University of Northumbria and Zetica Limited KTP 24_25 R4

A train’s trackbed—the layer of stone and ballast beneath the rails—is inspected by bouncing radar waves off it, but interpreting those radar signals is currently slow and imprecise. This project builds a computer simulation that models how ground penetrating radar (GPR) behaves across different trackbed conditions and antenna setups. The problem is that today’s GPR analysis relies heavily on human judgement and trial-and-error, making it hard to spot hidden defects like water pockets, voids, or degraded ballast before they cause track settlement or derailments. If the simulation works, railway engineers could run virtual scans of any trackbed scenario, train algorithms to recognise specific faults automatically, and shift maintenance from reactive repairs to targeted, data-driven intervention. That would mean fewer disruptive track closures, lower inspection costs, and longer track life—improvements that keep trains running on time and reduce the risk of failures that ripple across supply chains and commuter networks. This is applied engineering research with a clear practical target: a digital tool that makes a routine but critical infrastructure task faster, cheaper, and more reliable.