Interval analysis for updating FEM parameters using uncertain experimental data

The goal of FE models updating techniques is to modify the model parameters to get the optimal match between experimental and numerical modal data. The optimal solution is that in which a distance function, in the parameters space, between the model and the structure is at a minimum. A typical drawback of classical methods is the lack of uniqueness in the solution that could lead to unrealistic values of the parameters. This is due to the presence of unavoidable experimental and modelling errors, In the work, in order to cope with these errors, the parameters are assumed as uncertain quantities whose range of values can be described by means of intervals. Accordingly, the updating problem is recast, in the framework of interval analysis, as a global optimization problem in a bounded domain of the parameters. Along with the presentation of the theoretical background and the main steps of the proposed methodology, the results are discussed by the help of numerical examples. Finally, the interval model updating is applied to the case of a simple real structure whose modal parameters are made available by a previous experimental research.