An effective distributed approach based machine learning for energy negotiation in networked microgrids

In recent years, the definition of the distributed energy management frameworks has become the core of research due to its distinguished advantages including the less time-consuming, more accurate and secure than the centralized frameworks. In this sense, this work proposes an effective energy management framework for a distributed networked microgrid (DNM) based on the primal-dual method of multipliers (PDMM). The networked microgrid is comprised of a wind turbine (WT), battery unit and photovoltaic (PV) system which supply the demands of the networked microgrid. The power negotiation process among the networked microgrid' segments will continue until obtaining an adequate solution to achieve the best performance of the proposed microgrid. A Reinforcement Learning (RL) approach is proposed that can increase the accuracy of modeling uncertainty of the parameters of the DNM compared to other manners. To make the secured communication platform, a secured energy transaction structure based on the directed acyclic graph (DAG) approach, is proposed as one of the other aims of this paper. The results indicate the effectiveness of the proposed model and its applicability for the networked microgrid.