Backside Penetration Modeling of Aluminum Alloy Pulsed MIG Welding Based on Welder Adjustment

Literature Overview

This paper, published in the journal Transactions of the China Welding Institution (Hanjie Xuebao), Volume 40, Issue 8, 2019, by Zhang Chuang, Xiong Jun from Southwest Jiaotong University, and Zheng Senmu from Sichuan Aerospace Long March Equipment Manufacturing Co., Ltd., addresses the challenge of controlling backside penetration in pulsed MIG welding of aluminum alloy thin plates. The study proposes a novel approach that combines passive visual monitoring of the backside weld pool with manual welder adjustment of welding current, and develops a nonlinear Hammerstein model through least-squares system identification to describe the relationship between current adjustment and backside weld width.

Technical Methodology and Modeling Approach

The methodology presented in this study is innovative in its integration of human factors with process modeling. Rather than relying solely on automated control systems, the approach leverages the welder's ability to observe and respond to visual feedback from the backside weld pool. The welding speed is varied using a random waveform to create dynamic fluctuations in the backside weld pool size, which are captured in real-time by a passive visual system. The welder observes these images and manually adjusts the welding current to maintain uniform backside weld width.

The use of a random welding speed waveform is a deliberate experimental design choice that excites the system across a range of operating conditions, enabling comprehensive system identification. This approach is analogous to the use of pseudo-random binary sequences in control system testing, where the input signal must contain sufficient frequency content to characterize the system dynamics across the relevant bandwidth.

System Identification Approach

The Hammerstein model is a particular type of nonlinear model consisting of a static nonlinear element followed by a linear dynamic element. In this application, the static nonlinear element represents the nonlinear relationship between welding current and weld pool geometry, while the linear dynamic element captures the thermal inertia and mass transport dynamics of the welding process. This model structure is appropriate for welding processes where the input-output relationship is nonlinear but the dynamic response can be approximated as linear.

Process Dynamics and Control Performance

The study conducts current disturbance and welding speed disturbance tests to evaluate the robustness of the proposed modeling and control approach. The results demonstrate that the nonlinear Hammerstein model can effectively control the backside weld width and exhibits good anti-interference capability. This is significant because it indicates that the model can compensate for process variations and disturbances, maintaining weld quality despite changes in welding speed or current fluctuations.

The backside penetration control problem is particularly challenging in aluminum alloy welding due to several factors: the high thermal conductivity of aluminum leads to rapid heat dissipation from the weld zone, the oxide film on aluminum surfaces creates difficulties in achieving consistent fusion, and the low melting point of aluminum increases the risk of burn-through in thin plates. The pulsed MIG welding process is well-suited for aluminum alloy welding because the pulsed current allows for controlled heat input and stable metal transfer, reducing the risk of excessive penetration and porosity.

Aluminum Alloy Welding Challenges

Engineering Practice Implications

The approach proposed in this study has significant implications for production welding of aluminum alloy components, particularly in aerospace and automotive applications where thin-walled structures are common and weld quality is critical. The integration of visual feedback with human operator adjustment provides a practical solution that does not require expensive automated control systems, making it accessible for small and medium-sized manufacturing operations.

However, the approach also has limitations. The reliance on human welder judgment introduces variability in the control response, and the passive visual system may not provide sufficient spatial resolution for very thin plates or very small weld pools. In high-precision applications, fully automated systems with closed-loop control may be necessary to achieve the required weld quality consistency.

From a quality assurance perspective, the backside weld width serves as a direct indicator of penetration quality, and its uniformity is essential for ensuring structural integrity. The study's demonstration that the Hammerstein model can maintain uniform backside weld width despite process disturbances provides confidence in the robustness of the approach for production applications.

Key Questions and Reflections

An important question is how the proposed approach scales to different aluminum alloy grades and thicknesses. The study focuses on a specific aluminum alloy and plate thickness, and the model parameters would need to be re-identified for different materials and geometries. The development of a generalized model that can be adapted to different welding conditions without extensive re-identification would be a valuable contribution to the field.

Another consideration is the effect of welding position on the backside penetration control. The study likely focuses on flat-position welding, and the approach may need modification for vertical or overhead positions where gravity affects the weld pool geometry and metal transfer. The human factor becomes even more critical in these positions, as the welder must simultaneously manage the torch angle, travel speed, and current adjustment.

Summary

This study presents a practical and innovative approach to controlling backside penetration in aluminum alloy pulsed MIG welding by combining passive visual monitoring with human welder adjustment and nonlinear system identification. The Hammerstein model provides an effective description of the current-to-weld-width relationship, and the demonstrated robustness to process disturbances makes the approach suitable for production applications. Engineers working with aluminum alloy thin-walled structures should consider this approach as a viable alternative to fully automated control systems, particularly where cost and flexibility are important considerations.