Fault Prediction of Track Circuit Based on Bidirectional Gated Recurrent Unit with Attention Mechanism

In railway control system, track circuit is one of the most important parts. The failure of it might have significance negative influences on the safety of railway transportation system. In order to enhance the maintenance of track circuit, the faults of track circuit needed to be forecasted. So, this research using bidirectional gated recurrent unit (Bi-GRU) and attention mechanism (AM) to realize the fault prediction of track circuit. In this paper, the original data of track circuit at certain intervals are recorded by a microcomputer monitoring system, and then the features are extracted by manual experience. After that, a time series dataset of track circuit fault features is established as the input of Bi-GRU model with AM to build the track circuit fault prediction model. Finally, experiments are carried out and the results show that the proposed fault prediction model is able to forecast track circuit faults precisely, and has better prediction performance compared with classical fault prediction models.