Abstract: The paper analyzes the virtual power plant's participation in the three-phase scheduling optimization process of the day-ahead, intraday, and real-time phases. Besides, it models the response speed and time of each part of the generalized energy storage, with which a virtual power plant with generalized energy storage three-stage scheduling model is established. In the day-ahead stage, using the C & CG to deal with wind and solar uncertainty. an effective and efficient electric vehicle charging and discharging management model is built to optimize the fixed electric thermal energy storage and the electric vehicle charging and discharging in the generalized energy storage, realizing a 24-hour overall optimization; In the intraday phase, adopting the mutual conversion between the alternative loads and the adjustment of the output of gas turbines and boilers to re-adjust the virtual power plant scheduling plan time by time; In the real-time stage, a time-decoupled air-conditioning cluster model is established, with whose rapid response characteristics and taking half an hour as the step length, the deviation is smoothed in the real-time scenery. The simulation example verifies the effectiveness of the purchased model. The results show that through the synergy of the multi-stage scheduling and the generalized energy storage, the operating costs of the virtual power plant can be effectively reduced and the energy supply stability of the virtual power plant can be greatly improved.