Microgrid Structure for testing a Real-Time Energy Management System Model

Abstract

Proper management of the tasks in a microgrid makes the energy management system successful. These tasks are based on analysis, control, and predictions in real-time, which makes the system capable of autonomous and guarantees its reliability and validity. In this paper, an experimental Microgrid testbed is proposed to allow emulating tasks in real-time that involve predicting energy consumption and generation in an emulated Microgrid cluster with an on-cloud energy management system. It involves machine learning methods in order to solve economic dispatch problems. Since the use of the supervised machine learning technique to estimate future consumption and generation in a system, helps the control system verify and modify itself. The Microgrid testbed includes system hardware in the loop interface, hardware power in the loop, and real inverters to have close emulation of the Microgrid cluster. A communication protocol is also implemented to connect to the on-cloud energy management system and the system hardware in the loop.