| In order to ensure the safety and reliability of the key components of the railway vehicle structure,the designer needs to accurately obtain the type and size of the external load in the engineering design.The identification of external loads is of great significance in vehicle strength design,structural health monitoring and fault diagnosis.In many cases,it is impossible to obtain the external load directly by installing sensors in practical projects,people need to take effective load identification methods to reconstruct the abnormal load.Especially when the vibration frequency of the external load covers the natural dominant frequency of the structure,the structural components may have major accidents such as resonance fatigue and fracture failure.In this paper,based on the load type,the load identification technology of periodic dynamic load,impact load and moving load is optimized(1)For periodic dynamic load identification,green kernel function is used to establish transfer function through impulse response.Tikhonov regularization method,truncated singular value(TSVD)regularization method and least square QR decomposition(LSQR)regularization method are used to reconstruct the load identification,and the effects of different load types and different levels of noise on the identification results are discussed.The experiment is used to verify the periodic dynamic load.The modal frequency of the structure is obtained by modal test.The finite element model is modified by the experimental modal modification method to keep the simulation model consistent with the experimental model.The necessity and effectiveness of the model modification are verified by the identification results before and after the modification.(2)For the identification of impact load,the convolution integral relationship between load and vibration response is used to establish the identification equation of impact load.The impact sparse deconvolution model is constructed by L1 norm regularization method.The search direction is determined by preconditioned conjugate gradient method and solved by the truncated Newton interior point method.Using the impact load experiment to verify the recognition effect of the method,the location of the measuring point and the selection of the regularization parameters are discussed,and verified on the high-speed train pantograph.(3)For the identification of moving load,the transfer matrix between the modal vibration response and the modal force is established based on green kernel function.The modal force is identified by the regularization method,and then the moving load is reconstructed by the modal coordinate transformation.The influence of the position and response type of the measuring point on the recognition accuracy is discussed.The influence of different vibration response and the position recognition effect of the measuring point is studied,and the noise resistance of the method is discussed.(4)An optimization method based on vibration response identification algorithm is proposed.This method includes the optimization of recognition algorithm,sensor layout and different model modification methods.Finally,several typical practical examples(cantilever beam plate structure,wheel structure and pantograph structure)are used to simulate and verify the algorithm of structural load identification based on vibration response.The research results of cantilever beam and wheel structure show that,for periodic dynamic load,the method of vibration acceleration signal combined with TSVD regularization has better recognition result,higher noise resistance and stability,and the correlation coefficient can be kept above 0.95 under 20% noise;the finite element model is modified by experimental mode correction method,and the result of experimental load identification is robust and stable Qualitative analysis improved significantly.The results show that the correlation coefficient is above 0.9 and the peak error is less than 30% when L1 norm regularization is used to identify the impact load.In the example of single span beam moving load identification,it is shown that the result is robust when the moving load is reconstructed by using vibration response and modal force method. |