Mean Field Game based Energy Flow Management for Grid Integration of Large-scale Electric Vehicles*

Abstract

To integrate the distributed energy of electric vehicles (EVs) into power grid, handling the collective energy demand of the EVs is a key issue due to the uncertainties in mobility behavior of the individual vehicle. This paper addresses this problem by using mean-field limit approach to challenge this issue. The system targeted in this paper consists of an electric energy supplier and a parking lot that hosts a electric vehicle (EV) fleet with sufficiently large population. The energy integration between the system and the power grid is formulated as two problems: a day-ahead electric trading planning problem of the supplier, where the benefit maximization is considered under the constraint of charging/discharging demand of the parking lot, and a decentralized charging control problem is solved for the individual EV with consideration of the collective behavior of the whole EV fleet. To decouple the two problems, the mean-field limit approach is introduced to predict the charging/discharging demand of the EV fleet, and this also enables one to design the decentralized control strategy by solving a mean-field game problem.