The Study Of Structural Vibration Detection System Based On Computer Vision

Vibration detection is essential in the fault diagnosis and health monitoring of bridge girder erection machines,bridges,and other structures.Theoretically,structure defects result in a variation of flexibility and bearing capacity.In recent years,noncontact methods represented by computer vision have become a research hotspot in structural vibration detection.Compared to traditional methods,the machine-visionbased vibration detection method has the advantages of simple equipment,nondestruction of measured structures,wide measurement range.And it doesn’t need to change the vibration characteristics of the measured target.This thesis focuses on the phase-based motion estimation and motion magnification algorithms,develops a vision-based vibration detection system.And the system is validated through a series of experiments.The contents of this thesis will be introduced by the following four aspects:(1)The research status of mechanical fault diagnosis and the application of computer vision in vibration detection is reviewed.(2)The image preprocessing is studied.And the image enhancement technology is introduced,which including gray-scale image enhancement,histogram equalization,image filtering and RGB video noise reduction preprocessing.(3)The imaging principle and detection processes of vision-based vibration detection is studied.For structural micro-vibrations in practical engineering,the featureless motion tracking technology and motion magnification algorithm which is based on the optical flow method is studied.This thesis compares three methods based on Euler’s perspective in the motion magnification algorithm in a comparison experiment.And finally,it verifies that the phase-based motion estimation and magnification algorithm can detect small vibrations and is good in noise suppression.Hence the method is chosen as the core algorithm of the vibration detection system.(4)A vibration detection system is developed,which incorporated phase-based motion estimation and magnification algorithms,in LabVIEW combined with MATLAB programs.The system is composed of four modules: the image sampling and file reading,image enhancement,vibration analysis,and result display.The thesis carried out vibration experiments for simply supported beams and simulated box beams respectively and compared the detection results from the system with those from traditional sensors.It proves that the proposed vibration detection system meets the design requirements.The research of the thesis is of general application for structural vibration detection.