Study on inverse identification of composites’ nonlinear shear mechanical behavior based on finite element model updating method and artificial neural network

Carbon fiber reinforced epoxy resin matrix composites have the advantages of high specific strength and high specific modulus, and are widely used in aerospace, automobile manufacturing and other fields. The mechanical analysis of carbon fiber reinforced epoxy resin matrix composites is usually based on the linear elastic constitutive assumption. However, in engineering, their shear deformation behavior often exhibits significant nonlinearity. The degradation of the shear modulus is related to not only the shear strain, but also the normal strain perpendicular to the fiber. Accurate characterization of the nonlinear shear behavior is of great significance for the full use of the mechanical properties of composites and the design of reasonable composite structures. In this paper, short beam shear tests were carried out for carbon fiber reinforced epoxy resin matrix composite laminates. The shear modulus field of the composite was obtained by combining the digital image correlation method with the finite element model updating method. On this basis, the nonlinear shear mechanical behavior of the composite was characterized using artificial neural network. The results indicate that the degradation of shear modulus is primarily dominated by shear strain, but the normal strain perpendicular to the fiber also plays a significant role. Specifically, at low shear strain levels, a large normal strain may induce matrix damage or microcracks, leading to modulus reduction; conversely, at high shear strain levels, its influence is negligible.