A spatio-temporal bilevel charging-discharging scheduling strategy for urban electric vehicles based on functional zoning

Electric vehicles (EVs) serve as moveable loads, storing energy during off-peak hours and feeding it back to the grid during peak periods. However, the rapid growth of the EV industry leads to random and disordered charging-discharging behaviors, affecting grid stability and supply-demand balance. Thus, this paper proposes a spatio-temporal bilevel charging-discharging scheduling strategy for urban EVs based on functional zoning. First, the EV charging demand, geographic characteristics and node coupling degree within each functional region are analyzed. A functional zoning method based on reactive voltage sensitivity and modularity coefficients is proposed. Second, a spatio-temporal bilevel charging-discharging optimization scheduling model is established. The upper-level performs rolling prediction of EV charging-discharging time window length, whereas the lower-level provides spatial guidance of EV charging-discharging locations. The upper-level utilizes Long Short Term Memory (LSTM) to predict EV charging-discharging time, minimizing the voltage deviation. The obtained temporal features are imported into the rolling optimization model, adjusting the EV charging-discharging time window length. Then, the lower-level inputs the optimal results of the upper-level. Additionally, differentiated EV user behavior guidance strategies are established based on the regret theory and the dynamic charging-discharging cost model, achieving spatial optimal scheduling. Finally, a Modified Binary Gravity Search (MBGS) algorithm is developed, integrating the binary coding mechanism, adaptive gravitational constant and dynamic particle update scheme. Case studies on an IEEE-123 bus system demonstrate that the proposed methodology optimizes the spatio-temporal distribution of EV charging loads and enhances grid efficiency, promoting vehicle-to-grid interaction.