Concept for sensor based self tuning gas metal arc welding process control

Abstract Arc welding processes, especially gas metal arc welding, are a common technology for joining pieces in industrial production. Therefore, welding quality depends heavily on process settings, such as synergic parameter settings and welding torch guide. Finding suitable welding parameters requires a high level of understanding of the process and experience. To support the identification of initial process settings for varying process boundary conditions, the concept of a sensor-based, self-tuning welding process is introduced. Therefore, optical melt pool images are acquired, and relevant key points are extracted through artificial intelligence-based image processing. The objective of control is to achieve a preset seam width, a symmetrical seam and correct path. In a first step, control algorithms are developed for the individual goals. The melt pool width is controlled by adjusting the wire feed speed, while maintaining a constant welding speed. Lateral path corrections are made to ensure correct positioning of the wire tip over the root, and the torch angle is adjusted to achieve symmetry of the melt pool. A T-joint in horizontal 2 F position is used for the development. The control strategies are then combined, and the parameters are tuned simultaneously. The results of the investigations show that optical melt pool observation can be used to create a welding system for automatic parameter tuning. With the help of the system, the effort required to find suitable parameters can be reduced by defining the requirements for seam width and varying component positions.