loading page

Method to Reduce Errors in High-Speed Simulation for Automatic Train Operation
  • Jungtai Kim
Jungtai Kim
Korea Railroad Research Institute

Corresponding Author:[email protected]

Author Profile

Abstract

The use of simulators is essential for developing control algorithms for prototype trains. Simulators allow for sufficient testing of algorithms before testing on real trains, saving time and money. Recently, researchers have been trying to apply algorithms that require many iterations, such as reinforcement learning, to automatic train operation. For such a large number of iterations, it is important to have a simulator with fast execution time so that more trials can be performed. However, when high-speed simulation is implemented, errors that do not occur in real-time simulation occur, especially the positioning error, and this paper describes the cause of such errors and describes the configuration and interface method of the simulator to reduce the error. The suitability of the proposed algorithm is verified by comparing simulation results with actual train tests and existing simulators.