Here is another interesting example of the powerful applications of the SensorTile.box. We presented some time ago the domestic usage of the tool, while in this post an important industrial application has been described (full article HERE:(
Project purpose: In some climates, people use Sump Pumps where basement flooding can regularly happen or help solve dampness issues. Sump Pump's use a float switch to determine the level of water inside the basin, when this reaches a pre-set level, it'll trigger the Pump to remove all water to a second turn off point. Over time, these floats can fail or get stuck, which can cause the motor to run longer than expected, thus causing potential damage to the motor. Most current Monitoring systems rely on detecting water seepage in area's you wouldn't normally expect, aka flooding. With SensorTile.Box it is possible to monitor the system in another way: by watching vibrations directly from the Motor or Discharge Pipe, one can watch out for anomalies or any other changes to operation of the Sump Pump.
Implementation: STMicroelectronics' FP-AI-SENSING1 4.0.01 firmware has been flashed onto the ST SensorTile.Box. A modified version of STMicroelectronics' example_ble_1.py application has been used, patched to log 2400 samples (2 minutes of raw BLE data), and auto-connect to the sensor to work around the current bluepy limitations. Data is stored locally in date stamped log files and served up by Apache for external access.
Temporary conclusions: A consistent vibration pattern from the discharge pipe has been detected. At the time of the field tests, the Motor is consistently running around 8 seconds, every 2 to 3 minutes, so it's currently unable to replace it with a broken Sump Pump for comparison purposes. It will take more data samples before we are able to take advantage of STMicroelectronics' AI library. This library would allow us to detect Normal and Ab-Normal vibration patterns.
