苏通大桥结构健康状态评估技术研究与应用(1):拉索损伤识别  

Technology and Application of Structural Health Condition Assessment for Sutong Bridge (1): Damage Identification of Stay Cable

作　　者：杨杰[1,2] 李爱群[1] 丁幼亮[1] 姚蓓[3] 杨军[3] 张维苏[3] 

机构地区：[1]东南大学混凝土与预应力混凝土结构教育部重点实验室,南京210096 [2]南京航空航天大学土木工程系,南京210016 [3]江苏省苏通大桥建设指挥部,南京210006

出　　处：《防灾减灾工程学报》2010年第3期325-329,共5页Journal of Disaster Prevention and Mitigation Engineering

基　　金：国家杰出青年科学基金项目(50725828);国家"十一五"科技支撑计划项目(2006BAJ03B05);国家航空科学基金项目(2008ZD52040)资助

摘　　要：采用改进的RBF神经网络建立了苏通大桥拉索损伤识别方法。2个不同阶固有频率之比是仅与损伤位置有关的结构振动参数,据此定义了用于损伤定位的损伤特征指标,并用其来训练神经网络;提出了基于R2+准则与Jackknife校验的改进RBF算法,以有效地控制RBF网络的过拟合现象。算例结果表明,所提出的方法可以较好地对苏通大桥斜拉索进行损伤识别。In this paper the damage identification method for stay cable of Sutong Bridge is established using the modified RBF neural networks. The ratio of two different frequencies of the main girder is a vibration character which varies only with the damage location. It is selected as damage analysis index to train the neural networks. And in order to control the over-fitting of RBF neural networks, a new modified algorithm is presented based on R~_ rule and Jackknife validation. The analysis results show that the presented method can effectively locate the cable damage of Sutong Bridge.

关 键 词：斜拉桥 结构损伤识别 神经网络 斜拉索 

分 类 号：TU435[建筑科学—岩土工程]