| With the increasingly mature of communication technology,especially radio technology,the train operation control system based on this technology has become a research hotspot.In the train operation control system,train positioning technology is one of the necessary technologies.In order to make the train run efficiently and safely,it is necessary to obtain accurate-train position information.Therefore,it is of great significance to study the train positioning technology.At present,there are many kinds of train positioning technologies,each of which can meet the specific needs of train positioning to some extent,but they also have their own inherent defects.For example,in urban rail transit and tunnel,the traditional GNSS positioning technology will be limited by the environment and cannot play a role.Currently,the widely used responder-query positioning method is intermittent in the transmission of information,which cannot meet the requirements of real-time acquisition of train location information,and a large number of sensors have to be placed.Single positioning technology cannot meet the requirements of low density rail side equipment,moving block,high precision and high reliability of train positioning.Multi-source information fusion to realize the complementary advantages of each location information source is the development trend of train positioning technology.In this paper,a new track geometry signal(track attitude and track gauge)is proposed to use for matching positioning.If the range of track attitude and track gauge changes greatly,it will become a "harmful signal" that should be avoided for the safety and comfort of train operation,but it is a signal resource that can be used for positioning application.This paper explores its positioning potential and uses it to assist the train positioning system to realize "turning waste into treasure".In this paper,the spatial and temporal variation characteristics of the geometric features of the track are analyzed.By unifying the data collected from different working frequency sensors and different time to the same range point,the data measured for multiple times has the position correlation.The method of calculating the maximum Pearson correlation coefficient is used to find the best matching position between the background data and the matching data.On this basis,the matching positioning accuracy of several geometric features of the track is analyzed,and the factors affecting the positioning accuracy are discussed.In this paper,a multi-information fusion train positioning algorithm is designed.On the one hand,in the case of a given initial approximate position(rough positioning),the IMU/ odometer autonomous navigation solution is used to obtain the predicted train mileage,which is used as the mileage index of track geometric features.On the other hand,the IMU combines with the track gauge sensor to measure the geometric features of the track and matches with the background feature map to obtain the distance external observation.Kalman filter is used to fuse inertial navigation and matching positioning,to obtain the accurate distance(that is,positioning information).The IMU sensor error is estimated and compensated online.The algorithm model is verified and optimized by data acquisition experiments with several sets of inertial navigation of different levels and other sensors mounted on the rail inspection vehicle platform and train platform.The attitude repeatability collected by different levels of inertial navigation is analyzed and the influence of different length feature matching Windows on the positioning results is analyzed. |
PDF Full Text Request