Refinement Generation Method of Renewable Energy Scenario Based on Information Maximizing Generative Adversarial Network

In the process of power system planning and operation, it is urgent to generate accurate and reliable renewable energy scenarios. To this end, this paper proposes a renewable energy scene refinement generation method based on infoGAN for the renewable energy generation uncertainty modeling problem. First of all, the historical output data and historical meteorological data of each renewable energy station are collected. The characteristic indicators of the output of each renewable energy station and the corresponding meteorological data are obtained as latent codes. Then, the information maximizing generative adversarial network is trained based on historical data. Finally, a numerical example of a provincial power grid in Northwest China is used to test the superiority of the proposed method. The results show that compared with the traditional probabilistic model-based scenario generation method, this method does not need accurate probabilistic modeling and complex scene sampling and can generate a scene set that is more consistent with the real scene data distribution law according to different scene preferences. The process is interpretable.