Free Surface Reconstruction of Magnetic Controlled Rotating Arc Overlay Welding Using Delaunay Triangulation

Overview of the Study

This 2018 paper, published in the Transactions of the China Welding Institution, presents a novel approach to free surface reconstruction for multi-pass overlay welding using magnetic controlled rotating arc technology. The authors, Hong Bo, Yao Qiang, Yin Li, and Lei Weicheng from the Hunan Provincial Key Laboratory of Welding Robots and Their Applications at Xiangtan University, address a critical challenge in automated overlay welding: the accurate reconstruction of weld bead geometry to enable adaptive control of subsequent welding passes. The research was supported by the National Natural Science Foundation of China and the Hunan Provincial Natural Science Joint Fund.

Technical Challenge and Problem Statement

In multi-pass overlay welding, each subsequent pass is deposited onto the previously formed weld bead. The geometry of the previous bead directly influences the deposition characteristics, bead shape, and final properties of the next pass. Traditional approaches rely on manual visual inspection and real-time adjustment of welding parameters and torch orientation, which is labor-intensive, subjective, and prone to error. The need for automated, accurate, and real-time weld surface reconstruction technology is therefore critical for achieving high-quality automated overlay welding.

The magnetic controlled rotating arc (MCRA) technology offers unique advantages for overlay welding, including high deposition rates, uniform bead profiles, and the ability to produce smooth, continuous overlay surfaces. However, the implementation of automated MCRA welding requires precise knowledge of the existing surface geometry to plan the deposition path and control the arc parameters effectively.

Methodology: Kriging Interpolation and Delaunay Triangulation

The proposed method combines Kriging interpolation with Delaunay triangulation to reconstruct the weld surface from discrete sensor data points. The approach involves several key steps:

Step	Description	Technical Basis
1. Data Acquisition	Collect 3D coordinate data from weld sensor	Triangulation or structured light scanning
2. Kriging Interpolation	Generate interior points through weighted interpolation	Geostatistical Kriging method
3. Interval-Growth Triangulation	Construct Delaunay triangulation incrementally	Computational geometry
4. Surface Projection	Map triangulated mesh to 3D space	Coordinate transformation
5. Surface Reconstruction	Form complete triangular mesh surface	Mesh generation algorithm

The Kriging interpolation method is borrowed from geostatistics and provides optimal unbiased estimation of values at unsampled locations based on the spatial correlation structure of the data. The key advantage of Kriging over simpler interpolation methods such as inverse distance weighting is its ability to account for the spatial autocorrelation of the data, resulting in smoother and more accurate surface reconstructions.

The Delaunay triangulation is a fundamental computational geometry algorithm that produces a triangulation maximizing the minimum angle of all triangles in the mesh. This property ensures that the resulting triangles are as equilateral as possible, which is beneficial for numerical analysis and surface rendering. The interval-growth approach allows for the incremental addition of points to the triangulation, which is computationally efficient and suitable for real-time applications.

Performance Evaluation and Real-Time Capability

The authors report that the proposed method achieves good surface reconstruction quality and high real-time performance. The real-time capability is essential for closed-loop control of the welding process, where the reconstructed surface geometry must be available before the next pass is initiated.

The accuracy of the reconstruction depends on several factors:

Sensor accuracy: The precision of the 3D coordinate data acquired by the sensor directly affects the quality of the reconstructed surface.
Sampling density: A higher density of sampled points leads to more accurate surface reconstruction, but increases computational load.
Interpolation parameters: The choice of Kriging variogram model and parameters influences the smoothness and accuracy of the interpolated surface.
Triangulation robustness: The Delaunay triangulation algorithm must handle degenerate cases, such as collinear or coincident points, without failure.

Engineering Practice and Automation Implications

The integration of surface reconstruction technology into automated overlay welding systems represents a significant advancement in welding automation. The ability to accurately reconstruct the weld surface geometry enables:

Adaptive torch path planning for subsequent passes
Real-time adjustment of welding parameters based on surface geometry
Automated quality assessment of deposited beads
Closed-loop control of overlay thickness and profile

For industrial applications such as large-scale component repair, pipeline overlay, and surface hardening of structural components, the proposed method offers a practical solution to the challenge of automated multi-pass welding. The computational efficiency of the Kriging-Delaunay approach makes it suitable for integration into real-time control systems, where surface reconstruction must be completed within the interpass time window.

The study demonstrates the successful application of computational geometry and geostatistical methods to welding engineering problems. This interdisciplinary approach, combining insights from computational geometry, statistics, and welding science, exemplifies the kind of cross-disciplinary innovation that drives progress in advanced manufacturing technologies. For engineers developing automated welding systems, the proposed method provides a robust and efficient solution for weld surface reconstruction that can be adapted to various welding processes and application scenarios.