Section: New Results

damage detection for mechanical structures

Damage detection and localisation

Participants : Michael Döhler, Laurent Mevel.

Statistical methods using output-only data have been shown to offer a robust solution to the damage detection task. These techniques have also been combined with sensitivities extracted from

finite element models to offer information on the location of damage accounting for uncertainties in the

finite element sensitivities. In some applications, however, the formulation of the

finite element model makes implementation impractical and this motivates the search for model-free damage localization alternatives. One option is to use experimentally extracted sensitivities but their computation requires a set of constants (usually absorbed in the normalization of the eigenvectors) that are not available in output only identification. The noted limitation can be circumvented by adding a known perturbation to the mass distribution and repeating the output only identification, a procedure that can be practical in some cases. Linking a null-space based subspace damage index with experimentally extracted sensitivities allows us to infer on the position of damage without formulating a

finite element model and without the need for input measurements. The performance of the algorithm is illustrated on simulated data [19] . Damage detdction has also been applied to a large scale example of an european project [23] , [15] .

Robust subspace damage detection

Participants : Michael Döhler, Laurent Mevel.

Subspace methods enjoy some popularity, especially in mechanical engineering, where large model orders have to be considered. In the context of detecting changes in the structural properties and the modal parameters linked to them, some subspace based fault detection residual has been recently proposed and applied successfully. However, most works assume that the unmeasured ambient excitation level during measurements of the structure in the reference and possibly damaged condition stays constant, which is not possible in any application. This work addresses the problem of robustness of such fault detection methods. A subspace-based fault detection test is derived that is robust to excitation change but also to numerical instabilities that could arise easily in the computations [21] .