Structural Model Updating Based On Modal Features And Modified Firefly Algorithm

In recent decades,large-scale infrastructures represented by bridges have emerged rapidly in China,and the health monitoring of bridge structures has gradually attracted more and more attention.As an important part of bridge health monitoring,finite element model updating provides necessary technical means and information support for condition assessment,response prediction and management decision of bridge structures.Due to the extensive existence of uncertainty in bridge monitoring process,the incompleteness of observation data and the complexity of actual structure model,the finite element model updating still faces many challenges.This thesis investigates optimization solution,uncertainty quantification and complex model approximation in the process of finite element model updating.The main contents and conclusions are as follows:(1)Finite element model updating can be transformed into an optimization problem.In order to make the optimization results more accurate and reliable,an improved firefly algorithm is proposed.Based on the original firefly global optimization algorithm,the Nelder-Mead local optimization algorithm is introduced to improve the local optimization ability of the algorithm,and the concept of diversity threshold is introduced to control the startup time of the local algorithm,so as to achieve a balance between the exploration and exploitation performance of the algorithm as much as possible.In addition,necessary improvement measures are proposed for the random step size,attraction coefficient and firefly individual boundary control mechanism of the algorithm,which avoids the problems of prematurity and non-convergence of the original firefly algorithm.The improved algorithm is compared with the original firefly algorithm,genetic algorithm and particle swarm optimization algorithm,and the superiority of the improved algorithm in accuracy,consistency and convergence is verified.(2)The improved firefly algorithm is applied to the finite element model updating based on modal features,and the deterministic updating method based on modal parameters and the Bayesian updating method based on modal flexibility are proposed respectively.The deterministic updating method based on modal parameters utilizes the least square principle to construct the objective function and uses the improved firefly algorithm to identify the parameters.Bayesian updating method based on modal flexibility takes advantage of Bayesian inference to construct the posterior probability density function of the model parameters,obtains the maximum a posteriori parameter by using the improved firefly algorithm,and calculate the the standard deviations of model parameters.Therefore,this method can not only avoid the difficulty of getting higer modes and the large noise effect on the accuracy of model updating,it can also evaluate the uncertainty of the results of model updating.(3)This thesis takes a 6-story steel shear frame structure test and several numerical simulation tests as examples and conducts model updating and structural damage identification by using deterministic updating method based on modal parameters and Bayesian updating method based on modal flexibility respectively.Both the methods obtained satisfactory results,but the latter has an irreplaceable advantage in uncertainty quantification.(4)A Kriging surrogate model based method is proposed for the large finite element model updating of actual bridges.Taking the Binhai Bay Bridge in Dongguan as an example,the Kriging surrogate model is established and the parameters of the bridge model are updated,which verifies the effectiveness of the proposed method in model updating of large and complex finite element models.