70% of electricity consumptions in the US comes from fossil fuels. Unfortunately, the electricity bill at the end of month does not explain where your energy usage is coming. So when a new roommate came to my house, our electricity bill went higher by 1000%. He blamed me for setting the thermostat temperature higher. After some investigation we found out it was his BitCoin and Doge Coin mining practices by several servers he used at home that caused that. We had no clue by the end of the month, How much energy spend air conditioning used? How much washer and dryer? Plus we want to schedule our (appliances) loads to run when the energy cost is the lowest because the cost of electricity is varying based on the time. Time of Use pricing has introduced another inspiration for our team to look for an interesting hack and solution.
What it does
Our device can disaggregate/detect appliance energy usage via analyzing the electrical signature of the appliances on a discretized FFT on the powerline carrier. It can also detects temperature which can help guide user change to adjust the set point if it is needed as most people don't have smart-thermostat.
How we built it
We used a Raspberry Pi in conjunction with a current transformer to read the current waveform. The waveform gets discretised and FFT algorithm extracts harmonics of the waveform. This unique FFT signature is fed to our backend server where it gets analyzed and it will get corresponded to the nearest item we have found in our database using a machine learning classification algorithm. This information then gets transmitted to our app where user can get informed how much energy different appliances have used.
Challenges we ran into
HARDWARE IS REALLY HARD...We have spent a lot of time on the hardware to make it work. It was a lot of frustrastrating trial and error. The waveform somehow was cut off many times and we have to go make the circuit from the ground up.
Accomplishments that we're proud of
We are proud that we could work on this amazing project and a have the demo.
What we learned
We learned that hardware is a challenging subject for the hackathon as debugging can be real cumbersome.
What's next for Green Seed
We hope that we can continue polishing this hack in future and make a product.