A microwave, even in standby mode consumes about 8W of electricity every hour. Considering there are about 126 million households in the US and half of them have microwaves if we add up this power wastage then the total number comes to be pretty huge. If we could find a way to cut down this wastage then we can do a lot of good for the environment.

What it does

The project takes the live usage data and determines if the appliance is in use or it is in stand-by mode. Based on this information it predicts at what times the appliance is on and when it is not. Then it uses REST APIs encrypted with an algorithm to securely command the relay to turn the appliance on or off based on its predictions. As an incentive to the user to install this device, we plan to pay 1 cent for every minute worth of electricity he saves because of our project.

How we built it

We downloaded the appliance usage data from uk-dale for 5 houses. We picked one house and determined the appliance usage times. To do this, we played with several models like ARIMA, Neural Networks etc. In the end, we determined that LSTM gives us the best results. Based on the usage data, the AI model sends commands to ESP8266 which then controls the relay connected to the appliance. Based on the number of minutes that the appliance was turned off, the user gets paid accordingly using Softheon payment API.

Challenges we ran into

We couldn’t work or apply our project on real-time setup Due to shortage of time, we couldn’t evaluate many models in depth

Accomplishments that we're proud of

  1. Were able to determine the times at which the appliance is supposed to be switched on/off with reasonable accuracy,
  2. Encrypting the communication between the server and the ESP
  3. Developing the Android App which could control the ESP 4.Successfully integrated the Softheon's Payment API

What's next for SmartOff

Use live data to train our models. ESP8266 should be able to communicate with each other. Improve the packaging and the Android App

Share this project: