This paper focuses on the application of AI for railway infrastructure asset management and the potential for AI to facilitate optimisation of the rail system. It is based on Cordel’s own automated sensor data acquisition (covering 35,000 track miles of data service per month) and our processing and analytics outputs, with our growing range of railway-specific AI/ML training sets.
We begin by characterising three types of AI, including the Machine Learning AI that we are experienced in applying for railway engineers. We then work through our methodology with a case study of a specific implementation applied to LiDAR and other data captured from Cordel sensor sets installed in October 2022 (and still running) on Class 165 passenger trains (owned by Angel Trains). These trains are operated by Great Western Railways in normal passenger service on Network Rail infrastructure. We highlight one of the asset management applications that we delivered for Overhead Line Equipment (OLE) engineers, which focused on reliably and repeatedly measuring the Heights and Staggers of catenary wire along the electrified route.
The key aim for Heights and Staggers was to build local engineers’ trust in the data and the AI; in order to show that it could efficiently assure standards compliance, and promptly detect change, enabling corrective action. Since this aim was demonstrated, we have embarked on the next stage: to embed the application of AI into NR’s asset management systems. This stage will integrate our actionable insights into established NR enterprise IT architecture to enable the most cost-effective pro-active prevention of impact on operational train service delivery.
