Feasibility and Uncertainty Analysis of Constitutive Parameters Identification for Composite Materials Using Displacement Field Data

Due to the normalized average strain error is remarkably higher than the displacement error, a new cost-effective method based on the Levenberg-Marquardt method, which is a gradient-based optimization method, was presented in this paper. The unknown constitutive properties are determined through minimization of the square difference between the reconstructed DIC-measured and the FEM-calculated displacements. The feasibility analysis of this method is presented by synthetic displacement data. The covariance matrix of the extracted material constitutive parameters has been given explicitly. It has been proved that this method can significantly reduce the uncertainty of the extracted material constitutive parameter induced by the approximation error due to reconstruction algorithm and the measurement noise. The material constants show strong robustness to the measurement noise. And the optimization procedure is not sensitive to the formulation and parameters in the initial approximation of the constitutive model. A sensitive matrix was derivation and evaluation iterative in this method to improve the convergence of the optimization procedure.