Laser Vision Based Adaptive Fill Control System for TIG Welding

Literature Overview

This paper by Fu Xi-Bin, Lin San-Bao, Fan Cheng-Lei, Luo Lu, and Yang Chun-Li from the State Key Laboratory of Advanced Welding Production Technology at Harbin Institute of Technology, published in China Welding (Vol. 17, Issue 4, 2008), presents a sophisticated adaptive control system for tungsten inert gas (TIG) welding. The fundamental problem addressed is the non-uniform fill of deposited metal caused by variations in joint groove geometry during welding. The system employs laser vision sensing for real-time groove profile measurement and implements a closed-loop control strategy to adjust welding parameters dynamically. This is a landmark contribution to the field of intelligent welding control and has direct relevance to pipe welding applications where groove consistency is difficult to maintain.

System Architecture and Technical Principles

The adaptive fill control system consists of four major subsystems: a modular development kit (MDK) serving as the real-time image acquisition unit, a computer as the central controller, a D/A conversion card for controlled variable output, and a DC TIG welding system as the controlled device. The laser vision sensor projects a structured light pattern onto the weld groove and captures the reflected image, from which the groove profile is reconstructed using triangulation principles.

The feature extraction algorithm processes the captured images to determine key groove parameters including groove width, groove depth, and the position of the weld torch relative to the groove centerline. These features are then fed into a control algorithm that adjusts the welding current, travel speed, and torch oscillation amplitude in real time. The control strategy is designed to maintain a constant volume of deposited metal per unit length of weld, ensuring uniform fill regardless of groove variations.

The following table outlines the system hardware configuration and its functions:

Feature Extraction and Control Strategy

The feature extraction algorithm is the intellectual core of this system. The laser vision system captures the groove profile by projecting a laser line onto the joint surface and imaging the reflected pattern with a camera. The resulting image contains the groove geometry distorted by the viewing angle, which must be corrected through calibration. The algorithm extracts the groove edges, calculates the groove width at multiple positions along the weld length, and determines the deviation from the nominal groove profile.

The control strategy employs a proportional-integral (PI) controller that adjusts the welding current based on the measured groove width deviation. When the groove is wider than nominal, the current is increased to deposit more metal; when narrower, the current is reduced. The control loop operates at a rate sufficient to respond to groove variations within a few millimeters of the welding arc. The system also adjusts the travel speed to maintain a constant heat input per unit volume of deposited metal.

Experimental results demonstrated that the system achieves adaptive fill with high stability, reliability, and accuracy. The weld formation quality meets relevant industry criteria, and the system successfully compensates for groove width variations of up to ±0.5 mm from the nominal value. This level of compensation is critical for pipe welding applications where groove preparation tolerances can be significant, especially for larger diameter pipes or in field welding conditions.

Relevance to Pipe Welding Applications

For steel pipe manufacturing and installation, groove consistency is a major challenge. In the production of longitudinal submerged-arc welded (LSAW) pipes, the bevel preparation of plate edges must be precise to ensure proper root pass and fill pass welding. In field pipe welding for pipeline construction, groove preparation is often performed manually or with portable beveling machines, resulting in significant groove geometry variations. The laser vision adaptive fill control system offers a solution that can maintain weld quality despite these variations, reducing the need for excessive groove preparation tolerances and minimizing weld repairs.

The following table compares conventional TIG welding with the adaptive fill controlled system for pipe applications:

Study Insights and Engineering Implications

The laser vision adaptive fill control system represents a paradigm shift from open-loop to closed-loop welding control. The key insight is that groove geometry variations, which are inevitable in practical welding operations, can be compensated for in real time if the system has sufficient sensing capability and control authority. For pipe welding applications, this technology has particular value in the following scenarios: root pass welding of pipe joints where groove fit-up varies along the circumference, fill pass welding of large-diameter pipes where groove preparation is difficult to maintain consistently, and field welding operations where environmental conditions cause groove geometry drift.

The system's practical implementation requires careful attention to several factors: the laser vision sensor must be protected from arc light and spatter, the control algorithm must be robust against noise in the image data, and the system latency must be minimized to ensure timely parameter adjustments. Future developments could include multi-sensor fusion (combining laser vision with arc sensing) and data analysis-based control strategies that adapt to different joint geometries and materials without manual parameter setup. This technology is a significant step toward truly automated, high-quality pipe welding with minimal operator intervention.