Optimal Scheduling Method of Virtual Power Plant Based on Model Predictive Control

In the environment of high proportion of new energy power system, one of the key points of virtual power plant dispatching is to consider the optimal control of distributed new energy prediction error. In this paper, a multi-time scale rolling optimization scheduling method considering the prediction error of distributed resources is proposed. Firstly, the long and short term neural network prediction method is used to predict the distributed photovoltaic output and the load rate of the communication base station, and then the bidding strategy is formulated on this basis. After the bidding, the decision sequence of each optimization variable within the day is constantly updated by the model based rolling prediction, and the particle swarm optimization algorithm is used to optimize the solution. Example analysis shows that the proposed method can reduce the dispatching cost by 21.47% compared with the theoretical optimal dispatching method, and is suitable for the optimization control of virtual power plant with a high proportion of uncontrollable power or load access.