Exploration of Machine Learning-Enabled Prediction and Control Algorithms for Railway Traffic Management

This study explores the application of machine learning-based prediction and control algorithms to improve the efficiency of rail traffic management. Inspired by the need to coordinate different public transportation modes to meet diverse travel demands, we propose a novel paradigm that integrates machine learning techniques. Through a review of innovative methods in railway traffic management, we identify the potential of machine learning algorithms in optimizing train schedules, mitigating disruptions, and improving passenger safety. Specifically, the study presents a Railway Traffic Direction Management System (RTDMS) that uses short-term traffic flow prediction as its cornerstone. The study develops a predictive model based on machine learning techniques to forecast passenger flow, enabling efficient resource allocation and improving overall system performance. Furthermore, the study constructs a model to facilitate RTDMS and provide a systematic experimental evaluation of its performance. The experiments show significant differences in prediction errors for different OD pairs at different temporal granularities, highlighting the importance of temporal granularity in traffic forecasting and planning.