【目的】为挖掘电动汽车（electric vehicle，EV）集群可调度潜力，解决可再生能源出力不确定与用户行为异质性考虑不足的问题，提出一种考虑可调度潜力的电动汽车集群与充电站Stackelberg博弈调控策略。【方法】从时间、功率和能量三个维度构建集群可调度潜力模型，基于一致性原则建立聚合模型；在此基础上，构建充电站主导、EV集群跟随的双层博弈模型，上层采用两阶段鲁棒优化应对光伏出力不确定性，下层建立计及用户响应意愿和离站充电需求的最优响应模型；利用卡需-库恩-塔克条件与鲁棒分解实现等效求解。【结果】基于真实充电数据的仿真表明，与无集群聚合和无V2G场景相比，所提策略使充电站收益分别提升9.6%和10.5%，用户成本分别降低11.3%和29.1%。【结论】所提策略能有效释放EV集群规模化调节能力，实现充电站与用户双赢；两阶段鲁棒优化增强了光伏出力不确定性下的调度鲁棒性；V2G模式及用户行为反馈保障了调控可行性与用户满意度。

[Objective] To exploit the dispatchable potential of electric vehicle （EV） clusters and address the insufficient consideration of renewable energy output uncertainty and user behavior heterogeneity， a Stackelberg game-based regulation strategy considering dispatchable potential is proposed for EV clusters and charging stations. [Methods] A dispatchable potential model for EV clusters is constructed across three dimensions： time， power， and energy. An aggregation model is then established based on the consistency principle to reflect the true regulation boundaries. On this basis， a bi-level game model with the charging station as the leader and the EV cluster as the follower is formulated. The upper level adopts two-stage robust optimization to handle photovoltaic output uncertainty， while the lower level establishes an optimal response model incorporating user response willingness and off-station charging demand. The bi-level model is equivalently solved using Karush-Kuhn-Tucker （KKT） conditions and robust decomposition. [Results] Simulations based on real-world charging data demonstrate that， compared with scenarios without cluster aggregation and without vehicle-to-grid （V2G）， the proposed strategy increases the charging station’s revenue by 9.6% and 10.5%， and reduces user costs by 11.3% and 29.1%， respectively. [Conclusions] The proposed strategy can effectively unlock the large-scale regulation capability of EV clusters， achieving a win-win situation for both the charging station and users. Furthermore， two-stage robust optimization enhances scheduling robustness under PV output uncertainty， while the V2G mode， combined with user behavior feedback， ensures the feasibility of regulation and user satisfaction.