Research On Model Updating Method Based On Features Of Frequency Response Function

Using finite element software to simulate the actual working conditions of the actual structure and analyzing the structural strength,vibration characteristics and structural dynamic characteristics of the structural system are of great significance to the structural design.However,due to many uncertain factors,there exists some errors between the established finite element model and the actual structure in the process of structural modeling.Moreover,the measured data in the actual environment will certainly be disturbed by noise.Therefore,in order to reduce the error between the finite element model and the actual structure,improve the ability of the finite element model to simulate the actual structure,and make the model updating method robust to noise.Based on the theory of frequency response function,the model updating method based on the features of the frequency response function is studied in this thesis.The main research contents are as follows:Firstly,a model updating method based on singular values of frequency response function is proposed.Aiming at the problem that the correlation coefficient in Kriging model is difficult to determine,the optimization algorithm is used to optimize the correlation coefficient,so as to improve the construction accuracy of Kriging model.The number of effective singular values is determined according to the mutation principle of the number of extreme points,so that the proposed model updating method has certain robustness under different SNR.The numerical example shows that when the signal is disturbed by weak noise,even if the effective singular values are not selected,it has certain anti-noise performance.When the signal is disturbed by strong noise,the effective singular values are selected according to the mutation principle of the number of extreme value points,and the updating error is less than 3%,which proves that the proposed method has strong anti-noise performance.Secondly,a model updating method based on the wavelet decomposition of acceleration frequency response function is proposed.The parameters of the structure to be updated are determined by the sensitivity analysis method.After the acceleration frequency response function measured easily in engineering practice is decomposed by wavelet,the wavelet coefficient with larger amplitude is used to represent the frequency response function for model updating.Numerical examples show that the proposed method can still obtain satisfactory updating effects under different noise levels.Even when the SNR is as low as 5d B,the updating errors of each parameter are still less than 4%,which proves the robustness of the proposed method to strong noise.Finally,based on the above model updating methods,the finite element model of highspeed train wheel is updated.High-speed train wheel model is established according to its parameters,and Kriging model is used to replace wheel finite element model for iterative calculation,which solves the time-consuming and laborious problem of wheel finite element model analysis and improves the model updating efficiency.Due to the measured data of wheels are susceptible to noise interference in actual operating conditions,the proposed two methods are applied to update wheel finite element model.It reduces the gap between the finite element model and the actual structure and improves the finite element simulation ability of wheel by using the presented two kinds of model updating method which provides a foundation for the analysis of wheel vibration characteristics by using wheel finite element.