Finite element and machine learning-assisted analysis of fiber–matrix interphase in single-fiber pull-out tests

Experimental and simulative single-fiber pull-outs were carried out to determine the interphase parameters and to investigate the damage process in the pull-out. For each experiment, a related simulation was performed, not only taking into account the corresponding fiber diameter and the individual embedded length, but also the stiffness gradient in the fiber–matrix interphase. Using a generalized traction separation law it was shown that interphase damage is probably a process of plastic material deformation. Additionally, the determination of mechanical interphase parameters was achieved through the utilization of supervised machine learning algorithms.