Innovative model updating procedure for dynamic identification and damage assessment of structures

The model updating technique allows the understanding of the dynamic behavior of a system and its damage state. In the last years, the structural monitoring has increased its applicability thanks to the decrease of the cost of sensors and improvements in the computational power. More and more structures are today instrumented in order to assess the intervention of progressive damages, understand their structural behavior and safety in almost real time. Nowadays, the real time identification of structural parameters and damage assessment is no longer unachievable. Moreover, the uncertainties evaluation is another important task required by the model updating procedures. Combining real time assessment and uncertainties evaluation, the algorithms can drive to a judgment about unsafety conditions in the buildings, with possible evacuation and securing of the structures, which is more and more required to structural health monitoring systems. The algorithms developed in this work are focused on these topics, especially on very quick model updating procedure, with uncertainties evaluation, which allows to estimate the structural parameters along with an error assessment. The quickness of the algorithm enables for its use in real time monitoring of actual structures. The algorithm itself is based on an innovative two steps procedure, with uncertainties evaluation, solving the inverse eigenvalues problem. The first step is achieved with closed form solution (without considering the determinant equations). If the solution does not satisfy the fixed thresholds, the second iterative step should be performed in order to improve the agreement between experimental outcomes and numerical ones. This procedure allows us to write the partial derivatives of the problem itself, with respect to the experimental outcomes, in closed form. Therefore, the parameters uncertainties are computed using the errors propagation. A second procedure is developed facing the complete problem entirely in iterative way, using a genetic algorithm with response surfaces.