Bee colony are threatened by predators, pesticides and lack of food, thus limiting the pollinisation of plants and reducing our harvest in honey.
The focus of this project is to monitor the health of your colony with sensors and send them to "The Things Network", allowing you to check all of the data on your colony in order to help you make decisions.
This project is an alternative of the "BEEP" project that help scientist and apicultor to do statistics on domestic bees in Europe for scientific purpose.
What is it :It is a prototype that measure different characteristic of the hive:
- Temperature inside the hive in 3 points
- Humidity inside the hive
- Temperature and humidity outside the hive
- frequency in the hives
- luminosity and battery level
- Presence of bees/hornet (not implemented but codes and advices are given below)
The DHT22 is not perfectly waterproof, which is why I would recommend you to put the one outside below the hive and the one inside should be in a corner to be protected from the bees.
My box before I was uploading my code on the card, and sealing it,
I was afraid that because of the movement the sensors would be disconnect so I put some cardboard at the bottom of the box to fix my card to it.
How does it work:The code is built around 4 steps: First connect to the lora notework by using the modem.begin from mkrwan.h library, then read the values from the sensors, after this stage we send the values measured as a byte vector to the lora Network and finally enter in sleep mode by cutting the power from the sensors with the power switch.
After reception of the byte vector on the things network we will decode the measure from each sensor using the payload formatter (see attached)..The decoded values will then be sent to the client's interface BEEP through a webhook.
The BEEP interface give a lot of insight on the fluctuation of the data in the time, while also giving some easy to read interface for real time data.
We are using a switch load to limit the current consumed by our captor while our card is on sleep mode (between two uplinks).
The wiring of the sensor is done according to the KiCAD schematics that you can find in "attachment".
With all of the optimization, we reach a total of 40 hours in theory, with measure at 20°C, we then put our prototype outdoor in a hive, it is self sufficient in energy with the solar panel, but would last 10 days without it (in temperature around 0°C).
Wiring:You should take the grove connector, from the DHT22 and solder it with the wires in the multistrand cable, try to respect the colors you chose at both sides of the cable.
For the multistrand cable with the 6 cables inside, cut the cables of the two DS18B20, solder them together (the two DS18B20 work with special protocol so they can share the same wire for data, power and ground, the other 3 cables will be soldered with another DHT22.
Try to respect the color you choose for the different cables (GND, POWER, DATA) on each part to ensure that your sensors work well (taking pictures of both end of the cable is a good idea ;-) ).
Do not forget to put heat-shrink tubes on all for the part you soldered and seal the wires coming out of your boxes in order to protect them from the rain.
IA and hornet detection:We trained an IA that detect hornet and count bees but didn't implement it for two reasons:
- Using an IA embedded on a board consume a lot of energy
- The communication between the board and the MKRWAN need a DC converter (3.3V to 5V) for the ESP32CAM which take up space on the PCB and consume more.
You will find in the attatchment section the trainned IA model or train your own with a no code solution using Edge Impulse, here are the performance of our trainned model (on a dataset from our hive):
It has trouble differentiating the bees from the background, which is due to the fact that when an Asian hornet comes close to the hive, the bees will all come to the landing board to protect the hive, making it difficult to label them
While it may seem problematic that 60% of the bees aren't detected, it isn't, as the number of bees is only useful when there is a swarm, to double-check if there really is a hornet.
Comments
Please log in or sign up to comment.