Seismic damage evaluation method of urban bridge based on artificial neural network

In order to adjust the existent methodology for seismic evaluation of bridges to the assessment of overpass and viaduct,as a parameter,the regularity of the bridge was considered in the assessment model.For the purpose of improving practicality of the methodology,improper influence factors which are taken into account in the existent methodology were removed.A three-layer BP neural network was built,and then trained with data of 54 bridges which were collected from literature of 1976 Tangshan earthquake and 2008 Wenchuan earthquake.The capacity of network generalization was discussed and the conclusion is that this methodology for seismic evaluation of urban bridges has proper accuracy and can be operated in practice.