Damage Identification Based On Three Dimensional Digital Image Correlation And Bayesian Operational Analysis

Damage identification has gained increasing attention from the scientific and engineering communities because the unpredicted structural failure may affect the state,efficiency and safety of the structures and may further cause catastrophic,economic,and human life loss.When a structure suffers from damages,one or more of its dynamic properties will change.Conversely,the damage could be identified using the changes in the dynamic response characteristics.More and more vibration-based damage identification methods have been proposed because they are reliable and effective nondestructive techniques.However,due to the low spatial resolution of the traditional modal testing(no dense enough measurement points),and extra mass added by contact sensors,the accurate mode shapes cannot be obtained.The local damage of structure cannot be identified without enough accurate mode shapes.In this paper,the threedimensional digital image correlation(3D-DIC)and Bayesian Operational Modal Analysis(BOMA)are successfully integrated to identify the local structural damage.Mode shapes measured by this method would be more accurate than those from the traditional pointwise measurement method.The 3D-DIC method uses 3D stereo-vision calibration and DIC method to reconstruct the three-dimensional coordinates of the structure.So it is able to measure responses in the form of displacements.Compared with the traditional optical measurement method,this method has the advantages of simple optical arrangements,white light source,and no need for vibration isolation.And compared with the traditional modal testing method of directly attaching sensors onto the structure,it also has many advantages as it is a noncontact and full-field measurement.BOMA adopts a Bayesian system identification approach for operational modal analysis(OMA)using Fast Fourier Transform(FFT)as data processing means.This method aims at identifying the modal properties(natural frequencies,damping ratios,mode shapes,etc.)of a constructed structure using only its output vibration response measured under operating or ambient conditions while the input excitations to the structure are not measured.By maximizing the probability density function,not only the optimal values of modal properties but also the uncertainties and the signal-to-noise ratio are obtained,the latter of which can evaluate the analysis results.In order to improve the computation speed,the modal identification is well separated with single mode.The singular spectrum analysis method is then used to optimize the power spectral density map.The single mode peak value and its range are accurately determined through the singular value spectrum.This paper combines these two methods 3D-DIC and BOMA,making full use of the advantages of these two methods.Undamaged modal properties of structure are usually unavailable in practice.Assuming that the undamaged structure is geometrically smooth and made of linear isotropic and homogeneous materials that have no stiffness and mass discontinuities,a mode shape of an undamaged structure is obtained from a polynomial of a properly determined order that fits the corresponding mode shape of the damaged one.Mode shape damage index(MSDI),mean curvature damage index(Mean-CDI)and Gaussian curvature damage index(Gaussian-CDI)are proposed correspondingly to assist the damage identification.The finite element method is used to simulate a circular membrane structure to validate the proposed method.In the experiment part,the damaged circular and rectangular membranes with different boundaries are tested too.And the damages which are razor cuts are also successfully identified through the three damage indices.The damage identification using modal strain energy also successfully detect the damage of the rectangular membrane.But the result shows that due to multiple numerical solution,the error is also larger than the other three damage indices.