The railway industry is crucial for a country’s infrastructure, enabling the efficient transport of products over long distances at a lower cost. One of the main objectives of modern railway engineering is developing automatic train control systems to enhance operational efficiency and safety, save energy, and reduce costs. This paper proposes a planner capable of creating driving models for heavy haul trains on a real railway section, respecting speed limits without compromising travel time and fuel consumption. The planner was trained using a reinforcement learning algorithm called Deep Q-Network (DQN) and an analytical train driving simulator also developed in this work, which models the dynamic of a railway composition. Results show acceptable travel times but there is room for improvement in respecting speed limits, specially when it is close to the end of the journey. Fuel consumption slightly increased compared to current standards, indicating the need to improve the DQN algorithm’s reward function.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Driving Planner for Freight Trains Using Reinforcement Learning


    Additional title:

    Communic.Comp.Inf.Science



    Conference:

    World Congress in Computer Science, Computer Engineering & Applied Computing ; 2024 ; Las Vegas, NV, USA July 22, 2024 - July 25, 2024



    Publication date :

    2025-05-03


    Size :

    16 pages





    Type of media :

    Article/Chapter (Book)


    Type of material :

    Electronic Resource


    Language :

    English




    Energy saving driving methods for freight trains

    Lukaszewicz, P. | British Library Conference Proceedings | 2004


    Driving techniques and strategies for freight trains

    Lukaszewicz, P. / Wessex Institute of Technology | British Library Conference Proceedings | 2000


    Longer freight trains

    Vogel, H. | British Library Online Contents | 2000


    Heavy freight trains

    Engineering Index Backfile | 1907