Public Traffic Passenger Flow Prediction Model for Short-Term Large Scale Activities Based on Wavelet Analysis

Abstract

The short-term large scale activities refer to various large-scale activities with a duration of several hours, with features of high peak passenger flow and short gathering time. The analysis of public transport passenger flow characteristics and travel demand prediction for large-scale activities can provide a targeted organization plan for public transportation security in the context of large-scale activities. Based on the smart card data of Beijing, the paper analyzes the spatial-temporal characteristics of passenger flow under the background of large-scale activities. The Discrete-Fourier transform is used to study the frequency domain characteristics of large-scale active passenger flow sequences. Then, through the steps of sampling, decomposition and reconstruction of passenger flow sequence features, the public traffic passenger flow prediction model for short-term large scale activities based on Wavelet analysis was established. And reconstruction steps to establish a short-term large-scale public transport passenger flow forecasting method based on wavelet analysis. The method overcomes the weaknesses that data detail information are ignored in large-scale forecasting during modeling, and improves the stability of forecasting results in short-term forecasting. A case study of Beijing was conducted to validate, and the result shows that the mean absolute percentage error (MAPE) and mean absolute error (MAE) are 0.22% and 1.47%, respectively.