Machine learning based demand response scheme for IoT enabled PV integrated smart building

Abstract

The short-term forecasting of electric power consumption and renewable energy generation with high efficiency and advanced demand side management is essential for improving the energy flexibility provided by the smart buildings. However, the recent literature fails to present the mechanism for developing demand side management for energy flexibility. This article addresses the importance of developing an electric power consumption and renewable energy generation short-term forecasting model using a supervised machine learning approach. In this work, random forest and decision tree regression models are developed and used to forecast future electric power consumption and renewable energy generation. To validate the forecasting model performance using the MAE, RMSE, MAPE, and R2. This article proposes a day ahead dynamic pricing model for the demand response scheme. The day ahead dynamic pricing minimizes the peak time demand and end-user electricity tariff. A smart grid system must know the future short-term electric power consumption and renewable energy generation to enable the demand response scheme. The educational institute smart substation and weather station are integrated into the internet of things devices to enable real-time monitoring and control of the electric power consumption and renewable energy generation. The results show that the proposed DR scheme benefits end electricity users and smart grid operators.