RAILWAY TRACK ANOMALY DETECTION USING INSTRUMENTED TRAIN DATA

This project will bring the latest advances in machine learning from computer science into structural health monitoring and railway engineering to develop an innovative solution for railway infrastructure assessment and maintenance. Instead of conventional railway track monitoring approaches such as inefficient method of ‘walking the track’ and expensive approach of using ‘track recording vehicles’, this project proposes using inexpensive, frequent and widely available data measured on in-service trains to monitor railway track condition. The proposed approach does not disturb service provision and can provide real time data on track condition which has a substantial potential to revolutionise railway
maintenance sector.