Research On Structural Modal Monitoring Method Based On UAV Video Technology In Real Environment

The identification of modal parameters of structural systems is an important part of structural health monitoring,which is mainly obtained through analysis of structural vibration response data such as acceleration and displacement.These data are usually obtained by sensors placed on the structure,which is cumbersome and time-consuming to install,and the monitoring is also inconvenient.In recent years,with the development of image data acquisition and computer vision technology,video monitoring has gradually become an efficient method for structural response monitoring.However,the current structured video monitoring technology generally requires the use of high-speed cameras to be performed in an ideal environment,which has poor adaptability to the real environment.In response to the above problems,this paper studies a comprehensive solution and proposes a computer vision analysis that can use consumer-grade cameras and drones to shoot and adapt to a variety of environmental disturbances,including complex backgrounds,lighting changes,fog occlusion,camera shake,wind interference,etc.The new method realizes high-precision structural modal parameter recognition.The main work and achievement of the thesis are as follows.(1)A technical method for the application of consumer-grade cameras to simultaneously monitor the dynamic displacement of multiple targets is proposed.A smartphone collects the structure vibration video,and multiple monitoring target points are automatically assembled from predefined overlapping areas.Apply Spatio-temporal context algorithm and characteristic optical flow algorithm to synchronously track and obtain high-precision displacement time history information of all monitored target points.Through the frequency sweep experiment of the small ball model and the simulation of the illumination change,the displacement monitoring effect of this method and the commonly used characteristic optical flow algorithm is compared and tested,and the feasibility of the method is verified.(2)A motion-enhanced Spatio-temporal context algorithm is proposed to improve the robustness of displacement monitoring.Based on the conventional spatio-temporal context method,this paper proposes deep mining of the information in the context area,applying dense optical flow algorithm to calculate the similarity of motion trends and constructing the context influence matrix,which enhances the robustness of the spatio-temporal context algorithm in tracking targets.Through the small ball model sweeping experiment and simulating the illumination change,fog interference,and camera shake,the displacement monitoring effect of this method and the commonly used characteristic optical flow algorithm and the conventional context algorithm are compared and verified,and the method has a stronger anti-interference ability.(3)The technical method of using UAV video for structural modal identification in the real environment is proposed.Aiming at the problem of camera dynamic displacement caused by UAV movement,an adaptive scaling factor update method based on Spatio-temporal edge detection is proposed,and combined with the motion-enhanced Spatio-temporal context algorithm,the accurate tracking of the multi-point dynamic displacement of the structure is realized.Then apply the data-driven random subspace method to complete the structural modal parameter identification.Through the small ball model sweeping experiment,and simulating the illumination change,fog interference and camera shake,the displacement monitoring effect of this method and the commonly used characteristic optical flow algorithm and the conventional context algorithm are compared and verified,and the method has a stronger antiinterference ability.