Data underlying the publication: A Bidirectional Long Short Term Memory Approach for Infrastructure Health Monitoring Using On-board Vibration Response
DOI: 10.4121/64600497-7ba5-4831-9c94-1b643f600987
Dataset
Categories
Licence CC BY 4.0
The dataset is an Axle-Box Acceleration simulation dataset based on the vehicle-track interaction model developed by Dr. Chen Shen, as described in the following reference. The rail and sleepers are meshed using Timoshenko beam elements, while the ballast and railpads are represented as discrete spring-damper pairs.
Clamps and bolts are not explicitly modeled; instead, their stiffness is incorporated into the railpad stiffness, a widely accepted simplification in railway track modeling. The wheel is simplified as a rigid mass, and the wheel-rail contact is modeled using a Hertzian spring. ABA measurements are simulated at an operational speed of 65 km/h by considering floating stiffness reduction. This is the raw dataset and additive white Gaussian noise scenarios are generated as the paper explains. For details on the dataset, please refer to the cited papers.
If you use this dataset, ensure proper citation of the following references.
*Shen, C., P. Zhang, R. Dollevoet, A. Zoeteman, and Z. Li, Evaluating Railway Track Stiffness Using20 Axle Box Accelerations: A Digital Twin Approach. Mechanical Systems and Signal Processing, vol.21 204, 2023, p. 110730.22
*R. R. Samani, A. Nunez, and B. De Schutter. A Bidirectional Long Short Term Memory Approach for Infrastructure Health Monitoring Using On-board Vibration response. Dec. 3, 2024. doi: 10 . 48550 / arXiv.2412.02643. arXiv: 2412.02643 [cs]. Pre-published.
History
- 2025-02-28 first online, published, posted
Publisher
4TU.ResearchDataFormat
.npyReferences
Organizations
TU Delft Faculty of Mechanical Engineering, Delft Center for Systems and ControlDATA
Files (3)
- 1,676 bytesMD5:
e1e8a1cce88f444b502ac6804c99c561
README.txt - 1,039,200,128 bytesMD5:
0be5e2ca9389298df4d9ae4461bba5e8
Input.npy - 2,400,128 bytesMD5:
b932562bc959c126a5177cc9a3086d55
Output_labels.npy -
download all files (zip)
1,041,601,932 bytes unzipped