Abstract:
This research aimed to analyze the energy consumption of a smart home in order to develop the energy consumption forecasting models for reducing energy consumption in the home. In this study, Python software was used as a tool for data analysis and a home energy management algorithm was designed for demand response analysis.
The data analysis in this study was performed in accordance with the energy consumption data of a smart home. In addition, data mining was used to develop the models for forecasting loads. Autoregressive Integrated Moving Average (ARIMA) model and Recurrent Neural Network (RNN) model were applied as prediction models in order to compare and determine the optimum average errors. The electrical loads of the smart home were determined using the demand response algorithm for reducing energy consumption of the house.
The data analysis and the prediction of energy consumption of the smart home were analyzed according to the mentioned models. The prediction of hourly, daily and weekly energy consumption using the ARIMA model showed the mean square error of 0.160, 0.076, and 0.179, respectively. On the other hand, the RNN model showed a mean square error of 0.349. This indicated that the ARIMA model provided the optimum average daily error. According to the simulation using energy management algorithm, the peak demand decreased from 12.83 kW to 8.31 kW. By using this algorithm, the loads were turned off according to the specified priorities.