Damage identification in steel girder bridges using modal strain energy-based damage index method and artificial neural network

Abstract

Health monitoring of infrastructures is of utmost significance, as they play a vital role in transportation. Identifying damages at the onset is essential to ensure the structure’s health state. The vibration-based damage identification methods use the vibration characteristics of structures to detect damages. Also, artificial neural networks (ANN) have been used to estimate damage magnitude with varied success. In this study, a two-stage damage identification technique was proposed to locate and estimate damages in steel girder bridges. The performance and feasibility of the proposed method were evaluated by applying several single and multiple damage scenarios to a validated finite element (FE) model of I-40 Bridge. Initially, the damage location was determined using the modal strain energy-based damage index method. For this purpose, the damage index was calculated separately for the first three bending modes of the bridge. The calculated damage vectors were combined along the length of girders. The peak value of the combined damage index showed the damage location along the length of steel plate girders. The results revealed that all the damage locations were detected with appropriate accuracy even if multiple damage scenarios existed in plate girders. In the second step, an ANN was used for estimation of damage severity. The damage index was used as input parameters to train the ANN. Then, the trained ANN could predict the severity of structural damages. The results indicated appropriate accuracy proposed by ANN for estimating the damage magnitude as well as proper performance of the proposed method.