Identification Of Large-Scale Structures By Unscented Kalman Filter With Unknown Input And System-level Seismic Fragility Analysis Of Bridges By Optimized Vine Copula

With the development of large-scale structural engineering and the increasing attention on earthquake disasters,higher requirements for the safety of engineering structures is appealed.The vibration response of large structures is determined by external excitation and structural parameters.In order to study the dynamic performance of large structure systems,the identification of parameters and unknown excitation,the importance of uncertain parameters and the fragility of structural systems need to be studied.According to the above requirements,the system and excitation identification of large structures are studied from linear to nonlinear.An efficient importance analysis of random parameters is proposed and the random parameters of bridge model are screened out according to their importance order.Finally,the seismic fragility of the bridge system is analyzed.The research contents of this paper mainly include the following four parts:1.In view of the problem that the identification of existing large structural systems requires full observation or the assumption of unknown external excitation,the UKFUI(unscented Kalman filter with unknown input,UKF-UI)which were proposed by our team is extended to investigate the assessment of damage and UI for large-scale structures.The unknown external excitations can be estimated by recursive nonlinear least squares of the errors between the true measurements and the predicted ones,so as to achieve the simultaneous identification of UI and parameters.A 2 dimentional and a 3 dimentional finite element(FE)frame models are taken as validation of the method.By the partially measured noise-polluted structural acceleration and displacement responses,the extent and location of damage for the large structures are assessed at the element level.The unknown external excitations are simultaneously identified without any assumptions on the time evolutions.2.A distributed plastic finite element model updating method based on substructure UKF-UI was proposed to transform the bridge structure into the parallel identification of several substructures.Through the Matlab-OpenSees co-simulation platform,the material constitutive parameters of substructures can be transformed in Matlab,and then transmitted to OpenSees,which rebuilds the model according to the modified parameters and updates the model in real time.The interface force between substructures can be taken as the unknown excitation,and the error between the updated model response and the actual observed response was obtained by nonlinear least squares.A damaged distributed plastic finite element model of a three-span continuous girder bridge is used to verify the effectiveness of the method,which overcomes the problem that the structural parameters of the multi-parameter distributed plastic finite element structure are difficult to converge.3.This dissertation presents a computational efficient global sensitivity analysis based on a probability assessment method to investigate the importance of the random parameters in the analyses of bridge seismic demands.In order to rapidly estimate the probability density function(PDF)of random parameters and avoid the large amount of finite element time history analysis generated by Monte Carlo(MC)method,A modified three-point-estimate method is derived from Rosenblueth’s two-pointestimate method.The shifted generalized lognormal distribution method is adopted to estimate the unconditional and conditional probability density functions(PDF)of seismic demands which are used for the moment-independent importance analysis and the entropy-based importance analysis.The two kinds of importance measures of the random material and structural parameters are estimated by only several times of nonlinear time history analyses at the point-estimate sampling points,and importance coefficients of random materials and structural parameters for seismic requirements of piers,bearings and abutments by the proposed method are compared with those by MC simulation.4.In order to consider the correlation between different component indeviduals in the bridge system fragility,this paper introduces the vine Copula to system fragility analysis.Based on the rank correlation coefficient between seismic demand samples of different components,the optimal vine Copula function is obtained by using the shortest Hamiltonian path method,which avoids a lot of Copula function solving and trial calculation of information determination criteria.The component fragilities and their complement are obtained by probabilistic seismic demand analysis,and the joint PDF of component fragility complements is described by the vine Copula function.Finally,the expression of the bridge system fragility is derived.The joint probability of the complements can avoid solving the joint PDF of different combinations of components.An application example of a damaged four-span continuous girder bridge is studied.With the method of Chapter 3,bridge seismic damage are simulated and with the research achievements of the Chapter 4,the screening of the important parameters of 10 are taken as random parameters.The proposed method is used to analyze the fragility of a bridge system with dependent ten components.The results are compared with those obtained by MC method and the traditional fragility method considering the correlation between component classes.