Distributed Intelligence for Coordinated Electric Vehicle Control in Decentralized Microgrids

Abstract

The present study outlines a distributed control architecture for electric vehicle (EV) coordination in decentralized microgrids using a multi-agent framework where electric vehicles (EV) cooperate with decentralised energy resources through secure peer-to-peer communications and real-time execution of lightweight optimising and reinforcement-learning algorithms. By doing this, the agents will optimally balance approximate grid stability, user mobility requirements, and integration of renewable energy. Each agent will use local measurements and knowledge obtained by communicating with a limited number of neighbouring agents to create real-time charging, discharging, and local storage schedules. The coordination protocol developed through consensus-based techniques allows for system-wide objectives (voltage regulation, peak shaving, frequency support, and fairness) to be achieved while avoiding centralised control of any kind; only the coordination protocol is used for this purpose. Performance of the approach is assessed through realistic simulations of microgrid operation incorporating both solar generation, heterogeneous EV arrival/departure characteristics, and communication constraints. The distributed scheme produces similar grid performance when compared with centralised control; is substantially more effective than naive local heuristics when there are communication drops; increases overall utilisation of renewable resources whilst still respecting user preferences; and is capable of addressing and analysing scalability, privacy preservation, and resilience to component failure. Implementation considerations for future hardware-in-the-loop testing and field deployment will also be addressed, as will the future development of research regarding multi-microgrid coordination, stochastic market participation, and economic dispatch analysis.