At present, there are almost no reports on vehicle scheduling methods for three-dimensional rail transit system, and the real-time performance of existing vehicle scheduling algorithms is poor. Aiming at the scheduling problem of three-dimensional rail transit vehicles, this paper studied an order allocation algorithm combining high and low frequency station judgment and a Dijkstra path planning algorithm combining time window, namely intelligent scheduling algorithms, to improve the operating efficiency of vehicles. Firstly, it used the order allocation algorithm to select the appropriate execution vehicle for the order to reduce the waiting time of passengers. Secondly, it added the judgment of high and low frequency stations on the basis of the order allocation algorithm, and scheduled vehicles to the high frequency stations in advance to ensure the balance of supply and demand. Then, it combined the ordinary Dijkstra algorithm and time window judgment to realize multi-vehicle conflict-free path planning. Finally, it redeveloped the OpenTCS software and simulated the scheduling algorithm with the software. The results show that the average waiting time of the passenger from calling the vehicles was 8.043s only using the order allocation algorithm. After the advance scheduling of vehicles combined with high and low frequency stations, the average waiting time was reduced to 5.724s, and the waiting time of each passenger was reduced by 2.319s. During the path planning, both the ordinary Dijkstra algorithm and the Dijkstra algorithm combined with time window took less than 1ms to plan. However, the Dijkstra algorithm combined with time window only increased the time by about 0.1ms, and solved the problems of the vehicle path conflicts. The studied intelligent scheduling algorithm can reduce the waiting time of the passengers and improve the running efficiency of vehicles. The algorithm has good real-time performance and can meet the scheduling requirements of three-dimensional rail transit vehicles.

[2023-12-08] Accepted Paper /doc/2023.09.0427/v1