Source-load scenario generation method based on conditional Wasserstein generative adversarial network with gradient penalty

The historical measurement data of renewable energy and loads can be processed in various ways to generate grid operation scenarios. However, with the continuous development of advanced forecasting technologies, a large amount of valuable prediction data accumulated during the scenario generation process is often overlooked. To address this issue, we propose a new scenario generation paradigm that fully considers the uncertainty of renewable energy generation, enabling multi-scenario generation for grid operations. This method models the uncertainty of renewable energy based on discrete probability distributions and employs the Wasserstein distance with gradient penalty as the objective function for the discriminator. It designs a network structure suitable for short-term source-load scenario generation. Through adversarial training of the conditional generative adversarial network, the generator gradually learns the mapping relationship between the noise distribution under forecasting conditions and the day-ahead scenario set. The proposed method is validated using IEEE 118-bus data, and the results demonstrate that the model can accurately capture the short-term uncertainty of renewable energy generation.