| Resumo: | In this article, we present the application of a global optimization technique, in particular the GlobalSearch command from MatLab®, in the updating of structural dynamic models. For comparative purposes, we evaluate the efficiency of the global method relatively to the local search method previously used in the Finite Element Model Updating program. The Finite Element Model Updating programs are designed with the primary purpose of validating and optimizing structural numerical models. The first step for structural optimization process is to idealize the desired behavior of the dynamic model to develop, or collect experimental data of a physical model considered as the reference model. The process begins with the construction, on a finite element program, of a numerical model with initial physical parameters, preferably close to the reference model parameters. The numerical model is then submitted, through a Finite Element Model Updating program, to a successive parametric updating until improving its dynamic behavior described by their natural frequencies, mode shapes and damping properties, be similar to the dynamic behavior of the reference model. The Sequential Quadratic Programming algorithm was already used in the optimization of the Finite Element Model Updating program, and the obtained solutions showed that it can't achieve the global optimal value of the objective function. This kind of methods, used for nonlinear constrained optimization problems, have, generally, difficulties to achieve the global optimum, since they are local optimization methods.
|
|---|