Research On Structural Model Updating Method Based On Wavelet Energy Of Frequency Response Function

The establishment of an accurate model plays a decisive role in the results of finite element analysis,but in the process of modeling,dimensional errors,and material parameter deviations are unavoidable.So it is necessary to make further updating to the established more accurate model,so that the error between the response obtained by the model and the response of the actual structure can be within an acceptable range.This can provide a reliable model for the mechanical performance analysis.Therefore,the model updating methods are investigated in this thesis.The method of model updating based on the frequency response functions wavelet energy and the principle of choosing the nearest is proposed,and then the stochastic model updating method based on lifting wavelet total energy is studied.The contents are as follows:The model updating methods that have emerged so far are summarized.The model updating method using the frequency response functions and modal parameters is introduced and compared.The process of model updating using surrogate model is described.The advantages of choosing wavelet energy instead of frequency response function as the response are explained.The theoretical knowledge of wavelet transform is introduced.The model of a two-dimensional truss is updated based on wavelet energy and the principle of choosing the nearest.Firstly,the modal participation criterion is used to get the best excitation point.The frequency response function is calculated according to the formula.The frequency response function is inversely Fourier transformed into a time domain signal,and then the energy of the first three layers of high frequency and the last layer of low frequency is extracted to form a feature vector of the frequency response function.Finally,the shortest Euclidean distance between the vectors is calculated according to the principle of choosing the nearest to complete the pattern recognition,and the updated model is obtained.The updating result is good,and the updated frequency response function curve basically coincides with the real curve.On the basis of the randomness problem in model updating,a stochastic model updating method based on lifting wavelet total energy and Monte Carlo sampling is proposed.Firstly,a more advantageous lifting wavelet transform is applied to decompose the frequency response function.The total energy of lifting wavelet transform as the output of the surrogate model is extracted.The elastic modulus,density,and cross-sectional area are chosen as the parameters(the inputs of surrogate model)to be updated,construct a response surface surrogate model to replace the original finite element model.The Monte Carlo sampling is used to obtain a large number of response samples,and the samples are screened.The smallest difference between the response predicted by the surrogate model and the real response obtained by sampling is taken as the objective function.Finally,the cuckoo optimization algorithm adopted to solve the problem,and the mean value of the updating parameters is obtained.A two-dimensional and a three-dimensional truss models used to verify the updating method.Furthermore,the proposed method is also verified by simply supported beam test.Both the simulation and test results show that the parameters error are reduced after updating,and the obtained frequency response function curve is in high degree of agreement with the real curve,and also show that the method has a certain practicability in model updating.