Determination of Difficult Parking Points in Train Running Section Based on UAS and BP Neural Network

Abstract

The trains of EMU are all electric locomotives. During the operation of EMU, many reasons such as bad weather, high voltage cable falling off, catenary failure, power supply system failure and so on will cause power outage of power supply network. The power of the train is lost, so it has to be passively parked for rescue or use its own on-board energy storage to carry out self-rescue to the nearest station. Once the train stops in the middle of the "V" terrain or in difficult rescue locations, the use of diesel Trailer rescue will consume a lot of energy and cause a lot of carbon emissions. To solve this problem, a BP neural network method based on Levenberg-Marquardt algorithm is proposed to determine the parking difficulties in train operation section using UAS simulation platform. Compared with UAS simulation data, the reliability of this method is verified.