| Supercharging stations , featuring high single-pile power and short charging duration, effectively satisfy the rapid energy replenishment demands of electric vehicles. However, their operation can induce large power peaks within short periods, exacerbating load fluctuations in urban distribution networks and threatening grid stability. Coordinated operation between SCSs and the grid is therefore critical, particularly under highly variable load conditions. To address this challenge, an intelligent scheduling approach combining a hybrid genetic algorithm with reinforcement learning is proposed. The method first employs HGA to globally optimize user charging sequences, establishing a rational basis for subsequent power allocation. Subsequently, the RL module dynamically adjusts charging power and timing according to real-time electricity prices, grid load levels, user state-of-charge , and remaining energy demand, increasing charging during low-price periods and reducing it during high-price periods. Simulation results demonstrate that this approach effectively mitigates peak charging loads, smooths grid load profiles, lowers total charging costs while maintaining user satisfaction, and significantly enhances SCS–grid coordination. The proposed strategy provides a practical solution for the safe and efficient operation of high-power charging infrastructure and grid load management, offering substantial application value and potential for wider deployment in urban power systems. |