Smart Railway Maintenance Using AI

02/03/2026

This work, conducted by researchers from the Information Processing and Telecommunications Center (IPTC) at Universidad Politécnica de Madrid (UPM) and the companies ADIF and SepsaMedha presents a comprehensive survey of Artificial Intelligence (AI) techniques applied to predictive maintenance in railway infrastructure.

Railway transport is a critical pillar of sustainable mobility, and its increasing digitalization requires advanced maintenance strategies to ensure safety, reliability, and operational efficiency. Traditional maintenance approaches, often reactive or rule-based, present limitations in detecting failures early and optimizing asset lifecycle management. This survey systematically analyzes AI-based approaches, including machine learning and deep learning techniques such as neural networks, support vector machines, random forests, and deep neural architectures. These models are applied to diverse data sources, including vibration signals, imaging systems, track geometry measurements, and sensor data.

The study provides a structured classification of AI models, railway subsystems, and maintenance strategies, covering infrastructure, energy systems, and signaling components. It also identifies key research challenges, such as data quality limitations, model interpretability, scalability, and integration with real-time operational platforms. Emerging technologies such as Digital Twins, Cyber-Physical Systems, and Edge AI are highlighted as promising enablers of autonomous and intelligent railway maintenance.

This research contributes to advancing intelligent asset management in railways and supports the transition toward fully digitalized, predictive, and prescriptive maintenance frameworks.

Potential applications include:

Early fault detection in tracks, signaling systems, and power infrastructure
Reduction of operational disruptions and maintenance costs
Extension of infrastructure lifespan
Improved safety and reliability of railway networks
Development of autonomous maintenance systems using Digital Twins and AI

These findings support the development of safer, more efficient, and sustainable railway transport systems.

Bibliographic reference:

Bris-Peñalver, F. J., Verdecia-Peña, R., & Alonso, J. I. (2026). A Survey of AI-Enabled Predictive Maintenance for Railway Infrastructure: Models, Data Sources, and Research Challenges. Sensors, 26(3), 906. https://doi.org/10.3390/s26030906

José Ignacio Alonso Montes: ORCID / LinkedIn

Randy Verdecia Peña: GS / ORCID / LinkedIn

Francisco Javier Bris Peñalver: GS / ORCID / LinkedIn

LinkedIn: https://www.linkedin.com/company/iptc-upm/

For more information: www.iptc.upm.es