Design of automatic tungsten inert gas (TIG) welding system using machine vision for AA1100 aluminum plate

The most widely used manufacturing process in the industry is welding. One example of developing technology is TIG (Tungsten Inert Gas) automatic welding. Automatic TIG welding is important, especially when welding aluminum because the process can maintain quality, consistency, and repeatability. This study aims to create a TIG welding system capable of automatically welding AA1100 materials with the help of machine vision and neural networks. This research begins with designing and assembling mechanical, electrical components and making a program so that the welding can run automatically. Then, welding visualization is obtained with the help of a CCD camera placed on the front and top. The dimensions of the specimen used in this study were 0.14 m length, 0.07 m width, and 3.8 mm thick. The system will carry out the welding process at a constant speed, and the speed changes with the step of increasing the welding speed. This test is carried out to obtain the characteristics of the welding results at varying speeds so as to obtain training data for artificial neural networks. Furthermore, this research will create a system to capture molten pool images used as data to carry out the training process on an artificial neural network. The results of this study are the initial stages of making an automatic welding system that uses machine vision and artificial neural networks, showing the system can perform manual welding with speed input and obtain results from measuring the width of the upper weld bead using machine vision which is only able to achieve an accuracy of 0, 1 mm or get an error of 1%.