Abstract
In order to propose a numerical discrete model on structural element stiffness identification, a general framework of inverse analysis is first developed for the static parameter estimation problem as a nonlinear minimization of difference between the response of a real structure and the prediction of a finite element model of that structure. To enhance the identifiliability of the identification model for a large-scale structural system, a localized identification concept is proposed, and, a statical condensation method based on the cholesky decomposition approach is introduced to predict all the unmeasured nodal displacements for preparing the input data during the iterative identification procedure. Next the Hopfieled neural network is used as a minimizator for identification procedure.
