Multi-agent System Based Energy Management Strategies for Microgrid by using Renewable Energy Source and Load Forecasting

Abstract

This article focuses on multi-agent system based hierarchical energy management strategies for maximum economic and environmental benefits for microgrids. First, a two-level multi-agent based energy management system is constructed, which consists of an upper-level EMA in the view of whole system, multiple lower-level unit agents in a distributed manner, and their interactions based on communication. Second, in the upper-level agent, the energy management strategies are mainly designed by constructing multi-objective functions and by using a particle swarm optimization method based on hybrid probabilistic forecasting of renewable energy sources and loads. Third, in lower-level renewable energy source and load agents, the forecasting approach regarding renewable energy sources and loasd is mainly researched by means of the ensemble empirical mode decomposition combined with sparse Bayesian learning, called the hybrid probabilistic forecast approach. Moreover, in lower-level schedulable generation unit agents, local control strategies are also presented to regulate the output power to satisfy the reference power that is set by the upper-level agent. Finally, the validity of the proposed multi-agent based energy management strategies is demonstrated by means of simulation results.