A privacy-friendly hybrid data-driven algorithm for modeling the local flexibility of the EVs

The increasing penetration of electric vehicles (EVs), brings flexibility opportunities into the power system, because of the adjustable charging demand, as well as operational challenges. The charging energy of EVs is profoundly related with the user behavior uncertainties, which makes controlling the flexibility of each EV cumbersome. Therefore, to aggregate the flexibility potential of the EVs, local flexibility characterization is required to consider user comfort while satisfying user privacy. In this paper, a hybrid model is proposed that not only extracts charging sessions based on the raw energy consumption data but also deploys machine learning models to predict local flexibility characteristics for each local EV. Data for four real households are considered as the case studies to evaluate the model performance. The numerical results illustrate the achievements in presenting the local flexibility of the EVs, while the user privacy is given priority.