Process Planning of Robotic Wire Arc Additive Manufacturing (WAAM) Using Machine Learning
Wire Arc Additive Manufacturing (WAAM) is an additive manufacturing technique that deposits metal layer upon layer to manufacture 3D parts which is popular for producing large metal parts for aerospace, marine, automotive, and architectural applications. As a welding process, most researchers considered weld bead width, height, and penetration as the characteristic performances in WAAM. For better dimensional control and improving the process efficiency by reducing the amount of machining, the surface roughness of parts needs deep investigation. If the roughness of a deposited layer can be reduced, less machining will be required, and material wastage will be decreased too. With traditional overlapped single pass beads, layer height is not uniform, making it difficult for multilayer deposition. With the weaving path, if roughness is taken into consideration, flat layers with uniform height will be produced. Moreover, the efficiency and accuracy of the whole WAAM process largely depend on successfully implementing the 2D tool paths. Only a few researchers who investigated roughness in WAAM used the straight path for material deposition, but the investigation of the weaving path, which has a great potential to reduce layer roughness, is even rarer. Hence, the deposition of weld beads with minimized roughness demands great attention. This paper proposes a method to model surface roughness by using the weaving path. Moreover, since welding is a highly nonlinear process, Random Forest, which is an effective solution for this kind of application, but not explored adequately in welding for regression problems, will be used in this research to model and predict the layer roughness for a given set of welding parameters. Robotic welding experiments were conducted with weaving path, and roughness was calculated by acquiring point-cloud data with a laser scanner. The experimental results demonstrate that layers with better flatness have been produced and Random Forest was able to accurately predict roughness from the process parameters-wire feed speed, travel speed, weaving wavelength, and weaving amplitude.
WAAM, Additive manufacturing, Robotics, Machine learning, Random Forest, Roughness, Path planning
Yaseer, A. (2021). <i>Process planning of robotic Wire Arc Additive Manufacturing (WAAM) using machine learning</i> (Unpublished thesis). Texas State University, San Marcos, Texas.