10145604PRISM: Predictive Rail Innovation for Supporting MaintenanceClosedSmall Business Research InitiativeInnovate UKOur project focuses on developing a cutting-edge decision support system aimed at improving rail maintenance, specifically by optimising the allocation of rail cleaning interventions. Low rail adhesion, a persistent issue caused by contaminants like debris, sand, and vegetation blown onto tracks, poses significant challenges for the rail industry. These contaminants can manipulate the friction between the rail and train wheels, resulting in diminished braking efficiency and increased operational risks. Traditionally, train operators address this problem by applying dry sand to the rail-wheel interface, which helps improve traction. However, this method is far from perfect; its efficiency is often unpredictable, and frequent use leads to wear and tear on infrastructure, as well as operational delays. In many cases, rail cleaning is carried out either on a routine basis---whether necessary or not---or reactively after operators notice issues with adhesion. This approach wastes resources, increases operational costs, and heightens safety risks. Our proposed solution aims to revolutionise this process by leveraging the power of real-time data, numerical simulations, and artificial intelligence (AI) to inform more precise, data-driven decisions about rail cleaning. The key innovation in this project lies in its ability to use real-time environmental and operational data to predict when rail cleaning is necessary. By collecting information such as weather conditions (e.g., temperature, humidity) and contamination levels, our system will be able to simulate various rail conditions and their impact on adhesion. The AI component of the system will then analyse the data, identifying patterns and predicting potential adhesion issues before they occur. As a result, cleaning crew will receive accurate, real-time recommendations on when and where cleaning or sanding interventions are required, ensuring that these actions are only performed when absolutely necessary. Currently, the project is in its early stages, with conceptual development underway. The foundation of real-time data collection and numerical simulation technologies is well established, and the application of AI in decision-making processes is becoming increasingly common. The next steps include integrating these technologies into a cohesive system and validating the approach through testing and simulations. Ultimately, this system has the potential to significantly improve rail cleaning, lower maintenance costs, and improve safety across the rail industry. By transforming rail cleaning from a reactive to a predictive process, our solution will enable rail operators to optimise their resources and maintain safer, more efficient operations.