Multiscale identification of parameters of inhomogeneous materials by means of global optimization methods

This paper deals with the identification of material parameters at a microscale on the basis of measurements at a macroscale. Inhomogeneous materials such as composites and porous media are considered. Numerical homogenization with the use of a representative volume element is performed to obtain a macroscopically homogenized equivalent material. The evolutionary algorithm is applied as the global optimization method to solve the identification task. Modal analysis is performed to collect data necessary for the identification. Different ranges of measurement errors are considered. A finite element method is employed to solve a boundary-value problem for both scales.