3D printed piezoelectric composite filament width and height prediction using individual and stacking ensemble of machine learning algorithms

Abstract

The dimensions of extrusion printed filament determine its processing resolution and efficiency. Filament width and height are related to printing process parameters. BaTiO3/PDMS piezoelectric composite was selected to study the effects of four key process parameters, namely nozzle diameter, nondimensional height, extrusion pressure, and printing speed, on the filament width and height. Five individual machine learning models are established to predict the dimension of printed lines. For the models demonstrating good prediction performance, an ensemble learning model based on stacking method is established. The findings indicate that the prediction performance of ensemble learning surpasses that of individual models. Specifically, the coefficient of determination for predicting filament width and height reaches 0.9370 and 0.9646, respectively.