Research On Structural Damage Identification Of Railway Station House Based On Bayesian Method

The structure will be damaged by environmental erosion,material aging and other factors during service.In this case,real-time monitoring is needed through health monitoring,and damage identification is an important part of structural health monitoring.The present damage identification methods are mainly the traditional deterministic damage identification methods.However,in practical engineering applications,the factors that cause structural damage or the calculation process and results of damage identification are often uncertain,so the deterministic damage identification methods may not fully consider these influences.Therefore,this thesis uses an uncertain damage identification method based on Bayesian model,and combined with stochastic response surface method and Markov Monte Carlo sampling method.Thus,the influence of uncertainty factors on damage identification can be considered,and the calculation of eigenvalues of model response is simplified in the sampling process to improve the calculation efficiency.The main research contents of this thesis include:(1)Aiming at the solution of the likelihood function in Bayesian damage identification method,this thesis uses matrix perturbation method,least square method and random response surface method to solve the problem,so that the relationship between structural response and model parameters can be obtained.And then,establishing a finite element model to verify the three methods.By comparing the errors of the three methods and the feasibility of the methods,it is concluded that the stochastic response surface method can obtain a small error when solving the relationship between the structural response and the model parameters,which can be used in the subsequent calculation.(2)Use finite element simulation to carried out the numerical simulation of the damage identification method based on Bayesian method,compare the experimental results with the theoretical errors to verify the feasibility.Firstly,introduce the Bayesian statistics theory mathematically,then,introduce the theory into structural damage identification,and combine the stochastic response surface method to derive the posterior probability density function of the parameters to be modified,and obtain the posterior distribution of the parameters by the Metropolis-Hastings sampling algorithm.Thus,the damage identification process based on Bayes is completed.Then establish the finite element model,substitute the calculation process into the finite element model to make numerical simulation calculation,and verify the feasibility of the damage identification method based on Bayesian method.Finally,the posterior distribution of the parameters to be modified can be obtained.(3)Introduce the finite element model of railway station house,verify the feasibility of the damage identification method based on the Bayesian method in the actual engineering structure health monitoring through the damage identification of the highly sensitive elements in the rail bearing layer of the station model,and we can obtain the relatively accurate results.The results show that the Bayesian damage identification method combined with the stochastic response surface method can identify the damage relatively accurate and simple,and is applicable to different models,and can be applied to practical engineering.