| dc.description.abstract | The search for a industrial method of production more systematic, with better quality control and increased productivity have driven the use of robots in various industrial sectors. And the welding, as an industrial and manufacturing process commonly used, has also been going through a process of robotics. However, to make the welding robot more efficient and economical it is necessary to reduce material waste and time spent on destructive and nondestructive tests by developing models that simulate the welding process and are able to predict the final geometry of the weld bead from the welding parameters. Therefore, the models should be highly reliable in predicting the weld bead geometry to allow a process carried out with the determined parameters to provide a bead with the desired mechanical properties. The desired reliability is achieved with models that have great accuracy, and for this it is necessary to know the real geometry of the weld bead reinforcement. Thus, considering the importance of the robotic welding and its application in an efficient way, this dissertation proposes a critical and comparative study of models for predicting the weld beads geometry described in the literature and proposes a model based on the best techniques of parameterization and simulation to predict the geometric shape of the weld bead reinforcement through the welding parameters. The model will be based on experimental data obtained in laboratory, using the GMAW process carried out by a six degrees of freedom industrial robot. Various weld beads were made with different values of voltage, wire feedrate and travel speed in order to analyze the prediction accuracy of the model results and actual results | |
