Damage Detection of Bridge Structures Under Vehicle Loads Using Moving-Window Correlation Coefficient of Signals with Limited Sensors

Abstract

Based on the moving-window correlation coefficient of signals, a damage detection method is proposed for bridge structures. The signals collected by two sensors during the vehicle crossing the bridge are intercepted as windowed pair time series, and the damage index is defined based on the correlation of them. The damage vector is composed of damage indices from the beginning to end according to the moving direction of the window, and the damage location can be determined by the peak information of all the damage vectors which are obtained by different sensor pairs. First, the damage detection method and corresponding detection steps are introduced. Then, the proposed method is validated by numerical simulation, and the influence of moving window length, moving step, the number of sensors, deviation of vehicle parameters and road-surface roughness on damage localization are discussed, respectively. Finally, the proposed method is validated by experiments using a two-axis vehicle and a steel–concrete composite beam. The results show that the proposed method can effectively identify the damage location. High-pass filtering of the raw data to remove trend items and wavelet noise reduction can significantly improve the accuracy of damage detection, and can accurately locate the damage in the presence of road-surface roughness. The damage detection result is not sensitive to vehicle parameter deviation, and the damage location can still be accurately identified when the vehicle weight or speed deviates 50% from the vehicle parameters in the reference data. In addition, the proposed method can make full use of the damage information between sensors, and can accurately identify the damage location with limited sensors. The number of sensors can be flexibly determined considering both accuracy and economy.