Reducing energy consumption of pulsed-gas metal arc additive manufacturing through machine learning algorithms

Gas Metal Arc Additive Manufacturing (GMA-AM), also known as Wire Arc Additive Manufacturing (WAAM), has gained significant attention from both industry and academia due to its cost-effectiveness and capability to produce medium- to large-scale metal components. While previous research has mainly focused on the effect of process parameters on the final part properties, this study explores how Artificial Intelligence (AI) can support parameter selection to improve part quality and energy efficiency. An AI-based optimisation framework integrating regression and classification models is proposed to predict and optimise layer geometry, power consumption, and deposition quality. Experimental data from 72 layers of 316 stainless steel deposited via a pulsed GMAW process are used to train a Gaussian Process Regression (GPR) model that estimates electrical power and layer geometry from process parameters. A Random Forest Classifier (RFC) identifies parameter combinations leading to non-compliant layers, while a sustainability-oriented fitness function is optimised using a Genetic Algorithm (GA). The proposed framework achieves a 14.4 % reduction in input energy without compromising layer quality and demonstrates a 15 % energy saving in the fabrication of a test component.