The study published in Engineering Fracture Mechanics investigates the application of machine learning to modelling the material properties of metal. Specifically, it concerns the fatigue life prediction of titanium alloy samples manufactured using 3D printing. The proposed algorithm was developed at VZLU and the machine learning model was trained on experimental data by co-authors from the University of Applied Sciences Upper Austria. The fatigue test data were taken from a previous publication .
The paper also presents an effective tool for solving the small dataset size problem for training a machine learning model. The suggested dataset augmentation technique can be applied in various fields of science where it is time and cost consuming to produce a large volume of experimental data, which is necessary to train a model with high prediction accuracy.