A Comparative Study On Damage Detection For Bridge Structure Based On Operational Modal Analysis

The main task of bridge health monitoring system is how to obtain the characteristic information of the structure from massive data in order to determine operational status of the bridge structure.For bridge health monitoring system, optimal sensor placement under the limited measuring points is the basis, modal parameter identification under ambient excitation is the core and damage identification method is the key to the research. This paper carries out research, development and application of the in-depth system,focusing on the three common key issues presented above of bridge health monitoring system.In view of the sensor optimization part, this thesis mainly describes the basic idea of eight kinds of classic optimal sensor placement methods, elaborates the technical characteristics of each method and calculation process systematically and summarizes the evolution and links between the various algorithms, then codes the corresponding program and provides basis for modal identification; For modal identification part, this thesis introduces two methods of signal pre-processing in detail and the basic theories, implementation process and application scope of six methods of operating modal parameter identification and summarizes technical characteristics of the various algorithms and numerical methods. Taking numerical analysis algorithms as a starting point, it discusses the recognition accuracy,the application scope and anti-noise performance of different algorithms. In the section of damage identification,based on damage identification methods,in-depth study of modal is been discussed, mainly including the basis of frequency, curvature mode and flexibility matrix,and it discusses the application scope and anti-noise performance of each damage index.However, currently, there is no existing of health monitoring data analysis software,and one of the few commercial software are black box models, which can not be customized; more crucial is, they can not meet the requirements of integration, automation and real-time among the bridge health monitoring system because of artificial intervention. With the platform of MATLAB, we need to develop the tool boxes of modal identification and damage identification. The function of tool box of modal identification,which can proceed modal parameter identification in real time, covers six classical modal identification algorithms, with each algorithm to which the function of parameter debugging. The tool box of damage identification, which clearly shows the location of structural damage tersely, has three major kinds of damage index algorithm function which is based on frequency, curvature mode and flexibility matrix. The user can directly use the tool box of modal analysis to do modal analysis and researches in real time and to find the damage location by using the tool box of damage identification.In the last part, we will use the tool boxes developed here to conduct modal identifications on a three-span continuous bridge by eight kinds of sensor optimal placement algorithm, and conduct parametric analysis aimed at influencing factors of three kinds of hybrid algorithm; the results show that EI method is of the worst, the effect of the progressive reduction method of the most stable one and the performance of hybrid algorithm is related with the selected parameters;with six kinds of modal recognition algorithm for modal identification,the results prove that among the modal identification methods, FDD method is of the best noise resistance, SSI method and ERA method are of the highest accuracy of identification.By the calculation of damage identification,it is shown that the damage index MFC is of the best noise resistance. In order to further verify the reliability and applicability of the program, we conduct modal analysis on the Forth Nanjing Yangtze bridge with the tool boxes of modal analysis. The results show that the tool boxes developed in this thesis do have a high practical application value.