Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Using In-Service Train Vibration for Detecting Railway Maintenance Needs

Irene Alisjahbana

TL;DR

This work investigates detecting railway maintenance needs (tamping and surfacing) from in-service train vibrations using the public DR-Train dataset. It shows that a simple k-NN classifier on signal-energy features derived from acceleration, especially from the transverse direction and tri-axial sensors, can achieve up to about 0.76 test accuracy for binary maintenance detection, and about 0.72 Exact Match in multi-label scenarios. The study highlights the benefits of applying ICP-based spatial alignment, PCA for dimensionality reduction, and multiple classification strategies, including problem-transformation and adaptive multi-label approaches. While promising for continuous monitoring, the approach currently relies on a single location and a limited set of maintenance types, with real-time deployment and broader generalization remaining as future work.

Abstract

The need for the maintenance of railway track systems have been increasing. Traditional methods that are currently being used are either inaccurate, labor and time intensive, or does not enable continuous monitoring of the system. As a result, in-service train vibrations have been shown to be a cheaper alternative for monitoring of railway track systems. In this paper, a method is proposed to detect different maintenance needs of railway track systems using a single pass of train direction. The DR-Train dataset that is publicly available was used. Results show that by using a simple classifier such as the k-nearest neighbor (k-NN) algorithm, the signal energy features of the acceleration data can achieve 76\% accuracy on two types of maintenance needs, tamping and surfacing. The results show that the transverse direction is able to more accurately detect maintenance needs, and triaxial accelerometer can give further information on the maintenance needs. Furthermore, this paper demonstrates the use of multi-label classification to detect multiple types of maintenance needs simultaneously. The results show multi-label classification performs only slightly worse than the simple binary classification (72\% accuracy) and that this can be a simple method that can easily be deployed in areas that have a history of many maintenance issues.

Using In-Service Train Vibration for Detecting Railway Maintenance Needs

TL;DR

This work investigates detecting railway maintenance needs (tamping and surfacing) from in-service train vibrations using the public DR-Train dataset. It shows that a simple k-NN classifier on signal-energy features derived from acceleration, especially from the transverse direction and tri-axial sensors, can achieve up to about 0.76 test accuracy for binary maintenance detection, and about 0.72 Exact Match in multi-label scenarios. The study highlights the benefits of applying ICP-based spatial alignment, PCA for dimensionality reduction, and multiple classification strategies, including problem-transformation and adaptive multi-label approaches. While promising for continuous monitoring, the approach currently relies on a single location and a limited set of maintenance types, with real-time deployment and broader generalization remaining as future work.

Abstract

The need for the maintenance of railway track systems have been increasing. Traditional methods that are currently being used are either inaccurate, labor and time intensive, or does not enable continuous monitoring of the system. As a result, in-service train vibrations have been shown to be a cheaper alternative for monitoring of railway track systems. In this paper, a method is proposed to detect different maintenance needs of railway track systems using a single pass of train direction. The DR-Train dataset that is publicly available was used. Results show that by using a simple classifier such as the k-nearest neighbor (k-NN) algorithm, the signal energy features of the acceleration data can achieve 76\% accuracy on two types of maintenance needs, tamping and surfacing. The results show that the transverse direction is able to more accurately detect maintenance needs, and triaxial accelerometer can give further information on the maintenance needs. Furthermore, this paper demonstrates the use of multi-label classification to detect multiple types of maintenance needs simultaneously. The results show multi-label classification performs only slightly worse than the simple binary classification (72\% accuracy) and that this can be a simple method that can easily be deployed in areas that have a history of many maintenance issues.
Paper Structure (19 sections, 3 equations, 11 figures, 6 tables)

This paper contains 19 sections, 3 equations, 11 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Dataset
  4. Dataset Description
  5. Location of Interest
  6. Maintenance Record Ground Truth
  7. Methods
  8. Pre-Processing
  9. Feature Extraction
  10. Machine Learning Algorithms
  11. Binary Classification
  12. Multi-Label Classification
  13. Experiments
  14. Evaluation Metrics
  15. Binary Classification
  16. ...and 4 more sections

Figures (11)

  • Figure 1: Sample raw acceleration data from sensor 1 of LRV 4313
  • Figure 2: Portion of track chosen for samples as compared to Google Maps location
  • Figure 3: Selection of passes with respect to ground truth maintenance work
  • Figure 4: The passes before alignment and after alignment using the iterative closest point (ICP) algorithm. Passes that have bad GPS data are removed.
  • Figure 5: PCA results of the explained variation vs number of components for the signal energy feature
  • ...and 6 more figures