Visual Image Sensing Based Precision Pulse TIG Weld Seam Tracking System

Literature Overview

Published in 2001 in The Journal of Welding, this paper by Chen Nian, Sun Zhenguo, and Chen Qiang from Tsinghua University, funded by the National Natural Science Foundation of China (Grant 59975050), presents a vision-based seam tracking system for precision pulse TIG welding. This work addresses a fundamental challenge in automated welding: maintaining precise electrode-to-workpiece alignment during the welding process, which is critical for achieving consistent weld quality in cladding and overlay applications.

Core Technical Content

System Architecture

The visual seam tracking system consisted of:

Component	Specification	Function
Image sensor	CCD camera (monochrome)	Arc and weld bead imaging
Lens system	Narrow-band filter + telecentric lens	Arc light isolation and distortion-free imaging
Image processing	Frame grabber + dedicated processor	Real-time feature extraction
Control unit	Microcomputer with motion controller	Servo motor drive for torch positioning
Welding power source	Pulse TIG (inverter-based)	Adjustable pulse parameters

Image Processing Algorithm

The core innovation was the real-time image processing algorithm that:

Pre-processing: Applied median filtering to reduce noise and narrow-band filtering to isolate the arc and weld bead from background.
Feature extraction: Identified the arc center, weld bead edges, and groove geometry through edge detection algorithms (Sobel, Canny).
Position determination: Calculated the lateral offset between the torch centerline and the groove centerline.
Feedback control: Generated correction signals for the servo-driven torch positioner.

Tracking Performance

Parameter	Performance	Significance
Tracking accuracy	±0.2–0.5 mm	Acceptable for groove alignment
Response time	50–100 ms per cycle	Fast enough for welding speeds up to 500 mm/min
Maximum correction rate	2–5 mm/s	Handles moderate groove misalignment
Robustness	Effective under arc spatter and oxidation	Practical for production environments

Pulse TIG Parameter Optimization

The system was integrated with pulse TIG welding parameters optimized for cladding applications:

Pulse current: 150–250 A (penetration control)
Base current: 30–80 A (heat input between pulses)
Pulse frequency: 5–15 Hz (thermal cycling control)
On-time ratio: 30–60% (heat input modulation)

Integration with Engineering Practice

For automated cladding and overlay welding operations, seam tracking is essential for:

Multi-layer cladding: Each subsequent layer must be precisely positioned over the previous layer to ensure uniform coverage and avoid gaps or overlaps.
Large component fabrication: Pressure vessels and heat exchangers require tracking over long weld lengths where thermal distortion can shift groove positions.
Curved surface welding: Cylindrical and spherical vessel surfaces require continuous tracking to maintain proper torch angle and position.
Groove preparation tolerance compensation: Tracking systems can compensate for variations in groove geometry caused by machining tolerances or fit-up errors.

Practical Implementation Considerations

When implementing visual seam tracking for cladding operations, the following factors must be considered:

Arc light interference: The intense arc radiation can saturate camera sensors, requiring specialized filters and exposure control.
Spatter and oxidation: Cladding materials (especially nickel-based alloys) may produce significant spatter that obscures the weld bead.
Thermal distortion: Real-time tracking must account for workpiece movement due to thermal expansion and contraction.
Calibration: Regular calibration of the camera-to-torch geometry is necessary to maintain tracking accuracy.

Key Technical Insights

This research demonstrated that visual sensing could provide reliable seam tracking for precision welding applications. The key insight is that real-time feedback control significantly improves weld quality by compensating for inevitable variations in workpiece geometry and thermal distortion. For cladding operations, where layer-to-layer registration is critical for achieving uniform overlay thickness and complete coverage, such tracking systems are indispensable.

The evolution from this early CCD-based system to modern high-speed CMOS cameras with advanced image processing capabilities has dramatically improved tracking performance. However, the fundamental principles of arc/bead detection, position determination, and closed-loop control remain unchanged. Engineers designing automated cladding systems should understand these principles to select and implement appropriate tracking solutions for their specific applications.