AI based energy management platform

AI based energy management ? [Software part]

Idea : The main idea is to come up with an intelligent way to reduce energy consumption in offices and everyday households.

How did we build it ? : The way we built the system is using Intel One API and a couple of data collection units that we developed.

We used the Intel one API to host a machine learning segment where we created a hourly model that processes an hour's data and creates a new model that will be used to generate optimal power values that will be used to control the Bulbs and fans in a room.

Basic Concept : The idea implements a power controller using machine learning and the end effector is a TRIAC based card that receivers these values serially and correspondingly controls 4 devices at a time.

How we are actually controlling the effective power going to the devices. The VRMS is different for different switching times. The more we delay the smaller is the value for the VRMS and hence lesser the power.

Hardware

The hardware I designed for this project is a BT-136 based TRIAC card which is a 2 layer PCB board with 4 TRIAC channels for 4 devices.

Also I designed a microcontroller base board to act as a bridge between the TRIAC card and the cloud uplink unit.

I have made all of these designs Open source and everyone is free to use them. They are available in my GitHub repo that I will be attaching to this project. It contains all the Gerber files for my designs.

Machine Learning model : The machine learning model is a Random Forest regressor that takes in the input feature vector and generates the optimal power value.