Train localization system is a vital part of train control system. A novel independent train localization method based on multi-sensor fusion is proposed in this paper. In conventional train localization system, the position determination is dependent on the reference information source controlled by rail operator, such as track circuit and so on. In some new train localization systems on test[1][2][3][4], the localization methods based on Satellite Navigation are very popular, so that the system’s reliability and safety are put outside of the rail operators’ control actually. The localization method proposed in this paper uses inertial sensor gyrometer and conventional odometer, which is mounted on the locomotive’s driven axle. Gyrometer is used to measure train’s heading angular rate. And odometer is used to measure train’s velocity. Usually, the dead reckoning method independent of outside reference information is selected for these two sensors’ fusion. But an obvious disadvantage of this method is the position error is increasing linearly according to time, so that this method can not work independently. Therefore Satellite Navigation system is often used to restrict the increasing error to an acceptable range. The method proposed in this paper overcomes the disadvantage of dead reckoning. Because the rail route or train’s moving track is fixed, the route’s curvature-mile curve can be obtained. According to this curve it is obvious, in the area where route is straight, the curvature value is zero; in the area existing a turning, the value is not zero. DGPS equipment is mounted on the test train to get and record the position data on the test route. Then the curvature-mile curve can be calculated from these data. We use Hurst coefficient to get the characters of train track such as turning from the gyrometer and odometer’s data. By matching these characters and the known route’s curvature-mile curve, the train’s real-time position can be calculated. If the dual-direction communication channel such as GSM-R is available, virtual block or moving automatic block in train control system could be achieved. Details of this process are presented in this paper with some results to illustrate the effectiveness of the methods.
Real-Time Self-Localization Method by Using Measurements of Directions of Two Landmarks and Dead Reckoning
