Indirect Bridge Health Monitoring Using Time-Frequency Analysis: Analytical and Experimental Studies

Abstract

AbstractBridges are among the crucial elements of public infrastructure and are inspected regularly for maintenance purposes. Often, these inspections are conducted visually, which can be particularly limited to detecting hidden and minor damage, for instance, fatigue cracks, delamination, and corrosion of embedded reinforcement. Ideally, the bridge inspectors need to identify any changes in dynamic parameters of the bridge, such as natural frequencies, damping ratio, and stiffness. Recently, there has been a shift from using fixed sensor networks to moving sensor networks that can detect changes in these dynamic parameters. Moving sensor networks rely on indirect measurements taken from within the vehicle while traveling over the bridge. The signal collected from within a passing vehicle contains the bridge’s structural response, vehicle suspension input, and surface roughness-induced vibrations. This paper investigated the feasibility of drive-by bridge monitoring using numerical and experimental assessments and addressed their challenges using the time-frequency method. The proposed methodology uses Wavelet Packet Transform (WPT), which extracted modal responses and delineated the bridge frequency components from the driving and vehicle frequencies using the wavelet packet coefficients. The performance of the proposed method was validated using both numerical simulations and a laboratory experiment. The effects of vehicle parameters on vehicle acceleration response were studied using analytical modeling. In the laboratory experiment, a moving cart was used as a vehicle traveling over a scaled bridge model. The results demonstrated that the proposed method could efficiently extract and separate the bridge dynamics from the vehicle response.KeywordsWavelet packet transformVehicle-bridge interactionIndirect bridge health monitoringTime-frequency method