Research On Modal Identification And Model Updating Of Bridge Structure Based On Dynamic Response

Bridges are the hinges of the transportation system.The safety of bridges has always been the main concern of the government departments and the public.In order to ensure the safe operation of bridges,many important bridge structures have installed health monitoring system.At present,the latest health monitoring system research is focused on the online structural modal identification,the online finite element model updating,and the online simulation.How to quickly as well as correctly identify the modal properties is a critical issue for online identification.Beside this,how to avoid the use of degree of freedom(DOF)matching technique,and how to update the damping related parameters,and how to improve the model updating efficiency are the issues in urgent need of study.Based on the characteristics of bridge structures,the following aspects related to the modal identification and finite element model updating have been studied in this thesis.(1)Based on the existing literature,the research status of the modal parameter identification algorithms has been summarized.The eigen-system realization algorithm(ERA)has been mainly studied and improved.A fast eigen-system realization algorithm(FERA)is proposed in this thesis,which uses the eigenvalue decomposition instead of the singular value decomposition(SVD)to save the calculation time and storage compared with the ERA.In addition,the selection rules for the number of rows and columns of the Hankel matrix in the FERA are discussed.In order to verify this method,a four-story steel structure with the precise size and known material properties was designed.The FERA method has been used for the modal property identification from the hammer tests.The results show that the FERA can obtain the same accuracy modal properties as the ERA,and the calulation speed can be greatly improved compared with the ERA.In addtion,the FERA was used for extracting modal parameters of an in-service pedestrian bridge through hammer test.The results show that FERA is also applicable to modal parameter identification of actual bridge structures and has good anti-noise ability and fault tolerance.(2)Most of the time domain modal parameter identification methods have the problems in model order determination and the spurious modes identification.Firstly,the common methods of model order determination based on singular value decomposition were introduced,and then a singular value angle change model order determination method with better antinoise performance is proposed for the FERA method.Mode accuracy indicators are also employed together for accurate identification of modal parameters.The modal identification process of the four-story steel structure and pedestrian bridge have proved that this method can effectively reduce the mode omission and eliminate the false modes.(3)To deal with the problem of noise interference and inadequate excitation in the operational modal test of a bridge,a multi-reference DOFs modal identification based on NExT + FERA method is proposed.By setting different reference DOFs in each data set,NExT+ERA is utilized to identify modal parameters.At the same time,using mode accuracy indicators to distinguish the true modes from the spurious modes,and then using frequency and damping ratio to cluster the identified modes.It can help obtain the most reliable modal parameters and prevent the mode absence effectively.The selection rules for the key parameters in this method are also discussed.Finally,the method has been verified through the white noise excitation test of the four-story structure.After appling this method on the natural excitation test of the pedestrian bridge,it is found that the modal identification results obtained through this method can get the even higher mode accuracy indicators than the identification reuslts through the modal hammer test.(4)In the finite element model updating section,the traditional frequency response function(FRF)analytical sensitivity-based approach is first studied.The Rayleigh damping model is suggested to take into account in the conventional method.The partial derivatives of the dynamic stiffness matrix to damping updating parameters in the FRF sensitivity matrix are derived.The commonly used DOF matching techniques are summarized for the traditional methods.Besides that,several solving technique of the traditional method are introduced,as well as the frequency point selection technique.In the final,the traditional approach has been verified by the four-story structure.However,the results show that it is difficult to modify all parameters of mass and stiffness simultaneously,and the coupling phenomenon will lead to the unreasonable model updating results.(5)A new model updating approach,the numerical sensitivity-based FRF residual approach,is proposed.The basic idea of this method is to decouple the analytical matrix expression of FRF to the individual expression,so that the analytical FRF can directly match with the measured FRF one by one;then,the updating parameter values can be obtained by minimizing the amplitude differences between the analytical FRFs and test FRFs through sateof-the-art optimization method.In this paper,the effect of non-proportional damping on the model updating of FRF is studied for the first time.As comparison,the influence on model updating parameters are studied by assuming different damping models.How to select the appropriate frequency bands for model updating and how to consider the accuracy of the analytical FRF expression are discussed.In addition,the analytical formulation of the FRF which suitable for shaking table test is derived in this paper.Finally,the algorithm has been validated through the four-story structure and successfully applied to the damage detection of a simply supported beam.(6)The Kriging model is introduced into the finite element(FE)model updating of the bridge structure,which avoids the must use of the FE model in each iteration of the tranditional model updating.The commonly used design of experiment(DOE)methods are summarized.Then,a numerical simulation of a simply supported beam was conducted to verify the Kriging model method.Comparing with the eigen-sensitivity method,Kriging model method is found to be more time efficiency.In the final,the Kriging model updating method is successfully applied to the pedestrian bridge.The physical parameters and boundary conditions are all updated.Finally,the basic conclusions are summarized and the outlook for future research is also made.