Using Deep Learning in The Development of A Welding Practice Training Support System

Phung, Pham-Tuan-Anh; Tran, Ngoc-Thien; Tran, Vu-Hoang; Huynh, Ton-Nghia

doi:10.1109/icsse52999.2021.9538464

Cited by 1 publication

(1 citation statement)

References 5 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are seam tracking problems, nonlinear variations in pipe gaps, fit-up misalignments, and the presence of tacks that need the welder's skill to adjust the welding promptly. State-of-the-art in welding control applications is evolving to capture and integrate the perception and cognition of skillful welders into the autonomy of robots [2][3][4]. Seam tracking was among the early challenges, and laser systems emerged about a decade ago to provide a solution [5].…”

Section: Introductionmentioning

confidence: 99%

Robotic Adaptive Algorithm for Solving Fit-up Variations in Welding at Industrial Scale

Ashoori¹,

González²,

Karimzadeh³

et al. 2023

Proceedings of the 10th International Conference on Control, Dynamic Systems, and Robotics (CDSR 2023)

View full text Add to dashboard Cite

Open root pass welding in Gas Metal Arc Welding (GMAW) is always challenging due to the nonlinear random variations in pipe gaps and the presence of tacks. Manual welding requires a lot of skill from senior welders to react and control many variables promptly. In the transition to robotic welding, tracking solutions based on laser or vision systems have emerged to address the tracking issue. However, adapting the welding parameters (e.g. wire feed speed) and motion parameters (e.g. travel speed) is still essential in getting a consistent, high-quality weld. This work presents an adaptive control approach to pipe welding. The method combines a visionbased system that replicates the perception of welders with real-time control to live-adjust welding and motion parameters based on the instantaneous pipe gap, learning about the tack and fusing it on the root pass -a critical challenge for robotic welding applications. The controller monitors the state condition and communicates the proper process and motion update with the robot according to the real-time gap and tack state. The resulting closed-loop system enables higher quality and consistency of weld throughout the pipe welding.

show abstract