섬유강화 복합재료의 계면 균열 검출을 위한 역 해석 방법

This study proposes inverse methods to detect interface cracks between the matrix and the fiber in fiber-reinforced composite materials. Finite element simulations of composite materials require high computational cost for mesh construction and analysis. To reduce the computational cost, the asymptotic expansion homogenization (AEH) method is used to obtain the effective elastic modulus of the representative unit microstructure. The AEH method is applied to fiber-reinforced composite materials with interface cracks, and macroscopic behavior of cracked fiber-reinforced composite materials are obtained to establish training data with respect to the size and location of interface cracks. Using these results, inverse analyses are performed to identify the size and location of interface cracks in fiber-reinforced composite materials by using RBF interpolation and deep neural networks.