Renewable Energy Scenario Generation Method Based on Order-Preserving Wasserstein Distance

With the increasing penetration rate of renewable energy generation, how to describe the uncertainty of renewable energy output is a key problem to overcome these challenges. Aiming at this problem, this paper proposes a day-ahead scenario generation method for renewable energy based on order-preserving Wasserstein distance. In this method, order-preserving Wasserstein distance is used as the loss function of discriminator to design a network structure suitable for day-ahead scenario generation. Through the game training of a conditional generative adversarial network (CGAN), the mapping between noise distribution and day-ahead scenario set under the prediction condition can be learned by the generator. In this paper, actual wind power data (including forecasting and actual generation data) are used to test the proposed method, and the results show that the proposed model can more accurately describe the day-ahead wind power uncertainty.