Multi-Agent System-Based Plug-and-Play Energy Management System for DC Microgrids

This paper discusses the design and implementation of a decentralized multi-agent based plug-and-play capability for an energy management system (EMS) of a DC microgrid. For the EMS, a consensus algorithm is used with a fully distributed optimization approach. The multi-agent system (MAS) is designed using the Java Agent Development Framework (JADE), taking advantage of its features and properties to implement the plug-and-plug capability. Agent negotiation abilities are implemented to enable the agents to react according to time-varying communication topologies in the microgrid. An interface based on TCP/IP is implemented between Matlab and JADE to create a bi-directional communication link so that signals are sent from Matlab to the microgrid agents to trigger certain events in the system. The EMS algorithm, with the help of the implemented plug-and-play capability, exhibits robustness and converges very fast. The results are validated through simulation.