Design And Research Of Bridge Intelligent Detection System Based On Machine Vision

In recent years,China’s high-speed railway construction has made significant progress.However,due to the frequent occurrence of safety accidents,problems such as damage to bridge structures and casualties have become increasingly prominent,seriously affecting the safe operation of high-speed trains.China’s bridges have passed the period of large-scale construction and entered a critical node where construction and maintenance are equally important.Bridge safety and bridge inspection issues are increasingly being valued.Traditional bridge inspection methods have problems such as low efficiency and low automation.The widely used bridge inspection vehicles also need to carry workers to inspect the condition of the bridge bottom,which consumes a lot of manpower.With the rise of machine vision,this article uses Solidworks,ANSYS,MATLAB and other software as platforms to design a bridge crack detection system based on machine vision.The main body of the detection system is an unmanned video bridge detection vehicle,which is equipped with high-definition zoom cameras to detect the condition of the bridge bottom.The thesis first introduces the current research status of bridge disease image recognition technology for bridge inspection vehicles at home and abroad.Based on this,the overall mechanical structure design of unmanned video bridge inspection vehicles is formulated,including the body part,rocker arm part,and walking device.Power calculation and motor selection are carried out for the walking device.At the same time,a new type of walking track system is introduced,which greatly saves the use of track steel materials;Perform static and modal analysis on the walking mechanism using ANSYS to verify the reliability of its strength;Take pictures of bridge cracks based on the planned walking route at the bottom of the bridge.Then the improved median filter is used to reduce the image noise,and the improved Canny operator is used to detect the edge of the image.According to the morphological open operation principle,the close before open operation is used to further remove the noise,and a clearer binary image is obtained.Then,the binary image is projected to the X axis,45 ° axis,Y axis,and 135 ° axis respectively,to statistics the characteristics of the cracks,and achieve the classification of crack types;Construct a BP neural network to classify crack images.Based on the comparison between the classification results and the actual situation,the accuracy of crack classification recognition is over 90%.The research results indicate that the bridge crack detection system based on unmanned video bridge detection vehicles has the advantages of high automation,fast detection speed,and high accuracy,providing new ideas for the development of subsequent bridge detection.