Study On The Dynamic Characteristics And Vortex-induced Vibration Performance Of Xihoumen Bridge Based On Structural Health Monitoring

With the development of bridge construction,large-span bridges have become an important research topic.As the span of the bridge increases,the structure becomes more sensitive to wind loads.Vortex-induced vibration(VIV)is one of the common wind-induced vibration diseases of bridges.Its occurrence mechanism involves fluidstructure coupling and nonlinear effects,which cannot be accurately predicted.With more than one ‘locked-in’ wind speed range,it can generation occurs and impact on structure and traffic,which cannot be ignored.The structure health monitoring system(SHM)provides a lot of data and new ideas for the study of VIV,but new challenges also follow.Based on the review of existing research,this paper studies the measured data of Xihoumen bridge in 2016.The vibration and wind speed data are used to study the dynamic characteristics and VIV performance of Xihoumen bridge.The main works and achievements of this paper are as follows:(1)Using finite element modeling(FEM)to analyze the vertical bending mode of the structure to obtain the theoretical frequency and vibration mode of the first few vertical bending modes;using the peak method to analysis the measured data.Consider the results from different locations and periods to get real mode.The results show that:the result of FEM is basically consistent with that of measured data.Due to the simplification processing and the algorithm the deviations still exist;in the measured data,due to the changing load and the relative position of the measuring point in the modal shape,it is necessary to analyze different data of each measuring point location in multiple time periods.All result should be considered comprehensively to obtain an accurate mode frequency.(2)Two numerical simulation tests are used to verify the accuracy and effectiveness of the AMD-RDT method in terms of damping ratio and frequency identification.Based on the measured data,this method is used to analyze the mode damping and frequency.The results show that: the identification result of the structural modal frequency is less affected by the change of wind speed,and the damping ratio is sensitive to the wind speed.In the case of high wind speed,the structural damping ratio identification result is inaccurate due to structural vibration and interaction with the incoming wind field.(3)Based on the environmental vibration data,the structural damping ratio is automatically identified twice,and the recognition results showed a large discrete state;then the deviations in the recognition results is analyzed.It is found that the wind speed and frequency distribution have a greater impact on the recognition results.Since the automatic recognition method cannot effectively choose the data,several criteria are summarized for selecting the appropriate data manually to analyze the damping ratio.Based on the manual work,a more accurate damping ratio is obtained,and it is found that the first-order mode differs at different measurement points.The data selection criteria are as follows: the wind speed should be less than 3m/s;for loworder modes,there should be two or more obvious peaks in the power spectrum at the same time,and the selected frequency should not much greater or smaller than other peaks;for high-order modes,the obvious peak of this frequency should be observed in the power spectrum;the AMD-RDT signal of the selected mode should conform to the characteristics of single-degree-freedom free vibration,and the fitted signal should not have obvious deviation from the original signal;the data of stagnation points should be avoid.(4)Taking advantage of the differences in the energy distribution of each mode in the two signals between the environmental vibration and VIV,the vortex vibration automatic recognition method constructed by the neural network technology is used to identify the VIV.This method takes the power spectrum distribution of the environmental vibration data of Xihoumen Bridge as input data,trains the BP neural network model,uses novel detection technology to design novel indicators for the network output data,and calculates the threshold.When the data of VIV is input to the network,the output data is obtained with a novelty index greater than the threshold.The results show that: more than 80 vortex-induced vibration events are obtained from the combined results of three different measurement points,and the first nineth-order vertical bending modes,including the seventh-order mode,are found;the wind direction at the periods of VIV is basically perpendicular to the direction of the bridge axis and each order of vortex vibration occurs when the degree of turbulence is less than 15%,but the different modal orders of VIV corresponds to different wind speed intervals.