Study On Bayesian Finite Element Model Updating Methods And Dynamic Reliability Analysis Methods For Isolated Structures

Aiming at the problems of poor accuracy and low efficiency of the existing dynamic reliability analysis methods for nonlinear isolated structures and inability of the finite element model of isolated structures to accurately reflect the true response of the structure,this paper proposes highly efficient and accurate dynamic reliability analysis methods and finite element model updating methods for isolated structures based on the orthogonal plane sampling method,equivalent plane method,Riemannian Manifold Hamiltonian Monte Carlo,BP neural networks and subset simulation.The method proposed in this paper was used to study the dynamic reliability problem of isolation structures with considering and without considering the randomness of isolation structures.The finite element models of isolation structures are updated based on structural measurement information,and the dynamic reliability of isolation structures was updated using the modified finite element model,some dynamic reliability analysis methods for isolated structures and some dynamic reliability analysis methods for existing isolated structures based on updated finite element models are obtained.The main research content are as follows:（1）To address the problems of low computational efficiency and poor accuracy of FORM-based structural dynamic reliability analysis method in solving the reliability problems of seismically isolated structures with a large number of random variables or a large degree of non-linearity of the limit state function,an OPS-FORM-based dynamic reliability analysis method for seismically isolated structures is proposed,which uses orthogonal plane sampling method to randomly generate samples located on the limit state response surface,and the secant hyperplane is constructed based on the random samples to modify the design point from FORM,and the structural dynamic reliability is evaluated according to the modified design point.In addition,the equivalent plane integral method is introduced to solve the multivariate normal distribution integral involved in the process of structural dynamic reliability solution,so as to improve the calculation efficiency based on OPS-FORM method.The dynamic reliability method based on OPS-FORM was used to calculate the failure probability of basic isolation structures under random seismic excitation,and the feasibility and accuracy of the proposed method were verified by comparing the calculation results with the FORM based method and MCS.（2）The OPS-FORM-based dynamic reliability analysis method for isolated structures is used to analyze the dynamic reliability of isolated structures with and without considering the randomness of structural parameters.The results showed that the failure probability of the structure with considering the randomness of structural parameters is greater than that without considering the randomness of structural parameters.If the structural safety performance evaluation is carried out according to this result,it can ensure that the structure can reserve a larger safety space.When evaluating the safety performance of isolated structures,it is necessary to consider the randomness of structural parameters.（3）To address the problem that the finite element models of isolated structures established based on the initial design values cannot reflect the real response of structures,a revised framework for the finite element model of the isolated structures based on Bayesian inference is established.To address the problem that the traditional sampling method has random wandering behavior and the probabilistic spatial explosion of Markov chains is low,which leads to sampling"stagnation"and slow convergence when solving,the Riemannian Manifold Hamiltonian sampling method is introduced to improve the spatial explosion of Markov chains,and the RMHMC-based Bayesian model parameter updating method is proposed.A BP neural network is introduced to establish a surrogate model of the parameter fit index function,and then the surrogate model is used for sampling to avoid time-consuming finite element calculations,further enhancing the computational efficiency of the proposed method.The parameters of the finite element model of a 7-storey base-isolated structure are updated using the method proposed in this paper.（4）In order to efficiently solve the dynamic reliability of isolated structures after updated finite element model,a RMHMC-SS-based dynamic reliability analysis method for isolated structures is proposed.Firstly,a framework of dynamic reliability of isolated structures based on subset simulation is established;on this basis,the Riemannian Manifold Hamiltonian Monte Carlo sampling method is introduced,and a sampling mechanism based on the Riemannian Manifold Hamiltonian Monte Carlo method for generating conditional samples is constructed,so as to eliminate the random walking behavior of the conventional MCMC method.To ensure a high acceptance rate of the conditional samples in the first subset,a method for determining the initial evolution time of Markov Chains is proposed.Based on the updated finite element model in Chapter 4,the feasibility and accuracy of the proposed method are verified.The results show that RMHMC-SS has high computational accuracy and the method is independent of the number of random variables and the shape of the probability space,and can be used to deal with structural dynamic reliability problems defined by Bayesian posterior distributions.（5）The RMHMC-SS-based structural system reliability analysis framework is established,and the principles of selecting the conditional failure probability₀ and the number of subset random samplesfor calculating the failure probability of the structural system using the RMHMC-SS method are discussed.The RMHMC-SS method is used to analyze the reliability of a 7-storey base-isolated structure system.The results show that the failure probability of the structure system is greater than the failure probability of the structure if only the isolation bearing is damaged or the superstructure is damaged,and that analyzing the reliability of the structure at the component level may underestimate the probability of failure of the structure.In order to ensure the safety and reliability of the isolated structure,it is necessary to analyze the reliability of the structure at the system level.