Skip to content

A Cloud-Driven IoT home Energy Management System : Utilizes a suite of Google Cloud Platform offerings to carry out Non-Intrusive Load Monitoring in near realtime

Notifications You must be signed in to change notification settings

wintercameearly/home-energy-management-system

Repository files navigation

Cloud Based IoT Energy Management System

A Home Energy Management System built for the purpose of improving Home Energy Usage Efficiency.

This system was built using Cloud Technologies and systems to enable consistent, location-independent access to energy information.

LIVE DEMO : https://www.youtube.com/watch?v=67l0wjL0xx0&t=7s

Flowchart

Features

  1. Monitoring of Entire Home Energy Usage via Online Dashboard
  2. Control of Appliances Remotely
  3. Disaggregation and Monitoring/Recognition of Individual Appliances based on Gross Power reading
  4. Notifications on Energy Usage

Technologies Used

This work was written in Python.

It utilizes a Current Transformer Clamp and the RPICT3T1 for CURRENT measurement and a ZMPT101B Voltage Transformer with the ESP8266 for Voltage Measurement.

Both of these are connected to a Raspberry Pi over serial to compute into power and send almost Real-Time readings to Google Cloud Platform and adafruit.io

Adafruit.io Dashboard as seen below shows the readings ot the Entire Home Power Usage

Dashboard

GCP recieves readings from RaspberryPi through Cloud IoT core and Pub/Sub GCP hosts an RNN-LSTM Model on Cloud AI Platform, adapted from (https://github.com/GoogleCloudPlatform/professional-services/tree/master/examples/e2e-home-appliance-status-monitoring )to carry out Energy Disaggregation. Responses from the Model are sent to Adafruit.io via its HTTP API on a Feed (As seen in "Appliance Status" above) , and sent as Notifications to users via the Twillio API with the use of Google Cloud Functions.

Also added some features with IFTTT for Voice Control and a diluted form of Demand Response

Scripts

  1. communication.py : This is the script responsible for sending information to GCP and Adafruit.io. Please note there is some authentication and configuration to be done to get API keys, JWT authentication and the likes. It runs on the Raspberry Pi

  2. cloudfunctions : this folder holds the two cloud functions responsible for sending to adafruit.io(adafruit.py) and twillio(sms.py)

  3. ml : This folder hosts the Tensorflow Model, start scripts and yaml files for GCP AI Platform to enable deployment of the model.

  4. gatewayservice : This folder holds the gateway service responsible for calling the deployed Model endpoint, pushing responses to PubSub and storing data in BigQuery.

This work is possible very much due to the demo's, documentration and examples provided by Google. Check out (https://github.com/GoogleCloudPlatform/professional-services/tree/master/examples/e2e-home-appliance-status-monitoring) for help with deployment of the Model and setup of Cloud Enviroment. Very Grateful to them !. I will release a tutorial on how i achieved this soonest.

About

A Cloud-Driven IoT home Energy Management System : Utilizes a suite of Google Cloud Platform offerings to carry out Non-Intrusive Load Monitoring in near realtime

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published