Time–frequency and wavelet-based study of an old steel truss bridge before and after retrofitting

Abstract

Signal-processing techniques have been widely used in structural health monitoring and nondestructive evaluation. Wavelet analysis, a relatively new mathematical and signal-processing tool, for damage detection in various civil and mechanical structures. It is a time–frequency analysis that provides more detailed information about nonstationary signals which traditional Fourier analysis miss. This paper aims to provide the damage identification in an existing 100-year-old deck-type steel truss bridge using-frequency- and time–frequency-based approaches. The dynamic testing of steel bridge was carried out using accelerometers for the damaged state and after partial retrofitting under similar environmental conditions and instrumental set up. The comparison is carried out using power spectral density, short-time Fourier transform, and wavelet packet transform with respect to both the upstream and the downstream trusses of the bridge. Higher and uniform dissipation of energy at resonating-frequency of the respective node after retrofitting showed intactness of joints. The variations of power spectral density in the first mode of the upstream and the downstream trusses clearly revealed improvements in the bridge signifying the importance of generating a signature of bridge before and after retrofitting. The status upgradations for the upstream and the downstream trusses obtained were different due to differential levels of damage in the bridge. Also, after retrofitting, the structural elemental behavior obtained was not the same as desired.