Development of a Learning Algorithm for Energy Management in a DC Micro Grid

Abstract

Combination of distributed energy resources, energy storage devices and electric loads connected through the DC bus form a DC microgrid. Effective utilization of available energy through smart scheduling can decrease the burden on energy storage devices. Rapid growth in electric vehicle market necessitates the integration of solar photo voltaic (pv) generators into the electric vehicle charging stations (EVCS) since the DC power from solar pv experiences less conversion losses. Machine learning (ML) algorithms are getting widespread acceptances in recent years in power system engineering applications. Intermittent nature of solar energy and the dynamic nature of load profile at the EVCS make the environment suitable for the ML techniques in achieving optimal utilization of available solar energy. Reinforcement learning algorithms became popular among the researchers in the last decade especially for efficient battery scheduling and cost minimization applications. newline