Predicting transitions across macroscopic states for railway systems

Area of Science:

• Socio-technical Systems Analysis
• Data Science
• Network Dynamics

Background:

• Railways are complex socio-technical systems with intricate interactions between technical and human elements.
• Unlike physical systems, railways lack governing laws for dynamic description, necessitating novel analytical approaches.

Purpose of the Study:

• To develop and validate a data-driven framework for analyzing macro-dynamics in socio-technical systems.
• To identify distinct operational states and predict transitions between them.
• To provide an early-warning system for large-scale disruptions in railway networks.

Main Methods:

• Dimensionality reduction to create a phase-space from micro-unit data.
• Clustering algorithms to identify system states (e.g., 'rest' and 'disrupted').
• Development of an early-warning metric based on state transition probabilities, evaluated using the Peirce skill score.

Main Results:

• Identification of 'rest' and 'disrupted' states based on deviations from the planned timetable.
• Successful prediction of large-scale railway disruptions up to 90 minutes in advance with significant skill.
• Demonstration of the framework's potential for real-time monitoring and management of network disruptions.

Conclusions:

• The proposed data-driven framework effectively analyzes macro-dynamics in socio-technical systems like railways.
• The early-warning metric shows promise in predicting and mitigating large-scale disruptions.
• The framework's principles are adaptable to other complex systems requiring real-time monitoring.