A new approach to inverse identification of mechanical properties of composite materials: Regularized model updating

Abstract

Inverse identification of constitutive parameters of materials might be adversely influenced by noise in the measured data. This study is concerned with an improved Finite Element Model Updating (FEMU) for accurate identification of mechanical properties of composite materials components from full-field measured displacement data. This numerical–analytical approach, namely Regularized Model Updating (RMU), is developed based on a hybrid constrained optimization algorithm. For this purpose, mechanical constraints, consisting of an appropriate homogenization model, are added as regularization factors to the optimization algorithm. The proposed method is validated by conducting several virtual experiments through elastic constitutive parameters identification of 2D composites. The sensitivity of the developed algorithm to different levels of noises of measured displacement fields is investigated. The identification results indicate that the proposed RMU methodology leads to higher accuracy of mechanical properties in comparison with FEMU method, particularly in the presence of random noise.