基于人工神经网络的城市桥梁震害评估方法  

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

作　　者：赵钊[1,2] 郭恩栋[1] 王再荣[1] 刘智[1] 

机构地区：[1]中国地震局工程力学研究所,黑龙江哈尔滨150080 [2]沈阳新大陆建筑设计有限公司,辽宁沈阳11000

出　　处：《世界地震工程》2011年第4期7-12,共6页World Earthquake Engineering

基　　金：地震行业科研专项经费项目(200808081)(201008005)(200808026)

摘　　要：为了使已有的桥梁震害评估方法适用于城市立交桥和高架桥,将桥梁结构是否规则考虑为评估模型的一项输入参数,并剔除了以往桥梁震害评估方法中不合理的影响因素以提高所提方法的可操作性。基于BP神经网络算法,以唐山地震和汶川地震中收集的54座梁式桥的震害资料作为网络训练样本,建立了梁式桥震害评估神经网络模型;通过讨论神经网络模型的泛化能力,得到了较满意的结果,表明该方法具有一定的准确度,可以应用于城市桥梁震害评估工作。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.

关 键 词：城市桥梁 震害评估 神经网络 汶川地震 桥梁震害 

分 类 号：X43[环境科学与工程—灾害防治]