Smart Beehive monitoring systems

European beekeepers have been reporting declining bee populations and colony losses for the past 25 years, and the problem is becoming worse. According to the EU Reference Laboratory for Honey Bee Health (EURL), several EU countries lose up to a third of their colonies each year

The pollination service provided by bees is the most essential contribution they provide to agriculture. The direct value of honey produced in the EU is projected to be €140 million, while the value of insect pollination for European agriculture is projected to be around €20 billion per year, with a global value of €153 billion.

The decline of pollinators results in a considerable loss of pollination services, which has negative ecological and economic consequences for the conservation of wild plant diversity, large-scale ecosystem stability, and crop productivity, food security, and human welfare.

Climate change-related change in bumblebee species richness from a baseline (1901–1974) to a recent period (2000–2014).

This is due to the rise in climate change. We need to encourage and protect bees in order to rectify this issue and control the endangering of the bees.

“If the bee disappeared off the surface of the globe, man would only have four years left to live” Albert Einstein

• How do we protect bees?
• We can use visual inspection regularly by experts
• We should not remove bees near buildings and towers [A type of bee called rock bee builds its hive on buildings instead of trees]
• We should not use pesticides against these bees.

Let's see how we can use machine learning to protect the bees.

Our solution is to use SensiML and quick feather hardware to collect the audio data and use that information to see the buzzing levels and analyse the audio information. Analysis of buzzing bees audio data can be used to infer a lot of information such as the presence of the queen, missing queen, healthy bees etc.

A bit of history:

In 1964, Eddie Woods (1901 – 1976) – a BBC radio engineer – created a device called the Apidictor. Eddie had a keen ear and had discovered that honey bees produce different sounds depending on conditions within the hive.

For instance, if you knock once on the outside of a hive you should hear a short, sharp hiss/buzz. The shorter and more intense this reaction indicates how defensive the bees are and that indicates that the hive is queen-right.

The other phenomena he was interested in was a warble produced when brood was no longer hatching and/or the queen was eating less. The nurse bees produce this warble and he discovered it was an indicator for swarming.

Using the apidictor a beekeeper would regularly measure the sound level and as it changed could decide when a hive inspection was appropriate.

“Sound engineers are familiar with a phenomenon known as the ‘cocktail party effect’. This is the ability of the human brain, in a room full of chattering people, to pick out and concentrate on one conversation, not necessarily the loudest. Eddie was blessed with this ability and it served him well when listening to the medley of sounds that his microphone picked up in the hive.

One sound that caught his attention was a sort of warbling noise that varied between the notes A and C sharp; that’s 225 – 285 Hz in terms of frequency. He noticed that this sound got steadily louder, then it stopped and a day or so later a swarm took off

What are the advantages of knowing these things?

To create young bees and keep the queen warm enough to survive the winter, the hive requires an optimum climate. A device that detects oscillations in these measures can provide beekeepers with an early warning of hive concerns.

Step 1: Gather the things:

Gather all the required things like a quick feather, esp32, jumper cables, Battery enclosure etc.

Step 2: Getting started with Quick feather

Here's a wonderful blog is written by Arduino"having11"Guy on this topic please refer to that and get the basic things ready

Step 3: Get the data:

Getting the buzzing bee data is very difficult in the current scenario of the lockdown due to covid here in India. This is causing issues in transportation and other things. Most of the component orders did not reach in time and since I'm in my native I don't have access to many things.

Luckily after reading many papers and information on the internet I found a huge database where these audio datasets of bee are prepared with proper annotations.

Database Link: https://zenodo.org/record/1321278#.W2XswdJKjIU

Here I could find all the audio data of bees in different conditions such as active day, queen missing, queen working actively, queen piping or the bees It means the hive is replacing its queen bee with a new queen.

I downloaded the .wav files and uploaded them under a new project on the DCL and started segmenting them accordingly so that the sampling would be easy.

Prepare the data on the SensiML cloud and then create a model.

All these procedures are explained in the getting started guide so it is easy to go through the procedure.

I have some video samples which I wasn't able to convert to the wav file so I will record them through the quick feather device.

Step 4: Build the model

After gathering the data and labelling it, using prepare data we need to give a proper query name and other things then we can go to build model and start building the model

Since the data set was big it took really long time to build the model. But good things need time, here's the confusion matrix.

Step 5: Download the model

Once you are sure of the results download the model onto the device and verify with the recorded sound or the beehive itself.

Step 6: Testing and sending the data to the cloud:

I'll be using the.bin for the testing as I'll be using theESP32 for connecting the info to the cloud. ESP32 will be in deep sleep mode and will send the information to the cloud only in the intervals of 10 mins so that the battery comes for a long time.

Serial information is read by the ESP32 and JSON unpack is done and finally, the information is sent to the thingspeak cloud.

Power consumed by Quickfeather: 7mA

Power consumed by ESP32:6 x 75mA * 2.5 / 3600 = 0.313mAH

In deep sleep mode, esp32 consumes very little power also, the solar cell can be used to charge the battery continuously which will help us to get an extra life.

Demo video:

Conclusion:

Pollinators are required by three out of every four crops that produce fruits or seeds for human use around the world. Bees, as one of the world's most significant pollinators, are vital to food production, human livelihoods, and biodiversity. Unfortunately, bees and other pollinators are in decline in many regions of the world, with recent estimates estimating that their numbers may be down by as much as 30%.

We need to protect them at any cost.

References:

Frequency Analysis of Honey Bee Buzz for Automatic Recognition of Health Status: A Preliminary Study

To bee or not to bee: An annotated dataset for beehive sound recognition

Why do we need to protect bees?

Bee buzzing audio datasets

Noninvasive Beehive Monitoring through Acoustic Data Using SAS® Event Stream Processing and SAS® Viya®

Beekeeping in a nutshell