Railway Track Alignment Inspection Method Based On Extended Kalman Filter

With the rapid development of railway transit,the smooth track state is an important basis for ensuring traffic safety and comfort,track alignment detection method is one of the key technologies for maintenance and maintenance of track lines.How to accurately measure and calculate the track base parameters of railway space is the key theory and technology to realize the track irregularity calculation.Multi-Sensor Based on machine vision and fusion technology inertial unit,to build a data set of binocular fusion gyro binding model,to provide a high-precision detection method based on extended Kalman filter algorithm calculate track parameters of railway.Firstly,based on the camera calibration principle,we calibrate the image collected by the camera,and calculate the internal and external parameters of the left and right cameras and the rotation and translation between the cameras.Secondly,the four element method is used to calculate the gyro attitude vector,optimization of gyroscope attitude vector by extended Kalman filtering algorithm,and the rotation relation between the camera and the gyroscope is obtained according to the robot hand eye calibration principle.Again,the extended Kalman filtering algorithm is established based on the system state matrix,using gyro attitude information correction camera position vector,uninterrupted update displacement direction of binocular vision system,camera motion in the world coordinate system,so as to obtain the parameters of railway track.In this paper,combining the high-speed dynamic detection performance of gyroscope with the excellent spatial detection expertise of binocular stereo vision,the Kalman filtering algorithm combines two kinds of measurement data to effectively suppress the gyro drift and make up for the shortcomings that the visual sensor is greatly affected by the environment,quickly estimate the three-dimensional coordinates of the vehicle in railway,attitude and speed.Experimental platform was set up for verification.In the experiment,the camera shot the calibration plate at a frequency of 5 Hz.The angular velocity of the gyroscope at a frequency of 100 Hz in the global coordinate system was used to write the program setup model with Matlab.The acquired gyroscope angular velocity and binocular Image processing,get the raliway track of the system motion,verify the accuracy and accuracy of the algorithm.The results show that the binocular vision and gyroscope data fusion can effectively realize the orbital linear space measurement.The displacement error of the measurement coordinate in the three directions of x,y and z do not exceed 0.381 mm.Simultaneously,the simulation results show that fusion model error does not accumulate with increasing distance measurements,can meet the requirements of long-distance measurement.