An online energy management system for AC/DC residential microgrids supported by non-intrusive load monitoring

Abstract Traditional electric energy systems are experiencing a major revolution and the main drivers of this revolution are green transition and digitalization. In this paper, an advanced system-level EMS is proposed for residential AC/DC microgrids (MGs) by taking advantage of the innovations offered by digitalization. The proposed EMS supports green transition as it is designed for an MG that includes renewable energy sources (RESs), batteries, and electric vehicles. In addition, the electricity consumption behaviors of residential users have been automatically extracted to create a more flexible MG. Deep learning-supported Non-intrusive load monitoring (NILM) algorithm is deployed to analyze and disaggregate the aggregated consumption signal of each household in the MG. A two-level EMS is designed that coordinates both households and MG components using optimization, forecasting, and NILM modules. The proposed system-level EMS has been tested in a laboratory environment in real-time. Experiments are performed considering different optimization periods and the effectiveness of the proposed EMS has been shown for different optimization horizons. Compared to a peak shaving strategy as a benchmark, the proposed EMS for 24-hour horizon provides a 12.36% reduction in the residential MG daily operation cost.