Unsupervised Machine Learning For Managing Safety Accidents In Railway Stations

Railway stations are high-traffic public environments where safety accidents pose significant risks to human life and infrastructure.Traditional safety management relies heavily on manual monitoring and incident reporting, which can be inefficient and reactive rather than proactive.This research proposes an unsupervised machine learning approach to manage and mitigate safety accidents in railway stations.By analyzing video surveillance data, sensor logs, and environmental inputs, clustering and anomaly detection algorithms such as DBSCAN and Isolation Forest are employed to identify irregular behavior or hazardous patterns.These models work without labeled data, enabling real-time identification of unusual activities, potential collisions, or unsafe human behavior.The use of unsupervised techniques allows for scalable and adaptive systems that can be implemented across diverse environments.Results from real and simulated datasets show that this method significantly improves early detection of safety threats, enabling timely interventions and minimizing accident severity.