News & Notice
Press releases and events to keep you updated about recent government innovation activities.
[Press Release] Real-time Subway Platform Congestion Prediction Model
- Admin
- 2024-01-09
- A real-time platform congestion calculation model for the subway in Seoul and the Gimpo Goldline was developed.
- With 90% accuracy, it can quickly detect crowd surges and allow immediate on-site measures following the manual.
- From November, the analysis model will be piloted in two subway stations in Seoul, and the Ministry of the Interior and Safety plans to standardize it for nationwide implementation.
The government has completed the development of an artificial intelligence (AI) based data analysis model that can monitor real-time congestion at subway station platforms, which will be piloted starting this month.
The Ministry of the Interior and Safety (MOIS, Minister Lee Sang-min) announced that the Integrated Data Analysis Center (IDAC) has completed the development of the "AI-based Subway Platform Congestion Prediction Model." This model has been under development using the subway in Seoul and the Gimpo Goldline as samples since June, and it will be piloted on the Seoul subways starting this month.
The model was developed based on AI-generated estimates of the number of people on the subway platforms. It calculates the density and congestion level, considering the platform area, and classifies it into four levels from Level 1 to Level 4 to represent the congestion.
The IDAC, the Seoul Metro, and the Gimpo Goldline joined the model development process, using 8 million pieces of data, including subway entry and exit tag data, transportation card settlement data, and train departure and arrival data.
Platform occupancy refers to the passengers waiting for the subways on the platform after passing through the entry gates and the passengers who move through the platform to the exit gates after getting off the train.
The real-time platform occupancy at the station is determined by a comprehensive analysis of factors such as the real-time number of people passing through entry and exit gates, the number of people who passed through entry and exit gates at the previous station, and historical entry and exit data for the current period.
According to the technical standards of the Railway Safety Management System (Design Guidelines for Urban Railway Stations and Transfer Facilities), the congestion level is calculated based on the percentage of people exceeding the baseline of 4.3 people per square meter (100% congestion).
Real-time platform occupancy: 300 people, platform area: 50 m², per-area standard occupancy: 4.3 people, → Congestion level: (300 ÷ 50)/4.3 = 1.395 (139.5%) → Congestion level: Caution
The two- rounds of performance verification results confirmed an accuracy of 90.1% for the analysis model.
The developed model has been integrated into the Seoul Metro's control room dashboard. Seoul Metro uses this model to monitor the congestion levels of two stations displayed in real time.
Along with the development and implementation of this model, the Seoul Metro has also revamped its congestion response system. In the event of unexpected crowd surges, the system is designed to automatically disseminate information and take proactive on-site measures following the congestion management manual.
MOIS and the Seoul Metro believe that the utilization of this analysis model will help effectively respond to congestion situations within subway stations. Scientific monitoring of platform congestion will enable proactive on-site measures at different levels, leading to a tangible effect in preventing accidents.
MOIS plans to standardize the platform congestion calculation model through a pilot operation process within the year and expand it to subway stations in the metropolitan area and four other cities (Busan, Daegu, Gwangju, Daejeon).
Vice Minister Ko Kidong stated, “As a concrete achievement of the Digital Platform Government, it will enhance the public convenience when applied to the subway.” “We will continue to support the effective prevention of subway crowd accidents as we have overhauled the response system for congestion,” he added.
#Big_Data #Real_Time #Subway #Congestion_Prediction_Model #Agile_Government #Data_Based_Government #Ministry_of_the_Interior_and_Safety