Disaster and Passenger Support for Public Transportation Hubs: Congestion prediction and real-time escalation

Large public transportation facilities such as airports and passenger terminals can become rapidly overcrowded when severe weather or unexpected emergencies occur. Legacy systems made it difficult to understand congestion precisely in real time, which created an information gap for travelers who needed timely guidance. At the same time, weak data linkage with related agencies made it hard to respond proactively with measures such as charter-bus deployment, staffing changes, and passenger support during large-scale stranding events.

The solution integrated reservation data from transportation operators, telecom base-station data for large-scale population flow, and on-site LiDAR sensor data through a real-time data-flow engine. Processed data was used to predict congestion by zone and floor at minute-level intervals. Based on predicted congestion levels and disaster phases, the system could automatically decide and trigger linked transportation support and guidance measures through a digital disaster-management architecture.

This architecture made it possible to target more than 90% accuracy in congestion prediction within large public facilities and reduce delays in transportation support and situation escalation by more than 50% during exceptional events. Real-time infrastructure visibility encourages voluntary crowd dispersion, while automatic escalation improves coordination speed across multiple organizations. The result is a standard model for intelligent safety operations that improves infrastructure reliability, reduces administrative cost, and helps operators respond proactively to dense passenger concentration.