Real Time Optimization Of Bead Geometry In Robotic Wire Arc Additive Manufacturing Integrated With Supervised Learning

Wire Arc Additive Manufacturing (WAAM) is a manufacturing technology that can fabricate a large-scale metallic part in a layer-by-layer fashion. It is receiving great attention from industries as a viable method of manufacturing due to its high deposition rate and cost-efficiency. However, there still exist numerous challenges that need to be overcome to ensure the geometrical accuracy of the part produced. WAAM process is highly non-linear and multi-dimensional and is difficult to model the input process parameters to the output geometrical quality of the final part, especially with an increasing number of materials introduced to WAAM. To overcome this challenge, a supervised learning control algorithm is implemented to search for a parameterized welding process while optimizing the geometry of a single-track multi-layer bead. The input parameters include torch travel speed, wire feed speed, previous layer's geometrical data, and dwell time. The output parameter is the geometry of the printed bead. The proposed algorithm is implemented and validated on a 3-axis gantry WAAM system.