Scenario Generation Method Considering the Uncertainty of Renewable Energy Generations Based on Generative Adversarial Networks

作者：Chao Huo, Cheng‐Yuan Peng, Zhiguo Hao, Xiuting Rong, Xuetao Dong, Songhao Yang · 年份：2024 · DOI：10.1109/iceeps62542.2024.10693230 · 被引用次数：4 · 研究领域：Smart Grid and Power Systems、Regional Development and Environment

The uncertainty of renewable energy output seriously affects the stable operation of the power system, and scenario generation is an effective method to characterize the output uncertainty. For the uncertainty scenarios of renewable energy output, it is necessary to consider both its stochastic properties and its actual statistical characteristics. So how to model the uncertainty in output is a key challenge for scenario generation. The deep learning-based scenario generation method can adaptively capture the high-dimensional nonlinear features of historical data and extract uncertainty information. Therefore, this paper establishes a generation model for renewable energy output based on the generative adversarial network (GAN). Through adversarial training, the model enables the generator to learn the distribution of the original data and generate extensive scenarios. Additionally, this paper employs statistical methods to evaluate the quality of the generated scenarios, including autocorrelation analysis and kernel density estimation. Utilize an improved K-means algorithm for scenario clustering to verify the uncertainty features of the generated scenarios. The calculation results based on the State Grid open dataset indicate that the established model can accurately describe the output characteristics of renewable energy.