Appliance Prediction from Total Energy Data — A Demand Response Method Using Simple and Complex Networks

Abstract

The electric energy consumption of a particular appliance is predicted using different machine learning and deep learning algorithms from the disaggregated energy consumption data of a household. This can be applied as a part of Demand Side Management (DSM) to educate the customer to shift the appliance use during peak and off peak tariff rates. The different algorithms used in the study are Support Vector Regression (SVR), k-Nearest Neighbour (kNN), Decision Tree Regression (DTR), Fully Connected Neural Network and Long Short Term Memory (LSTM). Two different performance matrices used for evaluation are Root Mean Square Error(RMSE) and Mean Absolute Error (MAE).