Real-Time Mobility Prediction and Management in Urban Railway Systems

Passengers are experiencing peak period crowding due to rapid urbanization in many cities. Providing real-time information is an effective way in managing crowding by informing smart travels. For example, passengers may change their departure times or adjust their routes to avoid peak crowding. Automated Fare Collection (AFC) data, recording origin and destination stations/times of passenger trips, provides opportunities to understand collective and individual mobility. The dissertation aims to develop real-time mobility prediction methods and incentives strategies for crowd management using AFC and related data. The main contents include missing data imputation, mobility pattern recognition, OD flow prediction and personalized incentives.