Study On The Intelligent Recognition Technology Of Surface Crack Of Ballastless Track Plate Of High Speed Rail Way

The ballastless track of high-speed railway generally appear different problems of surface crack on different levels,Automatic detection technology for the surface crack of ballastless track has become one of the key technologies for high-speed railway maintenance management.The existence of the crack has influence on the service life and the state of the track plate,so it is vital important to detect the surface quality of the slab.The existing detection theory and methods are more based on the processing and identification of cracks in 2D images,which cannot meet the requirements of high precision,high speed,high accuracy and full automation.Ultra-high speed cloud point sensing technology based on 3D laser,through the development of system testing platform,achieved the rapid collection of three-dimensional data of ballastless track slab surface cracks with high precision,synchronous and real-time storage.Based on the foundation of high precision 3D data acquisition and accurate positioning,developed the data acquisition and analysis platform of high-speed railway ballastless track surface cracks,realized the visualization of all the data which obtained by the system,including the query and the management.Combined with actual situation of high-speed railway ballastless track in our country,proposed typical surface crack automatic detection method and automatic recognition algorithm which are based on 3D data of track slab,realized crack automatic recognition based on three-dimensional image.And we have realized the application of crack automatic recognition in our developed system software,verified the reliability of the detection system by indoor model test.Our detection system can achieved the data collection and intelligent identification of track plate crack,having achieved a high precision intelligent crack identification.The comparison results show that the system can obtain the high precision of the fracture data.Test data show that the relative error of the length,width,and depth of the crack is about 10%.