A Moving Authority Generation Switch Decision Method for Virtual Coupling High-Speed Train

Abstract

Virtual coupling is an effective way to improve the capacity of high-speed railway. Moving authority generation is important for high-speed train operation safety and efficiency, which would have great significance for train operation with small distance in virtual coupling. In this paper, a framework of moving authority generation switch between the existing train operation control mode and train-to-train communication based mode has been established to the high-speed train operation control under disturbance in virtual coupling. Then, different logit models have been used to analyze the train-to-train communication mode applicable condition and mine the causes of moving authority switch to select the feature factors of moving authority switch decision. And a Convolutional Neural Network-Gated Recurrent Unit neural network module is designed and used to make decision for train moving authority generation switch. A simulation dataset of virtual coupling train trajectory with disturbance is collected from the model predictive control based train operation control. The experiment has been conducted with the simulation dataset, where the neural network feature factors of moving authority switch decision are selected. On the basis of moving authority switch causes analysis, the effectiveness of Convolutional Neural Network-Gated Recurrent Unit module in moving authority switch decision has been verified and tested. Moreover, the comparison among different moving authority generation conditions have been taken, and the effectiveness of the proposed decision method in train operation safety and efficiency has been verified.