Towards the automation of WAAM process: bead geometry prediction using machine learning alghoritm

This study focuses on the development and implementation of a predictive algorithm utilizing machine learning techniques for the Wire Arc Additive Manufacturing (WAAM) process. The algorithm is designed to predict in real-time the bead width and height based on parameters received from embedded sensors, enabling detailed insight into the deposition process. The primary aim of this research is to enhance the precision of bead geometry in aerospace applications, addressing challenges related to non-standardized positioning in robotic additive manufacturing. While the algorithm demonstrates considerable potential, it is not yet capable of real-time application for dynamically correcting the deposition head position or key performance indicators (KPIs) during manufacturing. Such functionality would require additional experimentation and a more extensive dataset. For now, the algorithm can be effectively employed to predict the number of beads required in a given layer to prevent defects such as lack of fusion or excessive layer height relative to predefined tolerances. This represents a promising step toward optimizing WAAM processes and ensuring the production of high-quality components.