Study Of Structural Dynamic Fingerprint And Damage Identification Based On Video Image

Structural health status can be reflected by its vibration characteristics,and the traditional sensors can only obtain the single point dynamic response from the structure surface.There is a new method to obtain the dynamic response of structures with the rapidly development of digital photography and image processing technology.The research is supported by the National Natural Science Foundation of China(NSFC)project "Research on holographic deformation and structural state deduction method of bridge obtained by image contour superposition analysis(51778094)".In this dissertation,cantilever beam and suspension model bridge vibration test are carried out.Modal identification and damage judgment of structure are realized based on machine vision meassurent.The results are verified by numerical simulation and traditional measurement.The main work is as follows:Firstly,the dynamic displacement of the structure is measured by machine vision.The template matching program written by the MATLAB is used for track the motion of the local area of the structure,and the Lucas-Kanade optical flow method by Python is for detecting the motion of a feature point of the structure.The reliability and accuracy of the visual measurement method is verified from a simple example.Compared with the vibration displacement measured by laser,the maximum error is7.67%,which shows the good accuracy of image measurement.Next,the dynamic fingerprint and modal parameters of the cantilever beam are obtained by machine vision.The template matching method is used to track the vibration displacement of each measuring points along the beam.The natural frequencies of each measuring point are obtained by fast Fourier transform.The first three vibration frequencies,damping ratio and modal parameters of the structure are calculated by COV-SSI(subspace modal identification method based on covariance).The results show that the dynamic characteristics of each point measured by machine vision are consistent with those by numerical simulation,the MAC values of the first three modes with the numerical simulation results are almost equal to 1.And the maximum difference between frequency and acceleration sensor is 0.13%.And then,the suspension model bridge teste was carried out,the displacement tracking results of different resolution image sequences indicates that the image definition affect the accuracy of measurement results.The error of visual measurement is explained from the image sampling quality.Finally,damage of cantilever beam was identificated based on machine vision.Several different degrees of damage of the cantilever beam under the two models were simulated by finite element method,and the curvature mode is calculated by the central difference method.The results show that the smaller division element is beneficial to damage identification,and the curvature mode difference is highly sensitive to the structural damage.The damage test of cantilever beam is carried out,and the dense measuring points are selected for visual measurement according to the numerical simulation experience.The frequency and curvature modal difference before and after the damage are compared.The test results illustrate that the structural damage can be identificated and located preliminarily by visual measurement.