Determining a Train Location at Level Crossing Approach Section by Track Circuit Input Impedance Processing Using ANN

Abstract

The possibility of determining a train location at level crossing approach section by measuring the input impedance of the track circuit with subsequent processing of the results using an artificial neural network has been carried out. The need to measure the coordinates and speed of the train at the section approaching the level crossing is associated with the need to improve traffic safety at the level crossing and reduce the waiting time for a train (especially for a slow moving one) for vehicles. The investigations of the track circuit input impedance dependences on factors such as the train coordinate on it, the resistance of the ballast layer of the track and the frequency of the signal current were carried out by modeling and experimental measurements on a real track circuit. The monotonic dependences of the track circuit input impedance on the train location make it possible to use these dependences to monitor the train location on the track circuit. However, the strong influence of the ballast resistance on the input impedance makes it impossible to directly use the input impedance for positioning the train. Two methods have been considered for determining the coordinates of a train on a track circuit by using its input impedance. The error in determining the train coordinates at the level crossing approach section by processing the track circuit input impedance using ANN shows that this method is promising for practical use.