We will present the development of a bee control application that helps the beekeeper in the efficient production of honey and other products. The application covers the field of beekeeping. We see the motivation in helping the beekeeper to control the bees and be as effective as possible. This would make it easier for the beekeeper to work long hours on individual hives. The idea is that based on temperature and humidity, the app offers insight into the state of the bee family in a particular hive and detection of a special event called swarming. This is an event in which the bee family is divided into two parts. One part stays in the hive and the other leaves the hive and finds a new home. The first part remains in the hive and waits for the new queen to hatch, while the second half leaves the hive together with the old queen. Here it is important that the beekeeper takes timely action. He would be helped in this by a bee control application, which recognizes such an event on the basis of the sound processing of bee buzzing.
SolutionSince an individual beekeeper usually has large quantities of hives and consequently also many beehives, manual inspection of an individual hive requires a lot of time. With the help of our application, the beekeeper connects to individual hives via a mobile terminal and Bluetooth connection, where he can view the health of the bee family. In addition, the application will warn the beekeeper in cases of swarms that he will be able to take timely action and the bees will not escape into nature, which would reduce honey production.
DescriptionThe system consists of an Arduino Nano BLE 33 Sense microcontroller, which also contains a microphone on its printed circuit board. With the help of an algorithm running on a micro-controller, the controller listens to the buzzing of bees and, with the help of a learned model, recognizes the difference in the buzzing of bees, when the queen is present in the hive and when it is not. In addition to birth detection, the Arduino also includes a temperature and humidity sensor. With the help of this data we can determine the condition or the health of the bee family located in the hive. Since the goal was low energy consumption, the system measures the condition only a few times a day, or in the time periods between 10 a.m. and 1 p.m., where the chance of swarning is greatest. The rest of the day, the device is mostly idle and does not consume energy.
Machine learning modelDescription EdgeImpulse procedure
- Capture data using a microphone
First, we captured the buzzing of the bees using a microphone to collect data that formed the basis for a learning model.
- Spectral analysis of sounds
The data were then processed using a spectrogram.
- Building a model using a neural network
The spectrogram was the input to the neural network, which was used to train the model. After a long recalculation, we got the results, which were given in a matrix showing the recognition performance of the model.
The graph below shows how the model performs based on the data captured.
- Create a library and upload to the Arduino
Finally, we created a library to be uploaded to the Arduino board.
- Arduino Nano BLE 33 Sense
- Battery power
- Android mobile terminal
For receiving data from Arduino to application on Android phone we used bluetooth connectivity option. Arduino Nano BLE 33 Sense offers bluetooth module on his circuit board. This communication allows you to connect to the Arduino inside the hive and have distance from hive where is no risk of bee stings.
Next, we have designed the Android app we need to connect to the Arduino Nano BLE 33 Sense and start downloading data and alerts about the status of the bee family.
1. Connecting to the device in the hive
2. Main screen with temperature and humidity data and event alerts.
Below you can see alerts that Arduino device send to the Android aplication.
Instructions for testing our system.
Step 1Downloading the.ino program environment for programming the Arduino ble 33 sense. Compile code and send it to the Arduino board.
https://www.arduino.cc/en/software
Step 2Download the app to your Android device (.apk file in attachment)
Step 3Install Arduino device in the hive.
Step 4Connecting to the device with bluetooth connection
- Improving the machine learning model by increasing the database of bee buzzing.
- Add extra features to the Android app
- We see improvements in building a database of hive information on the LoraWan network, where data could be sent to a server and accessed anywhere, anytime.
We are happy to present our idea and share with you a project that you can try in your own environment. We believe that we are on the right track to making beekeeper's work easier with further improvements. You can also contribute to improving the model by increasing the database of bee buzz recordings. This will make the system more accurate and less sensitive to interference. Thank you!
Comments
Please log in or sign up to comment.