Model Updating And Damage Identification Based On Surrogate Model And Frequency Response Function

At present,most model updating methods take frequency as the response of the model.In some cases,the identification error of modal parameters is larger than modeling error of finite element model.Model updating method on the basis of frequency response function avoids structural modal analysis.The frequency response function can provide more data and each curve of it can be used as an objective function to update the model.At the same time,in the process of model updating,using surrogate model to replace finite element model can reduce computational cost,which is one of the effective ways to solve the problem of complex engineering structure optimization.Based on this,the model updating method based on radial basis function(RBF)model and frequency response function is studied firstly and the updating accuracy and consumption time is compared with those of the other surrogate models about.After that,the frequency response function matrix is calculated to obtain a new matrix,the element of which is taken as the damage response.The Kriging model is selected as the surrogate model to identify the damage.The main tasks are as follows:The theory of frequency response function and surrogate model is introduced.The advantages of frequency response function as output response are expounded.The specific theory of frequency response function is introduced.The relevant theories of excitation point and measuring point selection principle are introduced.Finally,the construction steps and theories of surrogate model are introduced.Combining with RBF model and frequency response function for the model updating theory.Aiming at a large number of computational cost problems caused by calling the original finite element model in the iterative process of model updating,RBF model is selected to replace the finite element model.The acceleration frequency response function is chosen as the output response.The excitation point and the measuring point are selected by the coefficient of variance of mode participation criterion and the modal kinetic energy method respectively.The sample set is constructed based on the uniform design table.The parameters of the RBF model is optimized to improve the prediction accuracy.Then the objective function is constructed with the minimum frequency response difference.The parameter updating value is solved by using the intelligent optimization algorithm—beetle antennae search algorithm.Finally,a 36-degree-of-freedom two-dimensional truss model is updated by this method,which is compared with the updated results of finite element method,response surface model method and Kriging model method.As a surrogate model,the updating accuracy of the RBF model has some advantages.Damage identification based on model updating method and new damage feature.The identification accuracy is low when the frequency response function is used for damageidentification.The frequency response function matrix is used for matrix operation and excitation points and measuring points are optimized.The corresponding matrix elements are taken as the damage response.In order to save the calculation cost,the Kriging model is selected for damage identification and its parameters are also optimized.At the same time,the cosine similarity with high power is introduced as the objective function.Cuckoo optimization algorithm is used to iteratively optimize the objective function.Finally,the damage identification of a cantilever beam model is carried out and compared with the original frequency response function as the response.It is found that the identification accuracy of the new damage response is better than that of the frequency response function response.