\r熊 文¹，张 愉¹，李飞泉²，侯训田²，沈旭东\r³\r
\r
AuthorsHTML:\r熊 文¹，张 愉¹，李飞泉²，侯训田²，沈旭东\r³\r
\r
AuthorsListE:\rXiong Wen¹，Zhang Yu¹，Li Feiquan²，Hou Xuntian²，Shen Xudong\r³\r
\r
AuthorsHTMLE:\rXiong Wen¹，Zhang Yu¹，Li Feiquan²，Hou Xuntian²，Shen Xudong\r³\r
\r
Unit:\r\r1. 东南大学交通学院，南京 211189；\r
\r\r2. 浙江省交通运输厅，杭州 310009；
\r
\r3. 浙江省交通规划设计研究院，杭州 310006\r
\r
Unit_EngLish:\r1. School of Transportation，Southeast University，Nanjing 211189，China；
2. Zhejiang Provincial Transport Department，Hangzhou 310009，China；
3. Zhejiang Provincial Institute of Communications Planning，Design and Research，Hangzhou 310006，China\r
\r
Abstract_Chinese:\r\r桥梁基础水文作用是引起桥梁结构功能及安全性能失效的首要原因，而桥梁冲刷又是水文作用最主要的表现形式．依托舟山大陆连岛工程金塘大桥主通航孔桥，提出一种基于时频分析与神经网络的桥梁冲刷动力评估方法．首先，利用白噪声地震波模拟环境振动激励，采用动力时程分析法模拟环境激励下的上部结构振动，并获得其加速度响应，同时在振动过程中实时模拟基础冲刷深度的连续发展，进行环境振动激励下桥梁模态参数与冲刷发展的关联性分析，研究该方法的识别敏感性．进而，利用数值动力参数分析研究上部结构既有局部损伤多种组合形式对桥梁结构模态参数的扰动程度，以确定非冲刷损伤形式对桥梁冲刷动力评估方法准确性的干扰．最后，以模态分析中的低阶模态参数作为样本输入，以冲刷深度与墩位的不同组合作为样本输出，建立\rBP\r神经网络，实现桥梁冲刷深度与墩位的同步评估．研究结果表明：利用自振频率构建的动力指纹对冲刷发展较为敏感；该方法无论对冲刷深度还是冲刷墩位均具有较高的识别准确度，且可忽略有限局部损伤对识别准确性的干扰．\r该方法不需要水下操作，不需要昂贵的测试设备，仅需要加速度传感器以及数据采集装置，便于融入常规桥梁检测项目中．\r\r
\r
Abstract_English:\r\rThe hydrological effect on the bridge foundation is the main reason for the loss of function and safety of bridges\r．\rBridge scour is the most important form of hydrological effect\r．\rA vibration-based bridge scour identification method was proposed based on \rtime-frequency analysis and neural network\r，\rwhich was applied in the Jintang Bridge of Zhoushan Island Connection Project\r．\rInitially\r，\rthe environmental effects on the bridge were simulated by inputting earthquake waves as the white noise\r．\rThe environment-induced vibration of the superstructure was simulated by the dynamic \r\r\rtime-procedure analysis\r\r\r，\rand structural acceleration and scour depth were simultaneously recorded\r．\rA correlation study was then conducted between the modal parameters and scour development to investigate the identification sensitivity\r．\rThe interference on the modal parameter test from different local damage combinations in superstructures was further studied by using simulated parametric analysis\r，\rthereby determining the interference from non-scour damage on the proposed scour identification method\r．\rA backpropagation\r(\rBP\r)\r\r\rneural network\r was finally established by setting the low-order modal parameters and the combinations of scour depth and foundation as the input and output\r，\rrespectively\r．\rBoth the scour depth and corresponding pier state can be correctly identified by using this BP network\r．\rThus\r，\rthe frequency-based dynamic \rfingerprint shows a high sensitivity to scour development\r．\rThe proposed method can provide accurate identification results for both the scour depth and pier state without considering the interference from local damage\r．\rThe proposed method can be conveniently integrated into a routine bridge inspection with acceleration sensors and a data acquisition system instead of underwater work and expensive devices\r．\r\r
\r
Keyword_Chinese:桥梁冲刷；时频分析；神经网络；自振频率；动力评估\r

Keywords_English:bridge scour；time-frequency analysis；neural network；natural frequency of vibration；dynamic assessment\r


PDF全文下载地址:http://xbzrb.tju.edu.cn/#/digest?ArticleID=6439