A high dimensional uncertain scenario generating method for wind power and photovoltaic considering spatiotemporal correlation

To address the challenge of accurately modeling the complex spatiotemporal correlations across multiple wind and solar power plant output scenarios, this paper proposes a high-dimensional uncertainty scenario generation method based on clustering partitioning. This method first divides multiple power stations into several clusters through cross-validation clustering and employs kernel density estimation to fit marginal distributions. It then proposes a high-dimensional C-vine copula structure construction mechanism based on spatiotemporal variograms to characterize spatial dependencies across clusters. Compared to traditional methods, fitting time is reduced by 23.87 %, and the VS metric decreases by 55.4 % in wind power scenario generation. Finally, a time series reconstruction mechanism based on wavelet denoising is proposed. The generated scenes with average DTW of 8.9 and standard deviation of 2.1, with the time characteristics consistent with historical data. Experiments based on the multi-energy complementary system in the lower Yalong River demonstrate that the method proposed for scene generation outperforms existing approaches in spatial correlation, temporal continuity, and scene reliability. It generates 300 consecutive scenes in just 23.53 s, significantly enhancing the efficiency and quality of high-dimensional scene generation.