Research On Dynamic Detecting Method Of Wheelset Lateral Displacement Based On Image Process

With the rapid development of China’s rail transit field and the sharp increase in railway operating mileage year by year,people’s travel convenience has been significantly improved,but the problems caused by management and maintenance are also increasing,bringing huge waste of manpower and resources to the country.The rise of digital technology has injected new vitality into all walks of life,and studies have shown that digital and automated management can greatly liberate labor and resources while railway automation management still has relatively broad room for improvement.The displacement of the wheelset is an important factor and evaluation index that affects the safety of train operation and passing capacity.It is necessary to detect the displacement of the train wheelset to ensure the safety of train operation.At present,there is no recognized method for the detection of the wheelset displacement,because it is difficult to detect through traditional measurement methods.In order to solve this problem,this paper develops a wheelset displacement measurement system based on image detection,which uses image processing technology and camera ranging technology as the framework to convert the detection of wheelset displacement into the trace of the relative position of the rail in the wheel-rail image,and completes the calculation of the wheelset displacement.The main research contents are as follows:For the wheel-rail image with uneven lighting,the adaptive brightness equalization algorithm is first used to fill or suppress the key edges in the image,and then on the basis of pre-processing operations such as grayscale,filtering and noise reduction of the image,the detection and extraction of the rail edge line and the wheel contour line are realized.This paper proposes a rail edge line fitting algorithm based on segmented model to detect the edge of the rail,extracts the feature points of the rail edge line and the wheel edge based on the detection results of the cannon edge detection and the Hough transform detection results,and further uses the segmented model to complete the fitting of the double edge line of the rail;for the wheel-rail image of insufficient illumination,a rail line detection method based on laser marking is proposed,through the calculation of laser line contour extraction and screening,laser line positioning and laser line relative position,Gets the coordinates of the rail edge line and the key pixels of the wheelset edge line.In order to solve the conversion relationship between real coordinates and pixel coordinates,according to the characteristics of the wheelset running image collected,a camera ranging model based on geometric relationship is established,and the camera pitch angle is solved in real time by using the vanishing point coordinates to avoid the impact of the camera pitch angle due to the slight vibration of the vehicle during the video recording process;for the running scene where the vanishing point is difficult to collect,a calculation model for the conversion relationship between the pixel distance and the real distance is established with the rail width as the reference.Combined with the algorithm of rail detection,the wheelset displacement measurement based on video image is realized.According to the rail edge line detection algorithm and the camera line of sight calculation model,the wheelset displacement measurement system based on image detection is developed with MFC as the application framework,and the corresponding wheelset displacement measurement system is calculated and output with the wheel-rail image as the input.After test verification,the system can be used as a means to detect the traverse of the wheelset on the basis of ensuring a certain accuracy and the efficiency of the detection of the lateral displacement of the wheelset.