202503 – Afshar – CBTC Signalling System & Emerging Technologies; AI, Machine Learning & Crowd Computing for Adaptive Real-Time Train Timetables

Urban rail networks face challenges in train timetable rescheduling, particularly in high-density areas with fluctuating passenger demand. Communication-Based Train Control (CBTC) has improved operations’ efficiency and capacity, but disruptions like technical failures, overcrowding, and delays still impact efficiency. This article explores AI-driven real-time train scheduling, leveraging surveillance cameras and ticketing data for adaptive timetabling.

Train Timetable Rescheduling (TTR) is an NP-hard problem, making real-time solutions complex. Traditional approaches struggle with dynamic disruptions, but AI and CBTC (if operating in GoA2 -GoA4) integration allows adaptive scheduling using real-time passenger data, optimised speed management, and flexible dwell times.

Crowd computing processes data from CCTV, ticketing, and sensors to analyse congestion patterns. AI-powered computer vision estimates passenger density and movement trends, while ticketing data helps predict demand and optimise train frequency.

AI techniques such as Reinforcement Learning (RL), Genetic Algorithms (GA), and Multi-Agent Systems (MAS) enhance scheduling efficiency by learning from disruptions and optimising train movements.

Integrating AI-driven crowd computing with CBTC might reduce congestion, improves punctuality, and optimises energy use. Challenges like privacy concerns, computational complexity, and infrastructure upgrades remain, requiring further research on AI-based railway optimisation and cloud computing solutions, considering network limits.