Identificação de danos em sistemas mecânicos considerando incertezas e utilizando métodos de otimização

Abstract

One of the main problems encountered in structural integrity analysis of mechanical systems is the process of damage identification. The existence of uncertainties in measurement data and in the mathematical modeling of such mechanical systems can lead to erroneous conclusions in the process of damage identification. This work aims to address a study of damage identification considering the presence of uncertainties both in the experimental data as in the mathematical model of the mechanical systems studied. This study is based on the comparison of four optimization techniques (Nelder-Mead Simplex, BFGS Quasi-Newton, Inner Point and Sequential Quadratic Programming - SQP) related to the damage identification process, as well as in the comparison of two uncertainty analysis techniques (Monte Carlo Method and Polynomial Chaos Expansion).