Research On Measurement Method For Frequency Response Function By DIC

The frequency response function(FRF)is a powerful tool for dynamic characteristic evaluation,vibration control,and fault diagnosis of mechanical systems.Digital image correlation(DIC),which is a non-contact and full-field computer vision measurement method,has become a reliable means for structural dynamic testing,and is especially suitable for FRF measurement of plate and shell structures.However,the DIC based vibration testing suffers from small data sample,high level floor noise,and low accuracy in high-frequency band.To address these problems,this thesis develops the FRF measurement method based on the full-field DIC displacement data,using Bayesian inference and substructure modeling methods.The main research work of this thesis is listed as follows:(1)For the problem of small data sample in DIC testing,a Bayesian inference method for FRF is established.The Bayesian method is introduced into FRF testing and estimation,and the FRF is modeled using a complex Gaussian process to establish a Bayesian estimation method for FRF in the frequency domain.The regularized estimation of FRF is obtained under the condition of small sample data,and resulting in the smallest mean square error of FRF estimation.(2)For the problem of high floor noise in DIC testing,a Bayesian estimation method for FRF based on the frequency-space model is proposed.By exploiting the spatial smoothness of structural modal shapes,a low-order polynomial model of FRF is established in the local space.Combined with the complex Gaussian process model of FRF,the Bayesian estimation accuracy of FRF is improved by using the rich spatial measurement information of DIC.In addition,for the problem of high matrix dimension brought by the joint modeling,the QR decomposition of matrix is applied to improve the numerical stability of parameter optimization and FRF calculation,which improves the computational accuracy and efficiency.(3)For the problem of low accuracy in the high-frequency band,a hybrid measurement method for FRF is developed.Based on the test data from the accelerometer,the modal parameters(natural frequency and damping ratio)of the structure are identified,and the FRF measured by the DIC is reconstructed.Further,the substructure modeling method is used to fuse the test data of the accelerometer-based on the equivalent model,which improves the reconstruction accuracy of FRF.(4)Numerical and experimental validations of the DIC measurement method for FRF are carried out.Based on the simulated impact response of simply supported steel plate,the effectiveness of the Bayesian FRF estimation is verified.Using the measured response of the propeller blade and light-weight aluminum plate under hammering excitation,the applicability of the proposed method is justified.Through the research work carried out in this thesis,the FRF estimation method for DIC testing is developed,which is valuable for the high-precision testing and modeling for FRF of plate and shell mechanical structures.