Machine Vision-based Bridge Health Monitoring And Condition Assessment

Bridge structural health monitoring(SHM)aims to measure the environmental and operational parameters,observe the static and dynamic structural features,execute the structural condition and safety evaluation,and give the instructive inspection and maintenance recommendations.In a bridge SHM system,different kinds of sensors need to be applied to acquire a variety of physical parameters of the instrumented structure,and the novel data-driven structural condition assessment approaches are applied to carry out the diagnostic and prognostic tasks.The limitations and drawbacks existent in the traditional sensing techniques are poor durability,low accuracy of measurement,low sample rate,inconvenient installation,and weak performance of anti-electromagnetic interference.It urgently needs novel sensing technology to solve these problems in the field of bridge health monitoring.In this thesis,a vision-based bridge health monitoring and condition assessment framework is established combining the machine vision technology with the traditional civil engineering monitoring methods.Its performance is evaluated.The main work is shown as follows:(1)The basic components and research emphases of bridge health monitoring are stated.The hot research issues and key monitoring techniques in SHM at present are investigated.The research progress of machine vision-based sensing techniques and methods applying in the field of civil SHM is reviewed.The research approach of this thesis is given.(2)Three kinds of image processing algorithms for structural dynamic displacement measurement are presented,i.e.,grayscale pattern matching(GPM)algorithm,color pattern matching(CPM)algorithm,and mean shift tracking(MST)algorithm,respectively.The accuracy,reliability and stability of the displacement measurement using these three methods are verified through a comparative study of frame vibration experiments with the magnetostrictive displacement sensor(MDS).The performance of these three methods is evaluated and the application suggestions are given in real bridge application accordingly.The feasibility and accuracy of the GPM-based muti-point structural dynamic displacement monitoring method are further validated by the moving loading experiments on a real bridge.(3)The vibration monitoring and dynamic characteristics identification are carried out on a simple supported beam simulating the bridge deck and a vertical bar simulating the bridge tower,respectively.A comparative study is conducted with the traditional accelerometer method.The feasibility and accuracy of the proposed method for vibration monitoring and dynamic characteristics identification are verified.(4)The vision-based steel cable force monitoring method is proposed by integrating the vision-based multi-point structural dynamic displacement monitoring method with the basic theory of material mechanics.The tensile tests and cable force monitoring of three different types of cable samples(steel bar,steel rope and parallel strand cable)are carried out on the universal testing machine(UTM)in the laboratory.The feasibility and accuracy are verified in comparison with the force sensor on the UTM.The measurement performance is further validated by measuring the cable forces on a scale arch bridge model under moving loadings.(5)Different kinds of environmental influencing factors confronted in site monitoing applications of the presented method are descriped.Environmental factors are simulated in the laboratory including the light illumination,elevation angle,vapor,wind and ground vibration.The performance of the vision-based system is evaluated under different conditions.Relevent solutions are proposed according the problems of different influencing factors.The feasibility of the proposed solutions is verified by laboratory tests.