Virtual power plants integrate distributed power generations into a whole to feed power to the grid by utilizing advanced technologies such as intelligent measurement, network communication, and intelligent decision-making, which is expected to become a supporting framework for large-scale new energy generation integrating into grids. The goal of the virtual power plant constructed is to maximize the economic benefits of grid-connection. At the same time, considering the adverse effects of the uncertainty of new energy output on the power grid, fluctuation factor parameters and mechanism for reward or punishment are introduced to limit the fluctuation of grid-connection power. On this basis, the energy storage system and interruptible load are used as scheduling resources to consume the new energy power generation completely. Finally, the particle swarm algorithm is used for optimizing under the constraints. The calculation results show that the method can effectively suppress the fluctuation of the grid-connected power virtual power plant on the basis of consuming all the new energy generation, and greatly improve the friendliness of new energy access.