Using machine vision for observation of phenomena during laser hybrid welding

Abstract

Laser hybrid welding process is widely studied today, with many research groups trying to clarify the phenomena during the process. Many of the changes occurring during laser hybrid welding process can be observed with the help of high speed imaging. The high speed of the droplet formation and their movement to the melt pool set strict requirements for the imaging frequency in order to provide reliable information about the process. Because of large number of images produced by the high-speed (over 1000 Hz) imaging, it is time-consuming, error-prone, and frustrating to inspect the process manually, using only the human eyes for the observation. These problems could be alleviated by an automated system for the observation.This paper presents an off-line machine vision system created and used to study videos of CO2 laser-MAG hybrid welding experiments in which welding parameters were changed one at a time. The quality of the welds was observed by visual and macrographic examination. In the study, the regularity of pulse frequency of the arc and direction of filler material droplet flight were measured with the help of machine vision. Due to stable nature of keyhole welding part of the process, these characteristics show the stability of the whole welding process. It was found out that the automatic observation can be used if the quality of the images remains good enough.Experimentally, the inspection system provided information that the welding parameters have an effect on direction of droplet movement. This kind of automatic observation of changes in the process could provide a valuable tool for process optimization. The gathered information can be utilized in achieving an optimized process giving the best possible productivity.Laser hybrid welding process is widely studied today, with many research groups trying to clarify the phenomena during the process. Many of the changes occurring during laser hybrid welding process can be observed with the help of high speed imaging. The high speed of the droplet formation and their movement to the melt pool set strict requirements for the imaging frequency in order to provide reliable information about the process. Because of large number of images produced by the high-speed (over 1000 Hz) imaging, it is time-consuming, error-prone, and frustrating to inspect the process manually, using only the human eyes for the observation. These problems could be alleviated by an automated system for the observation.This paper presents an off-line machine vision system created and used to study videos of CO2 laser-MAG hybrid welding experiments in which welding parameters were changed one at a time. The quality of the welds was observed by visual and macrographic examination. In the study, the regula...