Damage Detection in Bridge Structure Using Vibration Data under Random Travelling Vehicle Loads

Abstract

Due to the random nature of the road excitation and the inherent uncertainties in bridge-vehicle system, damage identification of bridge structure through continuous monitoring under operating situation become a challenge problem. Methods for system identification and damage detection of a continuous two-span concrete bridge structure in time domain is presented using interaction forces from random moving vehicles as excitation. The signals recorded in different locations of the instrumented bridge are mixed with signals from different internal and external (road roughness) vibration sources. The damage structure is also modelled as the stiffness reduction in one of the beam element. For the purpose of system identification and damage detection three different output-only modal analysis techniques are proposed: The covariance-driven stochastic subspace identification (SSI-COV), the blind source separation algorithms (called Second Order Blind Identification) and the multivariate AR model. The advantages and disadvantages of the three algorithms are discussed. Finally, the null-space damage index, subspace damage indices and mode shape slope change are used to detect and locate the damage. The proposed approaches has been tested in simulation and proved to be effective for structural health monitoring.