. The algorithm's ability to process images in under 70 ms demonstrates that real-time feedback control is feasible, enabling closed-loop welding process optimization.

Engineering Practice Implications

For TIG welding applications in pipe and fitting fabrication:

Real-time molten pool monitoring enables immediate detection of process deviations such as excessive penetration, insufficient fusion, or weld bead width variation.
The visual sensing approach can be integrated into automated welding systems for closed-loop control of welding parameters.
Molten pool shape parameters such as width, length, and area provide direct indicators of weld geometry and fusion quality.
The technology is particularly valuable for TIG welding of thin-wall pipes and fittings where precise heat input control is essential.

The 70 ms processing time represents a practical threshold for real-time control in welding applications, allowing sufficient time for parameter adjustment between consecutive image frames.

Study Insights and Reflections

This research represents an important milestone in the development of welding process monitoring technology. The combination of optical filtering, adaptive imaging, and real-time image processing establishes a complete framework for in-situ molten pool observation. For engineers working on welding quality control, this work demonstrates that visual sensing is a viable and practical approach to monitoring critical welding process parameters.

The methodology developed in this study has direct relevance to modern welding monitoring systems. The principles of spectral filtering, adaptive imaging, and real-time processing remain applicable to contemporary systems using higher-resolution cameras and more sophisticated image analysis algorithms. For pipe fabrication engineers, the ability to monitor molten pool geometry in real time provides a powerful tool for ensuring consistent weld quality and detecting process anomalies before they result in defective welds.