Zhuojin Pipe Fitting Co., LtdUnivariate Adjustment of Pulse MIG Welding Parameters Using Least Squares Fitting
Literature Overview
This paper by Xue Jiaxiang and colleagues from South China University of Technology, published in the Transactions of the China Welding Institution (2014, Vol. 35, No. 8, pp. 75–78), addresses a long-standing practical challenge in digital welding power source development: the complexity of pulse MIG welding parameter management. In conventional pulse MIG welding, operators must independently adjust multiple interdependent parameters including base current, pulse current, pulse frequency, on-time, off-time, and wire feed speed. The paper proposes a univariate adjustment method based on the least squares fitting technique, which allows a single master parameter (welding current) to govern all other pulse parameters through pre-calibrated fitting curves.
Core Technical Approach
The fundamental idea is to treat welding current as the sole independent variable and express every other pulse parameter as a function of this current through polynomial fitting. The authors conducted systematic process experiments with large parameter steps to collect ideal welding data across a wide range of welding currents. They then applied the ordinary least squares (OLS) method to determine the optimal polynomial coefficients for each parameter-current relationship.
The method applies to both single-pulse and dual-pulse welding modes. For single-pulse operation, the fitting relationships typically take the form:
- $I_p = f_1(I_b)$ — pulse current as a function of base current
- $f_p = f_2(I_b)$ — pulse frequency as a function of base current
- $t_{on} = f_3(I_b)$ — on-time as a function of base current
- $t_{off} = f_4(I_b)$ — off-time as a function of base current
- $v_{wfs} = f_5(I_b)$ — wire feed speed as a function of base current
where $I_b$ is the base (average) welding current.
| Parameter | Typical Fitting Form | Engineering Significance |
|---|---|---|
| Pulse current ($I_p$) | Linear or quadratic in $I_b$ | Controls droplet detachment energy |
| Pulse frequency ($f_p$) | Linear or quadratic in $I_b$ | Governs heat input rate and deposition rate |
| On-time ($t_{on}$) | Linear in $I_b$ | Determines peak heat concentration |
| Off-time ($t_{off}$) | Linear in $I_b$ | Controls inter-pass cooling and spatter |
| Wire feed speed | Linear in $I_b$ | Ensures stable arc length and metal transfer |
Process Verification and Results
The authors validated the fitting method through actual welding trials on low-carbon steel plates. The results demonstrated that the welding process remained stable across the entire calibrated current range, producing welds with:
- Uniform fish-scale bead pattern indicating consistent droplet transfer
- Minimal spatter, confirming that the pulse parameters remained within the stable transfer window
- Aesthetically acceptable weld bead profiles without undercut or excessive reinforcement
The key advantage of this approach is that all experimental data points are fully utilized in the fitting process, unlike piecewise linear interpolation which only uses adjacent calibration points. The algorithm is computationally simple, requiring only polynomial evaluation at runtime, making it well-suited for implementation in digital welding power source controllers with limited processing resources.
Integration with Engineering Practice
From a manufacturing standpoint, this univariate adjustment approach has significant implications for the production of steel pipe and pipe fittings. In pipe welding operations — whether for ERW/HFW pipe manufacture, girth welding of large-diameter line pipes, or the welding of butt-weld fittings — operators frequently need to adjust parameters rapidly when transitioning between different pipe diameters, wall thicknesses, or joint configurations. The univariate method reduces operator training burden and minimizes the risk of parameter mismatch that could lead to:
- Incomplete fusion in thick-walled pipe joints
- Excessive penetration or burn-through in thin-walled pipe and fitting fabrication
- Poor bead profile affecting dimensional tolerance of formed fittings
For pipe fitting manufacturing — particularly for elbows, tees, and reducers formed by welding — the ability to quickly transition between welding positions and joint geometries while maintaining stable pulse parameters is particularly valuable. The method also facilitates integration with robotic welding cells where parameter scheduling is automated based on joint geometry databases.
Key Reflections and Limitations
While the method is elegant in its simplicity, several considerations merit attention. First, the quality of the fitting depends entirely on the quality and coverage of the initial calibration experiments. If the calibration range does not encompass the full operating envelope of the production process, extrapolation beyond the fitted range may produce unstable welding. Second, the method assumes that the relationship between welding current and each pulse parameter is unidirectional and single-valued, which may not hold for all material thicknesses or joint configurations. Third, the method does not inherently account for environmental factors such as wind, shielding gas purity, or wire diameter variation that can shift the stable transfer window.
In practice, I would recommend combining this univariate fitting approach with a closed-loop arc voltage or arc length feedback system to compensate for real-time disturbances. The least squares fitting provides the initial parameter set, while feedback control maintains stability during actual welding. This hybrid approach has been successfully implemented in several modern digital welding power sources for pipe and fitting fabrication.
Summary
The least squares univariate adjustment method presented in this paper offers a practical and mathematically rigorous solution to the parameter management challenge in pulse MIG welding. Its simplicity, data efficiency, and computational economy make it well-suited for industrial implementation in pipe and fitting welding operations where rapid parameter transitions and consistent weld quality are essential. The method's primary limitation — sensitivity to calibration data quality and range — can be mitigated through careful initial characterization and supplementary feedback control, making it a valuable tool in the modern welding engineer's toolkit.