Energy-Quality Process Parameter Map using Machine Learning in Wire Arc Additive Manufacturing

This paper proposes a method of generating a process parameter map to visualize the energy and quality availability graphically using machine learning in wire arc additive manufacturing (WAAM). In the proposed method, a machine learning model is generated to predict heat input by training numerical voltage data, while the heat input represents energy performance. Another machine learning model is generated to classify the normal or two defect types of the current state by training the predicted heat inputs. The results of the two models are combined and visualized in the form of a three-dimensional map to project heat input and normality distributions with regard to travel speed and wire feed rate process parameters. A case study is demonstrated to evaluate the performance of the models and the feasibility of the proposed method. The energy-quality process parameter map enables operators to select the two process parameters correctly for energy reduction simultaneously with quality assurance in WAAM.