| description abstract | An investigation on stiffened isotropic and composite plates has been conducted to determine the geometric and material parameters for the plate, as well as the stiffener from experimental modal data and finite element predictions using a genetic algorithm (GA). The problem is formulated as a global minimization of the error function defined by the difference in undamped eigenvalues and eigenvectors, as predicted from the finite-element modeling to that obtained experimentally. The parameter estimation problem is solved using a GA implementing selection, crossover, and mutation operators to obtain the global minimum solution. Because stiffeners contribute substantially to the overall rigidity of the plate assembly, their position, physical properties, and orientation create considerable variation of the modal properties, as compared to the bare plate with similar construction. This makes each of the stiffened plate identification problems rather unique. GAs have been the subject of considerable interest in providing a robust search procedure for a global optimum solution for such difficult minimization problems. The method is demonstrated on a few simulated examples on stiffened plates to investigate the uniqueness and convergence of results. The methodology, although slow in execution, is found to be very robust, even in the presence of noise, for isolating interesting zones of the search space. Unlike many traditional optimization techniques, it does not get stuck at a particular local minimum due to its parallelism. | |
