Real-Time Train Wheel Defect Detection with Fiber Optic Acoustic Sensing

Wheel defects pose a significant threat to the operational safety and integrity of railway lines up to derailments and rail damage. Current detection methods like Wheel Impact Load Detectors (WILDs) are point-based, limiting their ability to provide continuous monitoring. Fiber Optic Acoustic Sensing (FOAS), which uses existing fiber optic cables alongside tracks and is already used for train tracking, offers the ability to detect wheel defects in real-time across large sections of track. This paper presents a FOAS-based system for automatic wheel defect detection and evaluates its performance against data from a commercial WILD system. Train movements were measured and analysed over an operational railway line. Train-induced vibrations were measured and analysed to generate a unique “acoustic fingerprint,” mapping the vibrational energy distribution along the train's length. Wheel defects were automatically detected by identifying prominent peaks within these fingerprints. Comparison with WILD data from 3,657 trains on two tracks indicated an accuracy of 95% and a false detection rate of 5%. While there is room for further improvement, these results demonstrate a clear potential for FOAS systems to enhance railway safety. Unlike point-based systems, FOAS could enable early detection of wheel defects anywhere on a railway line, preventing damage, saving costs and increasing operational efficiency.