Full-scale bridge damage identification using time series analysis of a dense array of geophones excited by drop weight

摘要

This paper presents a simple and inexpensive technique for damage identification of bridges using drop weight vibration data of bridges recorded by an array of geophones, highly sensitive sensors to record vibration, and time series analysis. The dynamic response of bridges obtained using drop weight as an excitation source is convolved with white noise to create suitable input for autoregressive (AR) models. A two-stage prediction model, combined AR and autoregressive with exogenous input (ARX), is employed to obtain a damage-sensitive feature. An outlier analysis method is developed based on the Monte Carlo simulation to identify the existence of damage. The proposed technique is verified using unique vibration data of two full-scale steel-girder bridges located on I-40 through downtown Knoxville, Tennessee, and subjected to progressive damage scenarios induced to steel girders. The results of the analysis for the vertical vibration data of the test bridges indicate that the proposed technique is able to detect the damage induced on the real bridge girders consistently even when the damage level is small and damage is located near a support; however, damage is not well localized or quantified in these two highly redundant bridges. Copyright © 2015 John Wiley & Sons, Ltd.