A Study on Short-Term Passenger Flow Forecasting for Urban Rail Transit

Abstract

Short-term forecasts of light rail traffic are an indispensable basis for adjusting timetables in real time, improving operational efficiency and providing passengers with informed travel recommendations. At the same time, the subway is the most convenient and widely used mode of transport for people today. The purpose of this paper is to forecast passenger traffic in the subway. To solve the problem of predicting short-term passenger flow at the entrances and exits of metro stations with nonlinear and stochastic characteristics, this paper uses a genetic algorithm (GA) based on ordered long-term and short-term memory (SLSTM) and builds a GA-SLSTM forecasting model to predict future short-term passenger flow by studying the patterns and characteristics of passenger flow in historical data. The model consists of five stages: data preprocessing, model training, model evaluation, and forecast application. Through these steps, the GA-SLSTM model is able to predict the future situation of short-term passenger traffic based on the characteristics and patterns in historical data, providing a reference base for planning and optimizing urban transport.