Omid Rasouli - Lecturer, Department of Civil Engineering, Miyaneh Faculty of Engineering, Miyaneh, Iran
Eysa Salajegheh - Professor, Department of Civil Engineering, University of Kerman, Kerman, Iran
Seyed Sadegh Naseralavi - Assistant Professor, Department of Civil Engineering, Vali-e-Asr University of Rafsanjan, Rafsanjan, Iran
Ali Kameli - Assistant Professor, Department of Industrial Engineering, Miyaneh Faculty of Engineering, Miyaneh, Iran

In this paper a method for damage identification in bridges employing neural networks is presented. In this work, in order to increase the speed and reduce the computational error for damage detection a newtwo stage method is introduced. In the first stage the damages are localized using a radial basis functionneural network which has the benefit of high learning speed. In the second stage, the exact location and severity of damaged elements found using a well-trained back propagation neural network which possesses high powerful learning capacity. In order to evaluate the proposed method Louisville trussbridge in United States of America is modeled by a finite element program and then changes in the responses is analyzed using MATLAB neural networks toolbox. Numerical results demonstrate the efficiency of the proposed method for correct damage identification

damage detection, neural networks, modal analysis, truss bridge