Home Energy Management System for a Residential Building in Arctic Climate of Norway Using Non-Intrusive Load Monitoring and Deep Learning

Residential buildings can actively participate in energy management strategies by integrating advanced metering infrastructure to have a reliable and stable distribution system. This paper introduces a novel home energy management system (HEMS) framework to minimize total electricity costs by optimizing the charging of an electric vehicle (EV). The methodology incorporates a non-intrusive load monitoring algorithm to extract the EV information from the total power consumption. To ensure accuracy in planning, there is a need for an accurate prediction model for both total power and EV charging profiles. Therefore, deep learning methods are used together with signal processing techniques to accurately predict aggregated power consumption and decompose the EV from it. The effectiveness of the proposed HEMS algorithm is validated using a dataset collected from Northern Norway. The proposed methodology is validated by the test dataset. The implementation of HEMS resulted in a decrease of 15.4% in the total cost of electricity consumption and a significant reduction of 52.2% for charging the EV compared to the base case. Furthermore, the peak of total power is reduced by 23.19% with the adoption of HEMS.