Forecasting Load-Shifting Flexibility of Electric Vehicle Charging Stations for Model Predictive Control

The growing number of electric vehicles (EVs) necessitates intelligent charging strategies, particularly in charging stations (CSs) integrated with photovoltaic (PV) systems and battery energy storage systems (BESS). Although load forecasting for CSs has been widely studied, the explicit consideration of charging load-shifting flexibility remains limited. This paper proposes a novel forecasting method that predicts future charging demand by modeling the aggregated flexibility potential of an entire CS. The approach combines three forecasts: a direct charging power profile assuming immediate charging at maximum power, a delayed charging profile representing the latest feasible charging time, and a forecast of station occupancy. These forecasts define a load-shifting envelope that can be integrated into a Model Predictive Control (MPC) framework. Simulation results based on German CS data demonstrate electricity cost reductions of 3.4 % to 4.3 % compared to scheduling strategies without
flexibility-aware forecasts.