Machine-Learning enabled Condition Monitoring

Many manufacturing, warehouse and distribution facilities account for condition monitoring solution or predictive maintenance for the larger motors, drives and gear boxes but fail to take into consideration mission-critical equipment like rollers, bearings, conveyors, water pumps, etc. For the large part, the maintenance for these happens on a scheduled schedule. An unplanned conveyor motor failure in an assembly line, in a manufacturing plant, can cost anywhere from a few hours to several days of downtime. While unplanned maintenance is more disruptive and expensive than time-based scheduled maintenance, the best condition monitoring solution is one which predicts failure or presents the deterioration rate for any equipment that has the potential to cause downtime.

Bluvision’s Advanced Condition Monitoring uses machine learning and Artificial Intelligence (A.I.) to predict failure on any motor or mechanical equipment, weeks or months before the failure happens.

Predicting The Future Begins With History

The sensors in BEEKs Industrial BLE beacon, collects historical vibration data and establishes a motion fingerprint of each individual motor or mechanical device. These fingerprints are collected over a short period of time where the monitored asset is operated normally and is run through all possible stages of operation. (Eg: Motor off/motor at low speed/motor at high speed, etc.)


Recognizing The Patterns

Bluvision’s Condition Monitoring is based on multiple events and not just when a single anomaly is detected. While evaluating the new RMS and peak-to-peak values against the training stage, policies and alerts can be created for:


  • When the actual motion of a motor is different than its fingerprint. Alerts are created when the new value exceeds the modelled values.
  • When the motion level is growing – trend over time. Alerts are created when there is a trend line of exceeding values over time.

How We Do It Differently

Bluvision’s solution, apart from being equipment agnostic, also requires minimal hardware – sensor beacons to mount on the equipment and BluFi –WiFi gateways. Each gateway can manage over 200 sensor beacons concurrently. Our Bluzone cloud allows for user-defined alerts and for fleet management so users can check status and health of thousands of beacons at the same time.


At Bluvision, we study the motion in all 3-axis – x, y and z. More precisely, we use RMS (root mean square), peak and peak-to-peak, which provides the entire range of motion. The machine learning calculations are performed within the individual beacons with only the peak-to-peak data (Low speed @10 HZ and high speed @ 800 Hz) transmitted to the cloud, thereby saving battery and ensuring the user doesn’t have to go through tons of unnecessary data to analyze and detect the motion anomaly.


Editor BluvisionMachine-Learning enabled Condition Monitoring