masterThesis
Desenvolvimento de metodologia para inspeção de soldas MAG em componentes automotivos por imagens de raios X industrial
Fecha
2021-03-31Registro en:
SOARES, Cristiane Maciel. Desenvolvimento de metodologia para inspeção de soldas MAG em componentes automotivos por imagens de raios X industrial. 2021. Dissertação (Mestrado em Física e Astronomia) - Universidade Tecnológica Federal do Paraná, Curitiba, 2021.
Autor
Soares, Cristiane Maciel
Resumen
Analyzing that non-destructive tests (END) by computerized radiography (RC) have great potential for application in the automotive industrial sector, since the quality control processes of automotive components and, in particular, previous components of automobile bodies are rigorous and must meet the acceptance criteria pre-defined in standards, the aim was to develop a proposal for a methodology using radiographic tests on automotive components of the bodywork and exhaust, using the technique of computerized radiography. For this purpose, studies on the optimization of the radiographic arrangement were necessary, as well as simulations in specimens with defects simulating discontinuities and real inspections of automotive components containing welded regions. Digital radiographic images from the image bank GDXray were also analyzed, he was not qualified as an image supply technician, with the aim of highlighting and delimiting the MAG weld bead and its discontinuities. Using the manipulation of the images from the database, the same processes were used in the images acquired in the radiographic tests, totaling 25 of which 28 MAG weld beads were extracted. Finally, the current application techniques versus the proposed testing technique were compared, with the intention of proportioning the testing methodology, for this type of application and use in automakers in Brazil. Of the 28 sales cords, 12 had discontinuities identified as porosities and cracks, 100 % of the discontinuities were correctly segmented (true positive), images segmented as false positive were not identified, which allows according to demand, where they can be estimated as process limitations, reliability and repeatability.