A Machine-learning Based Energy Management System for Microgrids with Distributed Energy Resources and Storage

An Energy Management System (EMS) for a microgrid system was developed to fulfill consumer load demand by maximizing distributed energy resource (DER) usage. A MATLAB model of different scenarios of the community was created to simulate the performance of the EMS algorithm. Extreme Learning Machine was used as the machine learning model for solar resource and load forecasting. A case study was conducted to validate the performance of the system. The integration of the EMS control resulted in an increase in the self-consumption of solar PV generation and a decrease in the grid dependence of the system compared to the scenario without EMS based on a system that uses a smaller battery capacity.