Accurate passenger flow prediction of rail transit is of great significance to improve the operational efficiency of rail transit and reduce operational costs, which are essential for the proper scheduling and optimization of rail transit. The existing research could not ensure that the model has a good prediction effect in different stations. To improve the prediction accuracy of different stations of rail transit, an urban rail transit passenger flow prediction with combining Gaussian mixture model (GMM) and Gaussian process regression (GPR) based on complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) is proposed. It integrates CEEMDAN, principal components analysis (PCA), GMM, and GPR, which will effectively extract the fluctuation characteristics of passenger flow, make stations clustering more reliable, and improve prediction accuracy. Results from an application to Harbin Metro Line 1 data show that the urban rail transit passenger flow prediction that combines GMM and GPR based on CEEMDAN can extract effective implicit information of passenger flow, can perform well and stably in the passenger flow prediction of each station, can reduce the prediction errors of different stations, and can improve the prediction performance of the model.
Urban Rail Transit Passenger Flow Prediction with Gaussian Process Regression Based on CEEMDAN
01.09.2025
Aufsatz (Zeitschrift)
Elektronische Ressource
Englisch
Urban rail transit passenger flow prediction method under emergency
Europäisches Patentamt | 2023
|Urban Rail Transit Passenger Flow Forecasting—XGBoost
ASCE | 2022
|Urban Rail Transit Passenger Flow Forecasting - XGBoost
TIBKAT | 2022
|