Skip to content

Filtering

Behind the scenes sensor data is being processed and filtered to extract the best insights for estimating the condition of the machines.

The system primarily uses vibration data to analyze the condition of the machines. This is due to the fact that vibration data is the earliest data source in which it is possible to predict failures.

Work Orders

The analytics backend is connected with the maintenance app, and will use the work orders as feedback for optimizing the analysis. Currently, completed work orders of specific categories will be used by the analysis for a re-calibration.

The categories are:

  • Corrective Maintenance
  • Repair
  • Support Maintenance
  • Shutdown Repair Work

When the re-calibration is triggered, the machine will change to "Calibration" status until the analysis are ready to position the machine in a new health zone again. This will often take a few days if there's a constant stream of data from the sensors.

Custom Filters

Additionally, it is possible for users to customize the filtering of data that is being used for analysis.

The available filters are listed below.

  • Threshold (minimum and/or maximum)
  • Boxplot (filter out data not within 1st and 3rd quantile)

The filters can be used on any data source which is being ingested into the system for the particular machine.

Example

On machine XYZ we want to apply a filter that removes all RMS measurements below 0.5 mm/s and all RPM measurements outside of a boxplot where Q1 = 0.05 and Q3 = 0.95.