UNSUPERVISED MACHINE LEARNING FOR REAL TIME SAFETY MONITORING IN RAILWAY ENVIRONMENTS

Abstract

Ensuring safety in railway environments is a critical concern due to the high risk of accidents, equipment failures, and human errors. Traditional safety monitoring methods, often relying on manual inspections and rule-based systems, are limited in their ability to process large volumes of real-time data and detect subtle anomalies. This study proposes an unsupervised machine learning (ML) approach for real-time safety monitoring in railway environments. By continuously analyzing data from sensors, CCTV cameras, and train operation systems, the system identifies abnormal patterns and potential hazards without requiring labeled datasets. Techniques such as clustering, anomaly detection, and autoencoders are employed to detect unusual events, including equipment malfunctions, track obstructions, and unsafe human activity. The unsupervised ML model enables early warning and predictive alerts, enhancing operational safety and reducing the risk of accidents. Experimental results demonstrate that the system can effectively identify anomalies in real time, providing a scalable, adaptive, and intelligent solution for railway safety management. This approach not only improves situational awareness for railway operators but also supports proactive decision-making to prevent accidents and ensure passenger.