Accurate prediction of freight volume data is directly related to the preparation of railway transport plans. This paper proposes the Informer model to predict freight volume at major stations of railway network. First, the freight volume data is transformed by temporal embedding, which considers whether the day is in the skylight maintenance period, to better explore the time characteristics. Next, the decoder structure containing the ProbSparse self-attention mechanism is applied to capture the long-term correlation from the embedded data. Finally, the decoder structure is used to generate forecast information at once. Using the freight volume data of China Energy Investment, the high-precision railway freight volume prediction of twelve loading stations in the next 14 days is achieved. Experiments show that Informer model exhibits better prediction performance with the growth of the sequence: the RMSE value is improved by 90/0-27.4% over LSTM, GRU, LSTM with Self-Attention and Transformer.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Improving the Accuracy of Multi-Step Prediction of Railway Freight Volume Based on Informer Model


    Contributors:
    Liu, Jiaqi (author) / Jing, Yun (author)


    Publication date :

    2024-01-12


    Size :

    1646137 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Research on Railway Freight Volume Prediction Based on ARIMA Model

    Zhao, Jianyou / Cai, Jing / Zheng, Wenjie | ASCE | 2018



    Railway Freight Volume Prediction Based on Support Vector Regression (SVR)

    Liu, Yan ;Lang, Mao Xiang | Trans Tech Publications | 2014



    Prediction Models for Railway Freight Volume Based on Artificial Neural Networks

    Sun, Yan ;Lang, Mao Xiang ;Wang, Dan Zhu | Trans Tech Publications | 2014