Real-Time Coordinated Operation of Electric Vehicle Fast Charging Stations With Energy Storage: An Efficient Spatiotemporal Decomposition Approach

Abstract

Fast charging stations (FCSs) have been widely adopted to meet the increasing charging demands of electric vehicles. The intermittent and impulsive nature of fast charging might significantly deteriorate the safe and efficient operation of the distribution power grid. Integrating battery energy storage systems (BES) in FCSs presents a promising option to mitigate these challenges. However, it is nontrivial to effectively coordinate multiple BES-equipped FCSs due to the highly stochastic charging demand and the spatio-temporal coupling nature of FCS operation. To address these challenges, this paper proposes a two-layer approach for real-time stochastic scheduling of multiple BES-equipped FCSs in a distribution grid. In the upper layer, we propose a computationally efficient dynamic programming method to determine the total power of all BESs at FCSs based on observed real-time fast charging loads and electricity price. Specifically, we derive analytical expressions for efficient offline training and online scheduling of the dynamic programming problem. This approach allows for direct training of value functions without iterative updating and obtaining scheduling decisions without redundant calculations. In the lower layer, we design a consensus-based power allocation strategy to coordinate power dispatch among individual FCSs following the reference power determined in the upper layer. In this way, real-time responses for each BES-equipped FCS can be given sequentially and distributedly. The superiority of the proposed method is validated via numerical simulations in comparison with state-of-the-art benchmarks.