EUROPEAN CONGRESS AND EXHIBITION ON ADVANCED MATERIALS AND PROCESSES - EUROMAT 2021

Machine learning algorithms have already been used to interpret indentation data. In this study, different machine learning models will be used on data with different fidelities, closing the gap from 2D and 3D simulations to actual experiments using Residual Multi Fidelity Neural Networks combined with transfer learning and other machine learning techniques. Finite element produced data will be used to find features in indentation curves and train machine learning algorithms accordingly to elasto-plastic parameters and tip radii of experiments. The estimation of tip-radii will give new insights into tip wear during indentation experiments. As a proof of concept, indentation mapping combined with electron backscatter diffraction will be used to interpret machine learning estimations of a polycrystalline copper foil. The approach will demonstrate possible usage of advanced indentation data evaluation for future high throughput materials science applications. Combining these techniques could lead to faster materials testing and characterisation for industrial applications such as production of modern microelectronics.