High accuracy navigation system based on multisensor information for high-speed train

Abstract

In the Korean rail industry, tachometers and track circuits are used for the train navigation system. However, tachometers and track circuits cannot provide a high-accuracy navigation solution; however, this is essential in the rail industry. If a high-accuracy navigation system is adopted for rail track inspection cars, the maintenance cost of train service will be reduced. Generally, information gathered by various sensors is used to improve the accuracy of railway navigation systems. This paper proposes a sensor fusion algorithm in which INS/Tachometer/Doppler radar/DGPS/RFID/NHC sensors are combined. In addition, a map matching algorithm for high-speed trains is proposed. The proposed INS/Tachometer/Doppler radar/DGPS/RFID/NHC sensors fusion algorithm called high accuracy navigation system based on multisensor information for high-speed train(HANST) is based on the federated Kalman filter. The federated Kalman filter consists of local filters and a master filter. In this paper, 6 local Kalman filters are used in the implementation of a federated Kalman filter. RFID local Kalman filters are specially designed for aperiodic measurement update. Moreover, RFID Kalman filters provide stability for whole navigation systems. Generally, the performance of federated Kalman filter-based navigation system depends on DGPS performance. If the DGPS signal is cut off, the performance of a federated Kalman filter-based navigation system decreases. However, if the proposed RFID filter is implemented, the accuracy of the whole navigation system remains stable in the DGPS signal cut off section. A map matching algorithm for high-speed train navigation systems is also proposed. Generally, map matching algorithms are studied for car navigation. In car navigation, car map characteristics are used to improve the accuracy of map matching. For example, if a turn course is in the map and the navigation solution history has turn trace, the map matching algorithm matches the corner ...