查看文章

It is essential to provide accurate real-time forecasting to manage the intense passenger inflow (IPF) of urban rail transit (URT) stations caused by special events such as concerts and football matches. The IPF is predictable due to the fluctuations in passenger outflow before the special event, which also allows the management department to take measures to control the situation. By combining individual travel card data with event data, station data and others, this article proposes a system for estimating URT station IPF before it happens. It consists of two parts: (1) Offline model training is responsible for modeling the relationship between historical special events information, affected station information and traveler characteristics; (2) Online Inflow prediction takes the current event and affected station information as the input of the model trained in the offline part to estimate the IPF. By using the Shanghai, China URT …