Damage Detection of Reinforced Concrete Bridge Considering Temperature Influence Based on AANN and FCM-PSO

Abstract

In order to eliminate the temperature influence on modal frequencies and identify structural damage in RC bridge accurately, a two-step damage identification technique is proposed without measuring temperature data. Firstly, based on the modal frequencies of undamaged bridge under varying temperature, the nonlinear principal component analysis using auto-associative neural network (AANN) is performed to extract the underlying mapping relationship. The difference between modal frequencies of damaged bridge and values reconstructed by AANN is used as damage identification index. Secondly, fuzzy C-mean clustering (FCM) algorithm improved by particle swarm optimization (PSO) algorithm (FCM-PSO) is used to identify damage existence and degree. The difference of modal frequencies from training cases are classified into several clusters based on FCM-PSO analysis, and the cluster centers for damage grades are constructed. Damage degree can be identified by calculating the fuzzy memberships between identification indicator vector and cluster centers of damage grades. This proposed technique is capable of identifying damage without measuring temperature data. Damage detection results show that the damage extent can be accurately evaluated after the damage is introduced.