MACHINE VISION IN GAS METAL ARC WELDING PROCESS: A CASE STUDY

Resumo

Machine vision is a trend in many industries, due to improvements obtained from its application and high precision on visual information extraction. In the welding researches, the use of machine vision is ample with application on the measurement of drop diameter, observation of molten pool oscillations, recognition of spatter pattern and many other dynamics. In this work, a study about machine vision is provided, where some machine vision algorithms, which are applicable for welding processes analyses, are discussed. As a case study, the discussed algorithms were applied over images obtained from a high speed camera by shadowgraphy technique of mass transfer in a MIG/MAG (Metal Inert Gas/Metal Active Gas) process by globular transfer mode. The aim of applied algorithms is turn evident the edges of weld droplet to identify their radius. This is an initial step of a bigger work, where this information will be synchronized with electrical signal of voltage and current transducers and weld bead image, creating a dataset. The dataset information will be crossed with a weld bead quality evaluator, which will point when, where and what is the possible cause for weld defect as oscillations in current or voltage for training of an intelligent controller. The information of droplet diameter also allow relate the material volume which is transferred for the molten pool, then with properly calculation, it will be possible to calculate the generated spatter.