The automatic train regulation is essential for train operation in urban rail transit with uncertain passenger demands. This paper aims to propose an algorithm that can minimize the delays and the travel time of passengers through learning the feedback from the simulator or environment. A traffic model for metro is developed as a simulator to train the algorithm, considering more passenger characteristics and uncertain disturbances. Furthermore, the state transition process is represented by a recurrent neural network, which enables the short-term predictive control and long-term planning. Finally, the real-world cases are used to test and verify the effectiveness and robustness of the proposed algorithm. The results show the algorithm can efficiently search the near-optimal solution online within one minute.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    An Intelligent train regulation algorithm for metro using deep reinforcement learning*


    Contributors:
    Liu, Yafei (author) / Tang, Tao (author) / Yue, Lei (author) / Xun, Jing (author) / Guo, Hui (author)


    Publication date :

    2018-11-01


    Size :

    547083 byte





    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English