THULab/praxis_crompton_greeves

Crompton Greeves Three Phase Pump Predictive Maintenance Dataset (TsFile)

Apache TsFile version of narteybrown/praxis-crompton-greeves.

Overview

Condition-monitoring telemetry for a Crompton Greaves pump unit: three-phase voltage/current, power, temperatures, pressures, flow, tri-axial vibration, motor speed/torque and derived power-quality indicators, sampled on a minute grid and labelled with a failure code.

• Tables: 2 TsFile table(s), 259,200 rows in total.
  • crompton_greeves_train: 207,360 rows, 2024-01-01 00:00:00 → 2024-01-03 23:59:00.047000 (UTC)
  • crompton_greeves_val: 51,840 rows, 2024-01-01 00:00:00 → 2024-01-03 23:59:00.011000 (UTC)

Schema (TsFile structure)

crompton_greeves_train.tsfile

• Time (INT64, milliseconds) — the source Timestamp column.
• TAG: none — the table is stored as a single device.
• FIELD (29): unit_id (string), cycle (int64), Phase1_Voltage (double), Phase2_Voltage (double), Phase3_Voltage (double), Phase1_Current (double), Phase2_Current (double), Phase3_Current (double), Active_Power (double), Reactive_Power (double), Power_Factor (double), Frequency (double), Stator_Temperature (double), Bearing_Temperature (double), Ambient_Temperature (double), Inlet_Pressure (double), Outlet_Pressure (double), Flow_Rate (double), Vibration_X (double), Vibration_Y (double), Vibration_Z (double), Motor_Speed (double), Torque (double), Insulation_Resistance (double), Oil_Level (double), Current_Unbalance (double), THD_Current (double), Bearing_FFT_Harmonics (double), Failure_Code (int64)

crompton_greeves_val.tsfile

• Time (INT64, milliseconds) — the source Timestamp column.
• TAG: none — the table is stored as a single device.
• FIELD (29): unit_id (string), cycle (int64), Phase1_Voltage (double), Phase2_Voltage (double), Phase3_Voltage (double), Phase1_Current (double), Phase2_Current (double), Phase3_Current (double), Active_Power (double), Reactive_Power (double), Power_Factor (double), Frequency (double), Stator_Temperature (double), Bearing_Temperature (double), Ambient_Temperature (double), Inlet_Pressure (double), Outlet_Pressure (double), Flow_Rate (double), Vibration_X (double), Vibration_Y (double), Vibration_Z (double), Motor_Speed (double), Torque (double), Insulation_Resistance (double), Oil_Level (double), Current_Unbalance (double), THD_Current (double), Bearing_FFT_Harmonics (double), Failure_Code (int64)

Conversion notes

• The source ships a train and a val split; each is converted into its own TsFile so the split membership is preserved.
• Time = the source Timestamp column (day-first DD-MM-YYYY HH:MM minute grid) parsed to INT64 milliseconds. The 48 repeated experiment cycles share each minute, so rows are ordered by a 1 ms step; the minute is unaffected.
• Single pump unit, no TAG: unit_id and cycle are ordinary FIELDs. No column is dropped.

Usage

Install the Apache TsFile Python SDK (pip install tsfile) and read a converted file:

from pathlib import Path
from tsfile import TsFileReader

path = Path("crompton_greeves_train.tsfile")
with TsFileReader(str(path)) as reader:
    schemas = reader.get_all_table_schemas()
    print("tables:", list(schemas))
    table_name = next(iter(schemas))
    table = schemas[table_name]
    columns = [column.get_column_name() for column in table.get_columns()]
    print("columns:", columns)
    field_names = [
        column.get_column_name()
        for column in table.get_columns()
        if column.get_column_name() not in {"Time", "time"}
    ]
    if field_names:
        with reader.query_table(table_name, field_names[:3], batch_size=1024) as result:
            batch = result.read_arrow_batch()
            if batch is not None:
                print(batch.to_pandas().head())

Source & license

• Original dataset: narteybrown/praxis-crompton-greeves
• Source card title / publisher: Crompton Greeves Three Phase Pump Predictive Maintenance Dataset
• License: other (as declared by the original dataset).