Non-cooperative coupled microgrid-transportation coordination system: A deep deterministic policy gradient-based approach

Abstract

The cooperation between multiple microgrids (MGs) take a substantial role in enabling the process of information and energy exchange, which can increase the reliability of multi-microgrid (MMG) systems. Since MGs tend to be selfish in real-world operations, it is vital to address the way to motivate MGs to cooperate efficiently in MMG systems. In addition, considering the participated connected electric vehicles (CEVs) in CMT systems, the synergy between the MG-transportation coupled networks shall be accounted. In this paper, we propose a Stackelberg game theoretic approach for the non-cooperative coupled microgrid-transportation (CMT) system by exploiting connected electric vehicles (CEVs). This formulated game can jointly trigger the MG energy trading and vehicle-to-grid coordination schemes in the system. To solve the CMT game with imperfect information, a Markov Decision Process (MDP) and a deep deterministic policy gradient-based methods are developed. The devised algorithm can learn a deterministic optimal response from the opponents. The game’s Nash equilibrium can be approximately converged when the game is with imperfect information. Simulation results show that the proposed model can guarantee the effectiveness of the energy trading process in the CMT system. In addition, through the efficient CEV coordination strategies, the V2G regulation service is provided efficiently while it can further alleviate the grid power fluctuations. Furthermore, large economic benefits can be achieved under the CMT system.

Keywords

• Coupled microgrid-transportation system
• Connected electric vehicles
• Deep deterministic policy gradient
• Markov decision process
• Microgrid
• Non-cooperative system