Short-term Passenger Flow Forecasting Based on Phase Space Reconstruction and LSTM

Abstract

In this paper, the chaotic characteristics of the railway passenger flow are considered, and the PSR-LSTM (Phase Space Reconstruction-Long Short Term Memory) model is proposed by the phase space reconstruction method to recover the hidden trajectory in the passenger flow. First, this model uses C-C method to calculate the time delay and embedding dimension, and carry out phase space reconstruction. Afterwards, the LSTM neural network is used to predict short-term passenger flow. In the experimental part, it is proved that the passenger flow data with chaotic characteristics are reconstructed by phase space processing can get more accurate predictions. Then, in order to further verify the accuracy of the model, this model is compared with the BP neural network model and the SVR model, which is also subjected to phase space reconstruction processing. The experimental results show that the model has high accuracy.