Ambient Vibration Based Damage Diagnosis Using Statistical Modal Filtering and Genetic Algorithm: A Bridge Case Study

Abstract

The authors recently developed a damage identification method which combines ambient vibration measurements and a Statistical Modal Filtering approach to predict the location and degree of damage. The method was then validated experimentally via ambient vibration tests conducted on full-scale reinforced concrete laboratory specimens. The main purpose of this paper is to demonstrate the feasibility of the identification method for a real bridge. An important challenge in this case is to overcome the absence of vibration measurements for the structure in its undamaged state which corresponds ideally to the reference state of the structure. The damage identification method is, therefore, modified to adapt it to the present situation where the intact state was not subjected to measurements. An additional refinement of the method consists of using a genetic algorithm to improve the computational efficiency of the damage localization method. This is particularly suited for a real case study where the number of damage parameters becomes significant. The damage diagnosis predictions suggest that the diagnosed bridge is damaged in four elements among a total of 168 elements with degrees of damage varying from 6% to 18%.