Research On Bridge Load Identification And Equivalent Estimation Method Based On Traffic Flow Monitoring Image

It is of great significance to carry out economical,practical,convenient and rapid load identification of vehicles passing the bridge for the health monitoring of bridge structure.This paper is supported by the National Natural Science Foundation project "Research on Bridge Safety State Monitoring Method and Early Warning Theory Based on Bridge Deck Shape Change(51708068)" and "Exploration of Bridge Holographic Deformation and Structural State Deduction Method Based on Image Contour Stacking Analysis(51778094)" undertaken by my supervisor.Using traffic surveillance cameras installed on Bridges across the country,the research on bridge load identification and equivalent estimation method based on traffic flow surveillance images is carried out.The main contents are as follows:(1)Study on the size of vehicles crossing the bridge based on monocular vision recognition.Aiming at the bridge traffic flow image under the overhead Angle,the theoretical relationship between the five coordinate systems in the computer image projection transformation model is derived,and the theoretical mapping equation between the three-dimensional world coordinate and the computer image coordinate is established.A monocular vision dimension measurement model based on the camera imaging geometry principle is constructed,and the scale conversion between the vertical and horizontal unit pixels in the computer image and the actual distance in the three-dimensional world is realized.The size recognition method of vehicles crossing the bridge under the overhead camera is proposed.Based on the control variable method,the effects of camera height,vertical Angle and physical focal length on pixel scale factor(PSF)were investigated.The results show that the vertical Angle of camera has the greatest influence on the measurement accuracy,and the camera installation process should choose a smaller physical focal length to obtain the maximum range of view.(2)The equivalent model between vehicle size and load is constructed.By collecting domestic and foreign vehicle classification standards and codes,the probability distribution of existing traffic load survey and monitoring data was statistically analyzed.A load axis parameter atlas was established to represent the limits of the codes,and an equivalent model between vehicle size and load was constructed.In order to solve the problem of identification accuracy of large-sized heavy-duty trucks,a load identification theory based on the correction of maximum deflection and influence line was proposed,and a load correction method based on the response of bridge structure was established.(3)The experimental study of vehicle load identification on the bridge based on image data is carried out.The monitoring image data of the model vehicle and the elevation image data of the test bridge at different measuring points under different working conditions were obtained by using a camera with a fixed overhead Angle,and the measurement of vehicle length and deflection of the test bridge under different working conditions was realized.The error sources,such as the position of the camera’s simplified optical center and pixel coordinates,were analyzed.It was found that the position of the theoretical optical center of the camera could be determined with the structural information of the bridge as the iterative constraint,and the size recognition accuracy could be improved by ignoring the target vehicles at the upper and lower edges of the image.The load-displacement parameter calibration of the test bridge was completed according to several standard loads,and the load inversion of the model car was realized in combination with the deflection of the test bridge.The results show that the average identification error rate is about 10% under the three working conditions,which verifies the feasibility of vehicle load identification on the bridge based on monitoring image.