Elastic-plastic properties of ceramic matrix composites characterization by nanoindentation testing coupled with computer modeling

Abstract

The nanoindentation test is utilized to measure the hardness and elastic modulus of various components within Ceramic Matrix Composites (CMCs), including continuous stiffness measurement (CSM) and quasi-static testing. To further characterize the elastic-plastic properties of CMCs, an efficient method is proposed by using backpropagation (BP) neural network inversion, which is based on the relationship between load and depth obtained by the nanoindentation experiment and Abaqus finite element simulation. As part of the analysis, the load-depth (p-h) curve is simplified to six characteristic parameters through the fitting of the experimental results. The pre-estimated elastic-plastic characteristics are utilized to model the nanoindentation test for CMCs. A BP neural network is used to perform deep learning on simulation data, which leads to a network structure that matches the trained parameters of the p-h curve and elastic-plastic properties. The experimental data for fiber is inputted into the network to acquire the elastic-plastic properties required in practice. The accuracy of the proposed method is verified by comparing the inversion results with the simulation outcomes.