Analysis of 3D Printing Performance Using Machine Learning Techniques

Abstract

Additive manufacturing (AM) or 3D printing is gaining momentum in the market compared to conventional subtractive technologies due to its ability to speedily produce complex and customized geometries with less waste of material. Some 3D printing parameters are influential or crucial as they affect the final part’s mechanical properties based on the technology used. These are the printing height, the printing rate, the nozzle diameter, the nozzle movement rate, the layer thickness, the temperature, the ventilator speed, the print precision, the layer thickness, the type of infill, and the extrusion. This paper proposes the development of a neural networks model (ANN) and a hybrid neural network trained by particle swarm optimization (ANN-PSO) to get an insight into the selection of 3D printing parameters and adjust them. To ensure the quality of the 3D printing, a parametric analysis has been performed to identify the best configuration of the models. Readily available data has been used to demonstrate the potential of the proposed approach. These data have been used to train and test the algorithms and build robust models able to predict performance. The ANN and the ANN-PSO models have exhibited good overall performance that demonstrates the potential for modelling and prediction of 3D printing performance using machine learning techniques.