The rapid charging demand of electric vehicles is required to be satisfied by supercharging stations. However, owing to the high-power and concentrated nature of charging, significant load peaks are liable to be generated over short-time intervals, whereby operational stability is adversely affected. Under time-of-use electricity pricing and grid capacity constraints, an intelligent scheduling approach integrating a hybrid genetic algorithm (HGA) with reinforcement learning (RL), denoted as HGA-RL, is developed, within which the supercharging station operator is defined as the decision-making entity. The charging sequence of users is globally optimised by means of HGA, whereby a rational basis for power allocation is established. Subsequently, charging power and scheduling intervals are dynamically regulated through RL, such that load shifting is achieved, with charging demand being increased during low-price periods and reduced during high-price periods.It is demonstrated by simulation results that charging load peaks are effectively mitigated and that the overall load profile is significantly smoothed. Meanwhile, the electricity procurement cost is reduced whilst user satisfaction is maintained, and the coordination efficiency between the supercharging station and the power grid is consequently improved.