An enhanced energy vibration-based approach for damage detection and localization

This paper addresses the enhanced identification and localization of structural damage by means of the recorded response induced by ambient excitation. Damage detection is based purely on the vibration energy in structural acceleration records, deriving thereof the normalized cumulative power spectral density as the characteristic damage sensitive quantity. As key aspect of this contribution, a method for “correction” of the recorded response is proposed, to account for deviations from perfect stationary white noise excitation. Based on an overdetermined system of equations, recorded spectra are modified to better fit previously recorded spectra. This fitting affects broader frequency ranges, while the damage sensitive feature captures changes in narrow frequency bands. Thus, the proposed “correction” method does not mask or remove the effects of structural changes in the response. Subsequent damage localization in cantilever-like structures is based on the changes of the drift of the fundamental mode shape amplitudes. The efficiency of the proposed two-step damage identification procedure is tested on a small-scale shear frame model in various damaged conditions. It is shown that the defined normalized cumulative power spectral density damage index is suitable to indicate most of the imposed damages, in particular when the proposed response correction methodology is applied. The spots of damage are successfully identified by the utilized mode shape damage indicator.