Condition evaluation for existing reinforced concrete bridge superstructure using fuzzy clustering improved by particle swarm optimisation

Abstract

Condition evaluations of old bridges are necessary for determining their health states and providing priority levels of maintenance. In this paper, a novel condition evaluation approach for reinforced concrete (RC) bridge superstructure is presented based on fuzzy c-mean clustering optimised by particle swarm optimisation (FCM-PSO) algorithm. It is equipped with the advantages of PSO algorithm in global optimisation and FCM algorithm in convergence acceleration, which greatly improves the effectiveness of clustering. Using this methodology, a reliable evaluation index system and a number of training samples from field-measured data of existing old bridges are prerequisites. In addition, the optimal cluster number for training samples can be determined by Xie–Beni validity evaluation index. Subsequently, condition grades and corresponding cluster centres can be determined based on the calculation of cluster centres and membership matrix for training samples. On the above basis, bridge conditions of testing samples can be evaluated based on the fuzzy membership to the cluster centres of condition grades. A case study was carried out to verify the feasibility and effectiveness of the proposed FCM-PSO method. Evaluation results reveal that the proposed method can effectively reduce the influence of subjective factors and will be favourable for condition evaluation of existing RC bridges.