Load-disaggregation of Household appliances using Deep Learning

Load Disaggregation is a broad term covering a range of techniques able to split a household's energy consumption by the individual appliances used.



This is a course project of EN 304 Energy engineering ,done in a group of 3 members under the guidance of Course Instructor and Project Mentor.
Course Intstructor:-Zakir Hussain Rather(Professor at IIT Bombay)
Project Mentor:-Dhiraj Khadka(PHD Student at IIT Bombay)

Project Tasks

1. Study in detail the load disaggregation and its importance.
2. Survey the existing techniques for load disaggregation in the literature.
3. Use your own household dataset or open public access dataset of appliance load consumption.
4. Implement a deep learning model for load disaggregation of various household appliances and compare results with other machine learning techniques.

Machine learning Models used:-

Improved Linear regression models

1.Ridge regression

2.Lasso regression

Support Vector Machine

3.Support vector regression

Nearest neighbour Regressor

4.KNeighborsRegressor

Ensemble models

5.Random Forest Regressor

6.Gradient Boosting Regressor

7.ExtraTrees Regressor

8.LGBM Regressor

Deep Learning Model Used

1. LSTM
2. Multi Layer Preceptron Regressor(Neural Networks)

Final Result

Deep learning model of MLP Regressor performed the best with a accuracy of 94.3%