A proximal point-like method for symmetric finite element model updating problems (Q747223)

The authors consider model updating problems (MUPs) which concern updating of a finite element model, by replacing the unwanted eigenvalues and eigenvectors by suitably given or measured ones. Hereby some important properties of the original system symmetry, the positive semidefiniteness and the sparsity are preserved. After reformulating the problem as a constrained optimization problem, a proximal point-like method is developed to solve the variational inequalities arising from the Karush-Kuhn-Tucker optimality conditions for the Lagrangian. Under mild conditions the global convergence of the method is proved. Numerical results for a model of the serially linked mass-spring system with damping and of a spring system consisting of coupled springs, dashpots and masses show that the proposed method works well.