Anomaly detection and system diagnostics

The goal is to develop a train based system for monitoring track defects and rail track components.

Sponsors: Trafikverket/JVTC

Researchers: Praneeth Chandran (PhD candidate), Matti Rantatalo (PL)

Duration: 2017-2021

Objective

The goal is to develop a train based system for monitoring track defects and rail track components.

Project status and results

This Project will use the concept of Eddy current based inspection method, that can overcome the major challenges mentioned above. The main goal of this project is to develop a train based system for monitoring track defects and rail track components by anomaly detection in the modulated magnetic field, generated and measured by a differential eddy current sensor. The sensor used to detect the fasteners utilizes the eddy current principle, where a varying magnetic field is created over a conducting material for two different excitation frequencies of 18 kHz and 27 kHz. This varying magnetic field induces local circular currents called Eddy currents on the surface of the material, which in turn creates an opposing magnetic field, which are picked up by receiver coils within the sensor. The eddy currents generated on the surface of the material are dependent on the conductivity (μ), permeability (σ) and the geometric form of the material.
The pickup coils are differentially coupled, which implies sensitivity only to changes in generated eddy currents in rail and vicinity, not any absolute value can be detected. To investigate the possibility of detecting fasteners, the sensor was mounted 65mm above the railhead on a trolley system and was pushed along the track. Different measurements were conducted along the railway track with healthy track sections with intact E-clip fasteners and were compared with measurements of a track with missing clamps.

Individual clamps are easily distinguishable from both 18 kHz and 27 kHz plots. The zero crossing in the signal from the positive to negative values indicates the center positioning of the fastening system for each sleeper. It can be concluded that the sensor can detect fastener signatures from a distance of 65mm above the railhead. Furthermore, the study shows that missing clamps can be detected by analyzing the fastener signatures. The future scope in this study involves quantification of rail defects and developing efficient condition monitoring techniques with the aid of machine learning techniques to detect and predict faults from big data.