Popular Mechanics (South Africa)
GOOD VIBRATIONS
Anya McGuirk (pictured right) is a research statistician at data and analytics company SAS. Based in North Carolina, she works in the IoT division and has just finished a study on vibration and how it can be used to monitor machines. She’s an avid beekeep
Popular Mechanics: What inspired you to use big data for bees?
Anya McGuirk: I was out inspecting one of my beehives and I opened it up and heard all this rumbling going on, and thought, ‘Oh geez, something’s up!’ I realised I needed to monitor the acoustics in my beehive more closely to assess what’s going on. I’m a beekeeper and, like so many others, I needed help, so this project naturally appealed to me.
PM: How did the project initially evolve?
AM: We started putting sensors on the beehives. Initially we placed a scale underneath the hives – that was pretty cool, but it required connecting your phone via Bluetooth while near the hives. Still, it enabled us to get readings every 15 minutes. I started collecting data, and then found some internal temperature-, humidity- and sound sensors, which we added in. We were now able to monitor weight, internal temperature, humidity and the acoustics.
PM: You researched the effect the queen’s presence has on the sound of the hive. Tell us about that.
AM: I wasn’t sure what we were going to discover, but I’d heard many experienced beekeepers say that you can tell when a hive is ‘queenless’ by the change in sound. So I made a split – where you take a big hive that has lots of bees, and remove a few frames of the brood and put it in a new hive. If there are already eggs there, the new hive will make a queen. But we wanted to see what was going to happen when the new hive figured out that it didn’t have a queen. Not knowing if we’d hear something, or notice if things would go crazy, we inserted the microphone and recorded it continuously for about 21 hours.
PM: What did you learn from the recording?
AM: The sensors are able to detect either queen piping (the sound she makes) following a swarm, or worker piping, which happens when a colony is queenless. This information can be hugely beneficial for beekeepers, warning them that a new queen may be emerging and giving them the opportunity to intervene before significant loss of life in the hive occurs, which happens when a queen bee dies.
PM: How did you isolate the hum of the hive from ambient noise?
AM: To ensure only the hum of the hive was being used to determine the bees’ health and happiness, we used robust principal components analysis, and it worked beautifully – it separated out all of the aeroplanes and sirens from the general hum of the hive, leaving us with distilled sound.
PM: Have you found that other beekeepers are interested in your work?
AM: They’re fascinated! They know you have to thermoregulate temperature, but when they realise you can actually monitor that without going into a hive, they get excited. With more companies selling products, sensors are becoming cheaper, making them more accessible to beekeepers.
PM: What’s next for your research?
AM: SAS has 49 hives on different campuses across the world, and I can’t wait to really let this take off at all of these sites, getting everyone to add sensors to their beehives. At SAS, we’re demonstrating how our algorithms are solving an important societal problem; we’re showing what we can do with beehives, but it’s applicable to many other industries.