A Kernel-Based Predictive Model of EV Capacity for Distributed Voltage Control and Demand Response

Abstract

Energy storage and reactive power supplied by electric vehicles (EVs) through vehicle-to-grid (V2G) operation can be coordinated to provide voltage support, thus reducing the need of grid reinforcement and active power curtailment. Optimization and control approaches for V2G-enabled reactive power control should be robust to variations and offer a certain level of optimality by combining real-time control with an hours-ahead scheduling scheme. This paper introduces an optimization and control framework that can be used for charging batteries and managing available storage while using the remaining capacity of the chargers to generate reactive power and cooperatively perform voltage control. Stochastic distributed optimization of reactive power is realized by integrating a robust distributed sub-gradient method with conditional ensemble predictions of V2G capacity. Hence, the proposed solutions can meet system operational requirements for the upcoming hours by enabling instantaneous cooperation among distributed EVs.