Intelligent Distributed Charging Control for Large Scale Electric Vehicles: A Multi-Cluster Mean Field Game Approach

Abstract

This paper presents an intelligent distributed optimal charging control for large-scale Electric Vehicles (LS-EVs) systems. The rising popularity of electric vehicles in smart cities is putting a strain on the power capacity, leading to high demand for energy and a surge in time-of-use (TOU) prices, which is becoming a pressing issue to address. When solving this issue for LS-EVs, the conventional cooperative centralized system encounters communication and computational complexity challenges. To address these issues, a distributed optimal charging control strategy has been developed. Specifically, the LS-EVs in a smart city has been divided into a finite number of clusters. Then, a mean field game (MFG) theory has been utilized for intra and inter clusters EVs to obtain the distributed optimal charging control. The MFG technique incorporates the state of charge (SOC) of all EVs from a particular cluster into a probability density function(PDF). Then, the PDFs of all clusters are employed to achieve the optimal charging decision for each EV in the city. In addition, a multi-actor-critic-mass (MACM) reinforcement learning algorithm is constructed to learn the optimal solution of the multi-cluster mean field game distributed charging control. Finally, to show the effectiveness of the presented schemes, numerical simulations have been performed.