Regional wind and solar resources clustering methods considering spatial-temporal characteristics to accelerate production cost simulation

Abstract

Production cost simulation plays a crucial role in system planning, particularly for power systems with high renewable energy penetration. Balancing computational efficiency and accuracy remains a challenge. This paper introduces a regional division approach for wind and solar resources, utilizing spatiotemporal characteristics to significantly reduce model variables and constraints. Resources are first grouped into geographic grids based on latitude and longitude, with daily wind speed and solar irradiation averages capturing key renewable features. Annual feature vectors and regional correlations serve as clustering indices. The Davies-Bouldin criterion determines the optimal number of clusters, while K-means is employed for classification. A case study in Northwest China demonstrates that this method achieves substantial computational savings without compromising accuracy.