Bridge Bottom Crack Detection System Based On Flying Climbing Robot Platform

In recent years,with the rapid development of infrastructure construction,the bridge construction in China has been developing to the direction of large span,heavy load and new materials.Among them,the construction of high-speed railway bridge,large-span highway bridge and sea bridge are constantly refreshing the world record.With the gradual breakthrough of bridge design and construction technical difficulties,the structural health detection and state evaluation of bridges have also been paid more and more attention.In recent years,with the increase of bridge service time and load pressure,bridge collapse accidents occur frequently,which has caused huge losses to the economy and people’s life safety of our country.At present,China’s bridge development planning has entered the stage of both construction and maintenance from large-scale construction.Aiming at the problems of low automation,low detection efficiency,high detection cost and lack of guarantee for life safety of detection personnel in traditional bridge detection,this paper proposes a bridge bottom crack detection system based on tethered flying climbing robot.After analyzing the characteristics of crack image and the engineering requirements of crack detection at the bottom of bridge,this paper studies the key technologies such as the mechanism design of flying climbing robot,the splicing of panoramic image of crack,the automatic recognition of crack target,the automatic recognition and measurement of crack at the bottom of bridge,and designs a visual detection device that can automatically detect the crack at the bottom of bridge,and analyzes its advantages through experiments Effectiveness and measurement error.The introduction part of the paper introduces the existing bridge crack detection technology at home and abroad.On this basis,a detection scheme for identifying the bottom of the bridge through the tethered flying robot platform is obtained,and combined with deep learning and image processing technology to identify cracks.The main work content of the main body includes:1.According to the dynamic characteristics of unmanned aerial vehicles and tethered cables,a flying crawling robot platform suitable for bridge crack detection is designed from the three aspects of power system,crawling system and image acquisition system.2.For the obtained bottom image of bridge,a feature image extraction method is proposed by using uniform ORB operator,and the crack image stitching algorithm is optimized by random sampling consistency and LM optimized affine transformation matrix to achieve the bottom image of bridge.splice.3.A two-stage deep learning model is proposed for the stitched panoramic panoramic image,which can accomplish the segmentation of the crack image and determine whether the image contains a crack target.4.The image containing the crack target obtained based on deep learning,after MSRCR enhancement and bilateral filtering processing,an adaptive threshold segmentation method and crack parameter calculation method for weighting grayscale images through deep learning results are proposed.The results of this method and manual measurement are compared to verify its effectiveness and accuracy through experiments.