Data-Driven Charging Load Estimation of Behind-the-Meter V2G-Capable EVs
Mehrdad Ebrahimi, Mohammad Rastegar
Abstract
In this article, a data-driven methodology is proposed to estimate the net charging load of behind-the-meter vehicle-to-grid (V2G)-capable electric vehicles (EVs) using data mining techniques. The net charging load is the aggregated charging load minus the aggregated V2G output of EVs located behind the meter. The proposed methodology is composed of two stages: 1) finding candidate EVs and 2) estimating the net charging load of all EVs using measured data of candidate EVs. In the first stage, charging behaviors of customers are investigated and a small subset of EVs is introduced as representatives by clustering techniques. In the second stage, a mapping function is presented to estimate the net charging load of all EVs according to the charging behavior of representatives. The proposed method removes the cost of continuously monitoring, collecting, and archiving large-scale EV data. Applicability and effectiveness of the proposed method are validated by numerical studies using real data related to the charging behavior of EV owners.