Damage Identification in a Multi-DOF System under Uncertainties Using Optimization Algorithms

Abstract

In this study, four optimization algorithms are applied to identify the damage in a multi-DOF dynamical system composed of masses, springs, and dampers. The damage is introduced artificially by choosing a lower value for the stiffness. The applied algorithms include Nelder-Mead Simplex, BFGS Quasi-Newton, interior point, and sequential quadratic programming - SQP. In addition, some different strategies to identify damage are applied. First, a deterministic analysis is performed to identify the damage; i.e, the values of the stiffnesses. Then, random forces are considered, and finally, the stiffness values are considered uncertain and the different strategies are compared.