"The railway system is a critical component of the transport sector, offering seamless connectivity between city centers, regional hubs, and distant rural areas. Its long-distance reachage and high-volume capacity enable the efficient movement of passengers and goods, while its energy efficiency and strong multimodal integration help alleviate urban road congestion and enhance overall network performance. Maintaining safety, efficiency, and operational capacity is essential to preserve these advantages and sustain the railway system’s competitiveness over time. Traditionally, maintenance strategies have been based on preventive inspections focused on scheduled reviews and manual record keeping, along with the predefined replacement of components and parts without integrity or predictive-based analysis. Moreover, prediction lackage usually leads to repair and replacement interventions executed after fault detection, consequently generatinghigh costs and operational congestion and unavailability. Hence, these maintenance strategies with marginal predictive approach, generate unforeseen events, significant labor costs, planning difficulties, operational impacts, and higher safety risks due to the lack of early detection.
Artificial Intelligence (AI) has demonstrated substantial effectiveness in processing and interpreting large-scale datasets, positioning it as a highly suitable technology for integration into railway maintenance systems. Driven by accurate analysis and prediction of real-time data with high parameter variability, enabling the early detection of anomalies, reducing the manual inspections dependency, and improving the infrastructure resilience. This study presents an optimization maintenance model based on a machine learning failure prediction, real-time AI data analysis, correlated with operational scenario and maintenance cost shared between corrective and predictive action, providing a maintenance plan action based on sensor, auscultation and predictive action, for the sake of minimizing the reactive/correctives interventions.

Predictive maintenance; Railway maintenance; Maintenance optimization; Artificial intelligence