Weld pool oscillation frequency in pulsed gas tungsten arc welding with varying weld penetration

Control of weld joint penetration is a fundamental issue in intelligentizing the automatic robotic welding to a level comparable with skilled human welders. The oscillation of the weld pool has been found to be closely related to the weld penetration but existing technologies that are based on fluctuation in arc voltage or arc length lack the fundamental to monitor the weld pool oscillation accurately. In this paper, an improved laser vision based measurement system is proposed to capture the reflected laser lines to accurately monitor its oscillation. In this novel system for surface oscillation monitoring, a low-power laser pattern, five lines, is projected onto the specular weld pool surface and the reflection is intercepted by a high speed camera. By observing and analyzing the change of the reflected laser lines, it is found that the change of laser lines is periodic and has a strong relation with the pool oscillation. An image processing algorithm has also been proposed for extracting the oscillation frequency from the image sequence. Experiments verified that the proposed method can successfully measure the weld pool oscillation. Moreover, the oscillation modes with varying weld penetration during stationary and travelling welding process have been investigated and it is found that the oscillation mode is different under different welding process and weld penetration. Practically, during travelling welding process, the weld pool is oscillated from up to down. The oscillation frequency of partial penetration is higher than the complete joint penetration, along with an abrupt transition. The proposed sensing method thus established the foundation for a novel method to monitor and control the weld penetration in real-time for intelligent welding robots which may possess partial human intelligence to not only automate the welding process but also to intelligentize the automatic robotic welding process.