Wind Power Scenario Reduction Based on Improved K-means Clustering and SBR Algorithm

The scenario method is important in the adaptation of the optimal dispatch of the power system with a high proportion of wind power. As a research hotspot of scenario analysis methods, the significance of scenario reduction is to describe a large number of complex scenario features with a small number of representative scenarios to achieve the purpose of reducing computational complexity. Aiming at the wind power output, a scenario reduction method based on the combination of the improved K-means clustering and the Simultaneous Backward Reduction (SBR) is proposed. Firstly, the original scenarios are quickly classified based on the improved K-means clustering algorithm. Secondly, the SBR algorithm considering Kantorovich distance is used to reduce the scenario sets in each cluster. Finally, an empirical analysis is carried out using the actual data from a certain province in Northwest China. The effectiveness and superiority of the proposed scenario reduction method are verified with the Brier Score (BS) indicator and the Gaussian mixture model of wind power fluctuations. © 2021, Power System Technology Press. All right reserved.