Research On Key Technology Of Machine-visional Diagnosis System For Rail Profile Grinding

The rail profile is an important feature of rail, it directly bears the pressure of thetrain, and, it ensures the safety of train running. Because of the train pressure, thecomplex and changeable natural environment and reason for the long-term use of rail,the profile of the rail world produce different levels of damage, such as wear,deformation and film, which would gradually reduce the height of rail, graduallyincrease the fat edge, and make the top surface uneven. It influences the safety oftrain fast running. So, railway maintenance such as rail grinding is the security of railcondition. In recent years, railway infrastructure is more and more perfect, so that,rail grinding become more and more important. But, process of rail grinding atpresent is complex as before, and, a majority of equipment for rail profile detection isof contact type. Therefore, the operations of rail grinding are low. For the purpose toimprove the precision and the efficiency of rail grinding, this paper developed annon-contact type diagnosis system for rail profile grinding with detection method ofmachine vision. In the measurement part of this system, it should be first extract thefringe center stripe in the image, It directly affects the accuracy of measurement, sothe paper realizes the algorithm for accurately extract the center line of stable. Itcontains handles such as Background difference, Smoothing, Threshold Division,Center line extraction, delete rags and hollows. Then, this paper use cameracalibration method of the direct linear transformation (DLT) model to find therelationship between the image coordinates and object coordinate system, so that, thefringe center line in the image could be used to get actually rail profile. Then, thispaper studies the smooth processing algorithm for rail profile data for eliminating theimage pixel grid and the curve of profile influence on measurement accuracy. Thepaper also studies Alignment of real rail profile and the reference profile. Thismethod of automatic alignment ensures the stability of the grinding mode tableproduced by system. At the same time, Automatic alignment results are more accuratethan artificial alignment. At last, the paper research and design of the test scheme ofmeasurement system, verified the system measuring accuracy and repeatabilitythrough a large number of experimental data, verified the feasibility and accuracy of system to guide the grinding operation in the railway field trial.The measure precision of the system has reached0.1mm. And Comparing withthe similar products, the system of this paper add the function of automatic analysis.It can accomplish measurement and analysis completely by automatic. Also, thesystem of the paper contains the module of guide for rail grinding, which cangenerate grinding mode table for the grinding train. Then, the work of people to writepattern could be not necessary, and, the grinding precision could be improved.Obviously, it improved the grinding efficiency and the grinding accuracy.