Controlling 3D weld pool surface by adjusting welding speed

Skills needed for critical manual welding operations typically require a long time to develop, and the shortage of skilled welders has become an urgent issue the manufacturing industry is currently facing. The authors envision an innovative augmented reality welder training system to help accelerate the welder training process, in which an unskilled welder perceives the weld pool image with an auxiliary visual signal (arrow with direction and amplitude) superimposed upon, and makes speed adjustment accordingly. A critical part of the envisioned training system is to determine the optimal welding speed for an unskilled welder to follow. This paper aims to establish a machine algorithm calculating the optimal welding speed given a 3D weld pool surface, referred to as "super welder." To this end, dynamic experiments were conducted to model 3D weld pool surface characteristic parameters in response to the welding speed. A model-based predictive control (MPC) algorithm is proposed to maintain 3D weld pool surface characteristic parameters at desired values. The proposed super welder can also be directly utilized to control 3D weld pool surface in automated welding. To demonstrate its performance, automated welding experiments are conducted for pipe GTAW. Results show the proposed super welder is able to track varying set points and is robust against different welding currents and speed disturbances.