Research On Bridge Damage Identification Of Displacement Impact Surface Based On Computer Vision

Bridges are an important part of urban transportation,connecting different areas of the city and providing ways to cross obstacles.It is unavoidable to cause fatigue damage or even damage due to design reasons,service life,so long-term structural health monitoring of bridges is required.Conventional bridge monitoring methods have the shortcomings of high economy,poor bridge versatility and difficult sensor maintenance,so it is urgent to find an efficient,intuitive and versatile bridge structure health monitoring method.This thesis relies on the project of National Youth Fund of Natural Science（51708068）,which is sponsored by the tutor,to carry out the research on damage identification of bridges with displacement influence surface using computer vision.Dynamic displacement response locations are obtained through the displacement influence surface of the whole field of the bridge under moving loads.Damage locating indexes based on the displacement influence surface and quantitative damage formulas are constructed to identify local damage of simply supported beams by computer vision.The main work of this thesis is as follows:Firstly,by combining the general mid-span displacement influence line theorem and the deflection influence line theorem,the formulas of the non-destructive displacement influence surface and the damaged displacement influence surface of the bridge under moving load are derived.Based on the definition of the longitudinal displacement response position Y of the displacement influence surface,a method of extracting the full displacement influence surface of a simply supported beam under dense measuring points by computer vision is proposed.A simple beam model with single box and two chambers is made indoors,and the moving load position-displacement curves of dense measuring points are extracted by camera calibration,Canny edge detection and program compiled by MATLAB software,and the moving load position-displacement curves of each measuring point are connected to displacement influence surface.Secondly,using the expressions of the lossless displacement influence surface and the lossy displacement influence surface,the damage location index based on the displacement influence surface is derived:the displacement influence surface difference damage identification index₁（,）,displacement influence surface first order difference damage identification index₂（,）,displacement influence surface second-order difference damage identification index₃（,）.and the sensitivity of the three indicators in damage location,damage degree,load size,and damage number was analyzed using the control variable method using a single box,three compartment,and simply supported beam model established by finite element method.The sensitivity of the three indicators in damage identification was compared.Then,based on the second-order difference damage identification index₃（,）of the displacement influence surface,a damage quantification formula within a certain degree of damage is established,the influence of different influencing factors on the quantification of damage degree was verified within the finite element method.The effectiveness of the damage quantification formula was verified by comparing the error between the theoretical damage obtained from the damage quantification formula and the actual damage.The differences between displacement influence surface identification by sensors and that by computer vision under dense measuring points are compared.The results show that only when the number of measuring points reaches a certain density,a better damage identification effect can be achieved.Finally,the effectiveness of obtaining the displacement influence surface under dense measurement points through computer vision was verified on a single box,three compartment,simply supported beam model in the structural laboratory.The feasibility of the damage location index and damage quantification formula proposed in this theise is verified through multiple load cases and multiple damage cases tests.At the second span of the approach bridge of the Douzibei Bridge on Caiyuan Road in Caiyuanba,Chongqing,which is about to be reinforced,the displacement influence surface extraction under dense measuring points before reinforcement was carried out,preliminarily verifying the possibility of using computer vision to extract the displacement influence surface on the real bridge.The method of bridge damage identification with displacement influence surface based on computer vision can locate damage efficiently and intuitively,and provide a new technical path for structural health monitoring of medium and small span bridges.