MIDAS: A Modular Ice Cream Factory Dataset on Anomalies in Sensors

Citation Author(s):
Tijana
Markovic
Miguel
Leon
Bjorn
Leander
Sasikumar
Punnekkat
Submitted by:
Miguel Leon Ortiz
Last updated:
Fri, 12/23/2022 - 06:24
DOI:
10.21227/wj1k-cy41
Data Format:
CSV
License:
Creative Commons Attribution
255 Views
Categories:
Machine Learning
Keywords:
Artificial Intelligence; Dataset; Machine Learning
0
0 ratings - Please login to submit your rating.
ACCESS DATASET CITE SHARE/EMBED

Abstract 

The dataset is generated from the ice-cream factory simulation environmen that is composed of six modules (Mixer, Pasteurizer, Homogenizer, Aeging Cooling, Dynamic Freezer, and Hardening). The values of analog sensors for level and temperature are modified using three anomaly injection options: freezing value, step change and ramp change. The dataset is composed of 1000 runs, out of which 258 were executed without anomalies.

Link to github: https://github.com/vujicictijana/MIDAS

 

Instructions: 

Files:

  • 1000 CSV files, one file for each run
  • file name contains the run id and type (Normal, Freeze, Ramp, or Step)

Dataset division:

  • Training: runs from 1 to 500
  • Validation: runs from 501 to 600
  • Testing: runs from 601 to 1000

Columns:

  • ordinal number of instance within one run
  • 13 parameters for Mixer module
  • 8 parameters for Pasteurizer module
  • 4 parameters for Homogenizer module
  • 7 parameters for AgeingCooling module
  • 16 parameter's for DynamicFreezing module
  • 6 parameters for Hardening module
  • Time stamp
  • Anomaly type
  • Sensor where the anomaly was injected
  • Actual sensor value

Classes:

  • Normal
  • Freeze
  • Step
  • Ramp

Instances:

  • Total: 36,124,859
  • Normal: 17,422,215 (49.67%)
  • Anomalies: 18,182,644 (50.33%)

Dataset Files

LOGIN TO ACCESS DATASET FILES

 

QUESTIONS?