Urban Rail Transit Passenger Flow Prediction Based on Machine Learning

Abstract

As the scale of urban rail transit networks continues to expand, accurate passenger flow forecasting has become a key technical support for optimizing operational scheduling and improving transportation efficiency. This study comprehensively summarizes the advances of machine learning in rail transit passenger flow forecasting, and elaborates the fundamental theories and characteristics of three representative technologies: support vector regression, long short-term memory networks, and spatiotemporal graph convolutional networks, and discusses the challenges faced by current methods in terms of adaptability to extreme weather, multi-source data fusion, and computational efficiency. This paper shows that spatiotemporal graph convolutional networks have become the mainstream technological paradigm in this field, but breakthroughs are still needed in dynamic graph construction, long-term dependency modeling, and industrial-grade deployment. Future research should focus on the integration of physical information, lightweight model design, and the improvement of interpretability to support the sustainable development of intelligent rail transit systems.