Demand Forecast of Railway Transportation Logistics Supply Chain Based on Machine Learning Model

Abstract

The deep learning method based on long short-term memory (LSTM), gated recurrent unit (GRU), and bidirectional LSTM (Bi-LSTM) was constructed by researching the factors affecting railway transportation logistics. Moreover, a simulation study on Tianjin Station was conducted. The deep learning model suitable for the logistics demand forecasting of Tianjin Station was established, and the changing trend of logistics supply chain demand in Tianjin Station in the future was analyzed. Moreover, a strategy for railway construction and regional cooperation was proposed. In this study, three deep learning neural networks, namely LSTM, GRU, and Bi-LSTM, were used to construct a demand forecasting model for the logistics supply chain in Tianjin Station. Bi-LSTM, which has bidirectional storage performance and the highest prediction accuracy, is superior to the traditional neural network structure in terms of period and fluctuation.