Research On Damage Detection Of Beam-type Structure By Using Bridge Indirect Measurement Method

With frequent collapse of engineering structures, and other disaster problems caused by earthquakes, landslides etc., researchers are paying more attention to the structure detection and monitoring. In this thesis, based on a research hotspot--indirect measurement method of bridge, a simply supported beam and a two-span continuous beam were used for research. First, vehicle-bridge coupled model was established through VBI(Vehicle-Bridge Interaction) element to extract dynamic response from the test vehicle. Then, mode shapes of the bridge were extracted from the acceration response of the test vehicle through various signal processing technologies, including the Fourier transform, band-pass filter based singular spectrum analysis algorithm, Hilbert transform etc. At the end of the thesis, the improved direct stiffness method(frequency domain) method was applied to calculate the bridge node stiffness, and so the variation of bending stiffness can be determined for assessment the damage location and damage degree preliminary; further on the basis of statistical moment theory, a bridge was divided into a number of test units, the test vehicle displacement fourth-order moments vector index(time domain) was put forward to perform damage diagnosis through the model update method.The main works and corresponding analysis results are summarized as follows:(1) Based on a simply supported beam and a two-span continuous beam, the indirect measurement method was deduced theoretically. The vehicle-bridge coupled models of the simple beam and two-span continuous beam were established through VBI element by using Matlab to extract the vehicle dynamic response correctly. Meanwhile, the finite elements were modeled by ABAQUS software to realize the indirect measuring method, which establishes the foundation research on complicated models of the bridge in future.(2)Constructing the mode shapes of the bridge from a passing vehicle was studied theoretically. Signal processing technologies, e.g., band-pass filter based singular spectrum analysis algorithm and Hilbert transform, were adopted to extract the low order mode shapes of the bridge. Considering measurement noise, dynamic parameters of the test vehicle, the length of the filter window, the speed of the test vehicle, the random traffic flow and other factors, the extraction of the 1st mode shape was studied. A simple beam and a two-span continuous beam were utilized to simulate respectively. Simulation results showed that when the environment noise level is less than 20% or the signal-to-noise ratio(SNR) is more than 20 dB, the 1st mode shape could be obtained accurately; the faster the speed of the test vehicles, the smaller accuracy the 1st mode shape is; when the ratio of the test vehicle's speed to the span length of the bridge was less than 0.25, the 1st mode shape would be accurate; it should avoid resonance between the test vehicle and bridge; with random cars driving on the bridge, 1st mode shape with indirect measurement method can be also used; for the selection of window length by using filtering technology, it should try to ensure the balance of the direction of the ranks and columns in the trajectory matrix.(3)Using the improved direct stiffness method(IDSM) to identify the bending stiffness of the bridge, which was embed into the OpenSeesNavigator System Identification Toolbox. Based on the extracted mode shape by the indirect measurement method, the bending stiffness can be calculated by the IDSM, the damage location and damage degree can be preliminarily evaluated by comparing the changes of node stiffness. A simple beam and a two-span continuous beam were utilized to simulate respectively. Simulation results showed that when the ratio of the test vehicle's speed to the span length of the bridge was less than 0.2, the identified results of stiffness could be relatively precise; the method proposed herein can preliminarily diagnosis damage cases with different locations and damage degree; under the influence of noise, identified damage areas would be expanded, but it was still concentrated in the actual damage location. Without blocking the passage of traffic on the bridge, frequency domain identification method proposed herein could be still performing online diagnosis to a certain extent. Although there were still many errors between the identification results and the real damage, the identified damage areas were still concentrated in the area of assuming position, and so the scope of manual detection in the field test could be reduced to some extents.(4)The tested bridge was divided into a number of units, and then displacement fourth-order moments vector index was put forward to perform damage diagnosis. Through the finite element model updating method, the element stiffness of every tested unit would be identified. Comparing the change of the element stiffness, the damage location and damage degree could be evaluated. A simple beam and a two-span continuous beam were simulated respectively. Simulation results showed that with different bridge damping ratio, the method proposed herein could accurately identify the damage location and damage degree of the bridge, but with the increase of noise level, it may cause mistakes in the boundary and joint part elements of the bridge.