With the extensive access of user-side distributed power generation, the resources that can be aggregated in virtual power plants are gradually transiting to prosumers. There are two main strategies for traditional virtual power plants to aggregate user-side resources：One is direct load control, the other one is to guide users to adjust their energy demand through load forecasting and then unilaterally setting the price of electricity. However, the first one cannot respect the user’s energy preference, while the second one cannot accurately predict the user behavior, so the price signal cannot accurately guide the user behavior. Moreover, the above strategies don’t consider the user’s power generation capacity. Aiming at the above problems, a stackelberg game model between the two sides is established, to set the purchase and sale prices inside the virtual power plant, guiding prosumers to adjust the power generation and consumption strategies, and finally the external output of virtual power plant is formed. The case analysis shows that both the virtual power plant and the internal producers and consumers can get considerable benefits through the model proposed. The case also shows that the change of the price of the spot market has a great influence on the determination of the internal market electricity price of the virtual power plant.