On Demand Response Baseline Estimation for Public Fast Charging Stations

Attributed to its rapidly growing installed capacity and increasing charging power, public fast charging station (PFCS) exhibits outstanding potential for demand response (DR) programs. Thus, accurate DR baseline estimation for PFCS becomes critical to implement a reliable, economic and fair DR program. In this study, we propose an Informer-based neural network to estimate the DR baseline of PFCS, which incorporates the charging patterns of individual electric vehicles (EVs) at the lower level into the aggregate charging model of PFCS at the upper level. Leveraging the synchronized information during the DR period, we discern distinct EV models and subsequently establish new features of the Informer model. These features are combined with other factors influencing the overall load of the charging station as input features to estimate its baseline. The numerical cases based on an actual dataset demonstrate that our proposed method exhibits greater performance compared to conventional baseline calculation methods.