Research On The Intelligentization Of Train Dispatching Plan Generating For The EMU Depot Based On Deep Reinforcement Learning

The EMU depot is a place for EMU trains’ parking and level 1&2 maintenance.With the rapid development of China’s high-speed railways,the total number of EMU trains keeps increasing,more and more EMU depots are built by different railway bureaus.The EMU depot usually contains lots of parking tracks,maintenance tracks and devices which are in complex situation.The quality of train dispatching plans would directly decides the efficiency of maintenance work in the EMU depot and has an influence on the use of EMU trains.At present,based on the EMU depot Centralized Control System(CCS)and Centralized Traffic Control System(CTC),train route has been executed automatically,but the train dispatching plans are still made manually.In order to improve the efficiency of maintenance work and fully utilize the capabilities of maintenance devices in the EMU depot,the intelligentization of train dispatching plan generating was studied in this paper.After a comprehensive analysis of status quo and difficulties of train dispatching work,an intelligent algorithm for generating train dispatching plan was designed based on deep reinforcement learning.Then the algorithm was programmed and an EMU depot simulation software was made to test the algorithm.The test result showed that the algorithm could generate train dispatching plans automatically according to the status of the EMU depot and maintenance requirements of EMU trains.The feasibility of the algorithm was proved.