In-process monitoring of weld quality in thin plate lap welding by using image sensing

Abstract

Automated welding process is expected to realize a high-quality and high-productive welding process without skill of welders. It is sometimes difficult to obtain demanded welding quality because of any disturbance such as variation of the assembly accuracy in welding object. One of the solutions to overcome the problem is in-process monitoring of welding phenomena using various sensors. In this research, we focus on the visual sensor. The weld pool shape during the welding process is obtained by a compact CMOS camera and the relationship between the weld pool shape and weld quality is discussed. The monitoring technique is applied to CO2 welding on thin plate lap joint. In the experiment, deviation of the target position and the gap between the upper and lower plates are changed as disturbance. The correlation between them and the welding quality are determined from the cross-section shape. The weld quality depends on the target position, the gap conditions. From the obtained images by a CMOS camera, it’s confirmed that the molten pool tended to tilt toward the lower plate as the target position changes from the lower plate side to the upper plate side. Therefore, we developed an image processing program to extract the molten pool right and left area ratio (RR, RL) that indicates the inclination of the molten pool. Relationship between the average value of them during welding time and target position deviation, and the gap size are correlated respectively. As a result, the target position deviation and gap size can detect from the image of the weld pool. Moreover, the relationship between welding quality and RL/RR is discussed. There’s the correlation between RL and welding quality. This result shows that in-process monitoring of weld quality is possible by calculating RL.