Enhancement of Railway Passenger Information Systems Using Temporal Fusion Transformers and IoT Data

Abstract

The integration of Temporal Fusion Transformers (TFTs) with IoT-driven data streams presents an effective approach for improving Railway Passenger Information Systems (PIS). This study introduces an innovative framework that utilizes real-time data from IoT sensors, including GPS, environmental, and ticketing systems, to predict train schedules, delays, and other significant events. A predictive model based on TFTs is developed to process and analyze temporal data, thereby enhancing prediction accuracy. Quantitative results indicate that the proposed system achieves a 15% improvement in delay prediction accuracy compared with conventional machine learning models, along with a 20% reduction in root mean square error (RMSE). Moreover, passenger wait times decrease by an average of 10% due to improved prediction accuracy. The platform supports real-time updates, enabling more dynamic and responsive passenger notifications. This study highlights the significant potential of integrating IoT data with TFTs to enhance operational efficiency, reduce delays, and improve the overall passenger experience in modern railway systems. The findings suggest that incorporating TFTs with IoT data can substantially transform railway Passenger Information Systems, leading to improved service reliability and passenger satisfaction.