Imagine being part of a railway maintenance team, where your responsibility is to assess the structural integrity of rail tracks to ensure train safety. The conventional approach of manually inspecting each track segment is quite labor-intensive. Vibrations occur as a train passes over the track, and any damage to the track can lead to variations in these vibration signals. The hypothesis is that by measuring the vibration signals from the sensors installed on a specific train vehicle component and enabled by machine learning, we can instantly detect the anomalies as the vehicle passes over damaged areas of the track.
This project aims to improve the technique that utilizes vibration analysis and integrates machine learning for detecting damages in railway tracks. It is an extended project from last year, in which a model train vehicle system has already been developed integrating the vibration measurement and analysis and damage detection algorithm. However, some issues exist:
· The existing controller cannot perform real-time vibration analysis.
· The current damage detection can only indicate damaged or undamaged condition without identifying the location of damage.
· The code and algorithm of the controller needs to be upgraded and optimized for higher efficiency, accuracy and robustness.
Undergraduate
Improve the existing model vehicle control system and algorithm to achieve real-time vibration data analysis, and promptly alert specific structural damages detected as the vehicle passes over them, with an emphasis on faster and more accurate damage detections.
None
Dynamics & Control Lab (405.852, Lab)