基于神经网络的混凝土斜拉桥健康状态评估  

Health Condition Assessment of Concrete Cable-stayed Bridge Based on ANN

作　　者：尚鑫[1,2] 徐岳[1] 马保林[3] 王春生[1] 

机构地区：[1]长安大学旧桥检测与加固技术交通行业重点实验室(西安),西安710064 [2]西安公路研究院,西安710054 [3]陕西省公路勘察设计院,西安710068

出　　处：《武汉理工大学学报》2011年第8期79-83,共5页Journal of Wuhan University of Technology

基　　金：国家西部交通建设科技项目(200531822316)

摘　　要：根据混凝土斜拉桥的结构特点,以双连体独塔斜拉桥为工程依托,通过建立实桥动力有限元模型,以模态分析结果为桥梁健康状态评估提供数值模拟试验模型及试验数据。借助Matlab图形用户界面,选择神经网络参数(网络的类型、网络拓扑结构、各层传递函数、特征参数、数据前处理方法及训练样本个数),构建BP神经网络,在对混凝土斜拉桥结构健康状态评估模型研究的基础上,提出了基于人工神经网络的评估技术,并通过实桥应用,完成对样本矢量的输入,以及对所建网络的训练,为实现混凝土斜拉桥健康状态评估提供合适的工具。结果表明,该方法能够有效反映在用混凝土斜拉桥结构的健康状态。According to the characteristics of concrete cable-stayed bridge and the data of a conjoined twin cable-stayed bridges sharing a single tower,the 3-dementional finite-element model（3D FEM） of the bridge is set up using structure analysis software,and the modes are analyzed,so the numerical simulating mode and test data for the bridge health evaluation is available.With Graphical User Interfaces（GUI） of Matlab software,artificial neural network（ANN） parameters（such as ANN type,web topology structure,transfer function in every layer,stamp parameter,data pretreatment and quantity of training examples） are selected and the BPANN is built.According to the research on concrete cable-stayed bridge structure health condition assessment,the calculation process based on ANN is presented and is used in the conjoined twin cable-stayed bridges,and then the examples vectors are input and the constructed network is trained,so that a suitable assessment tool for concrete cable-stayed bridge is presented and the network is effective.The result shows that the exiting concrete cable-stayed bridge structure condition can be reflected effectually by this kind of assessment method.

关 键 词：混凝土斜拉桥 有限元模型 模态分析 健康状态 人工神经网络 分类评估 

分 类 号：U446.3[建筑科学—桥梁与隧道工程]