MetroPT2 — Train Compressor Predictive Maintenance Benchmark [7.1M instances]
Catalog Summary
"Real train-compressor sensor dataset from Porto metro with 7,116,940 time-series instances and 21 attributes. CSV files support anomaly detection, failure prediction, and RUL research."
Catalog Notes
Overview
MetroPT2 is a benchmark dataset for predictive maintenance collected from an urban metro public transportation service in Porto, Portugal.
It captures analogue sensor signals, digital signals, and GPS information from a train compressor system.
The data were logged at 1 Hz by an onboard embedded device from 2022-04-28 to 2022-07-28.
Column Schema
| Column | Description |
|---|---|
| timestamp | Time of the 1 Hz logged sensor observation. |
| pressure_signals | Analogue pressure readings from the air production unit. |
| motor_current | Current-consumption signal associated with compressor operation. |
| oil_temperature | Temperature signal related to compressor oil condition. |
| flowmeter | Flow measurement signal from the compressor system. |
| digital_control_signals | Discrete or control-state signals recorded from the equipment. |
| gps_latitude_longitude_speed | GPS location and speed information for operational context. |
| failure_event | Failure or event information used for failure prediction and RUL analysis. |
Key Statistics
- Total Records: 7,116,940 instances
- Features: 21 attributes
- File Format: Dataset files on Zenodo, commonly used as CSV time series
- File Size: Not summarized as a single file size in extracted metadata
- Time Period: 2022-04-28 to 2022-07-28
Use Cases
- Train compressor predictive maintenance
- Online anomaly detection
- Remaining useful life prediction
- Failure prediction with multivariate sensor streams
Source & Attribution
Created by Bruno Veloso, João Gama, Rita Ribeiro, and Pedro Pereira. Dataset DOI: 10.5281/zenodo.7766691. Described by Scientific Data article DOI 10.1038/s41597-022-01877-3.
View Data Structure
To explore column names, data types, and sample rows, visit the official dataset page on Zenodo.
Preview on ZenodoCite This Dataset
Veloso, B., Gama, J., Ribeiro, R., & Pereira, P. (2022). MetroPT2: A Benchmark dataset for predictive maintenance. [Dataset]. Zenodo. https://doi.org/10.5281/zenodo.7766691
Source metadata: Zenodo (2022) · DOI: 10.5281/zenodo.7766691
Indexed by IoTDataset.com on Jun 06, 2026
Review the Source Record
Confirm the licence, version, access conditions, file format, and provenance at the source before use.