OSBH:acoustics: Unterschied zwischen den Versionen

Aus fablab Cottbus
Zur Navigation springen Zur Suche springen
 
Zeile 49: Zeile 49:
 
* Sample rate should be at least about 2.5 time the signal frequencie
 
* Sample rate should be at least about 2.5 time the signal frequencie
 
* With Analog Read it would be possible to analyse frequencies up to 3.2 kHZ (is that enough?)
 
* With Analog Read it would be possible to analyse frequencies up to 3.2 kHZ (is that enough?)
 +
 +
== FFT Frequency analysis ==
 +
* fight leakage of the signal with windowing functions (e.g. hamming)
 +
* time smearing is not that big problem because of the probably constant sound inside the hive
 +
** for detecting leaving and incoming bees it might be important to look at that issue again
  
  

Aktuelle Version vom 25. Januar 2015, 19:04 Uhr

This Page belongs to the Open Source Bee Hive Project


what for?

Acoustic analysis can be used to determine the health status and mood of the bees.

apidictor patent

frequencies to measure:

  • 180-190 Hz ("normal"): healthy bee sound in quiet period (night?)
  • ~220-290 Hz ( "warble"): if that sound occurs, queen's rate of egg laying has diminished.
  • >3000 Hz ("hiss") : sound occurs by clapping or knocking on the hive.
    • short sharp hiss: healthy queen
    • sluggish rise in hiss with slow fade: abnormality

interpretation of the sounds

warble hiss interpretion
yes sluggish rise and fall, max intensy much less than normal. "at time of issue it has vanished" swarming activity. possibly noticeable 15 days before swarming.
yes short sharp hiss age or deformity of the queen
yes on hiss vanishes completly death or removal of queen

So practical usage would be alerting if warble sounds occure. Also a LED signal responding to the hiss would be a quite nice on-field use and interaction: knocking on the hive: "everything alright?" - and the sensor answers with green (shor sharp hiss), red (sluggish hiss) or no light (nobody there!)..?

Acoustic sensor for beehive monitoring (US 8152590)

The patent says that it's possbile determine quality and quantity of air pollution and desease (mites, foulbrood) by analysing sound spectrum frequencies in bee hives. Quite complex statistical analysing required and basic data. Question would be, if every hive responds in the same way to pollution/desease, so that collected data could be compared with new positioned hives.

If that's possible the OSBH could set up a world wide pollution map with beehives! Also determinig bee's desease would be important data to analyse bee die off.


Arduino audio capabilities and sample Rates

  • Arduino can sample up to 8 kHz with analogRead.
  • bypassing AnalogRead gives about 38,5 kHZ
  • Sample rate should be at least about 2.5 time the signal frequencie
  • With Analog Read it would be possible to analyse frequencies up to 3.2 kHZ (is that enough?)

FFT Frequency analysis

  • fight leakage of the signal with windowing functions (e.g. hamming)
  • time smearing is not that big problem because of the probably constant sound inside the hive
    • for detecting leaving and incoming bees it might be important to look at that issue again


Links