An experimental study of structural damage detection using incomplete measurements

Abstract

In this paper, algorithmic approaches to enhance damage detection capability when faced with incomplete measurements are addressed. The incomplete measurement problem arises in part because of the mismatch in the number of degrees of freedom included in the structural math model versus the number of degrees of freedom instrumented during structural testing. Studies comparing model reduction, eigenvector expansion, and a hybrid model reduction/eigenvector expansion are performed using experimental data. These approaches to the incomplete measurement problem are evaluated within the frameworks of optimal matrix update theory (both sparsity and non-sparsity preserving algorithms) and minimum rank perturbation theory, which are both applicable for model refinement as well as damage location. Experimental evaluation of the proposed approaches utilize data from the NASA LaRC 8-bay and MDA 10-bay truss facilities.