Optimal charging scheduling for vehicle-to-grid regulation services

Abstract

Balancing power supply and demand in real time is critical for stable and reliable operation of power grids. Vehicle-to-grid (V2G) frequency regulation, which aims to utilize electric vehicles (EVs) to provide active power compensation for power grids, becomes a promising and economical solution for real-time power balance. To provide the V2G regulation services, most existing studies used single-level control structure and centralized approaches to coordinate the charging/discharging behaviors of EVs, which are not scalable and efficient when the population of EVs is large. To handle this issue, a generic hierarchical V2G system architecture should be designed, and scheduling algorithms are needed to coordinate the charging/discharging behaviors of EVs efficiently. In this thesis, we focus on the architectural design and scheduling approaches for the future V2G system with a large population of EVs.

First, we propose a hierarchical system architecture and a scheduling algorithm for a large-scale V2G system. The hierarchical V2G system applies a treelike control structure and employs smart V2G aggregators (SVAs) to facilitate the V2G control on EVs, which ensures the scalability of the V2G system. An online scheduling algorithm is devised to implement V2G regulation in the proposed hierarchical V2G system. The performance evaluation results show that the proposed algorithm can achieve the near-optimal performance of the V2G regulation while its computational time is much less than the traditional centralized algorithms.

Second, we further propose and study a hierarchical game-theoretic framework for providing V2G regulation, called V2G game, so as to motivate EVs to participate in the V2G system. We study two game-theoretic approaches to V2G regulation based on the nature of games. To study the competitive behaviors between the EV aggregator and EVs, a non-cooperative Stackelberg game approach is proposed. We also propose a cooperative approach based on the potential game model to consider the cooperation in the V2G system. We devise a smart pricing scheme in these games to motivate EVs to provide the V2G regulation services implicitly. The numerical results show the efficacy of the pricing scheme for providing V2G regulation in these games. They also indicate that, through cooperation, the social welfare of the system can be optimized and V2G regulation can yield the near-optimal performance.

Finally, we study a V2G game with imperfect information to address the privacy issue in the V2G operation. In the imperfect information game, each EV is unable to know any private information of other EVs. To find the Nash equilibria of the game, a reinforcement learning-based algorithm is devised for each EV. The numerical results show that the proposed algorithm can achieve the approximated Nash equilibrium in the game. This indicates that EVs can maximize their utilities and yield good performance of V2G regulation in the game, without exposing their private information.