Power Prediction Model of Wind-solar Combined Power Generation Based on Cluster Scene Reduction Method

With the large-scale construction of wind power and photovoltaic units, the output prediction of such generating units has become a significant research trend, considering that the distribution of natural resources is relatively concentrated and geographically close to the wind power units and photovoltaic units have a great correlation. Therefore, in this paper, considering the ideal fitting degree, the Frank-Copula function is chosen as the combined probability distribution of wind and solar power, and the joint output scenario of wind and light considering the spatial correlation is generated. Based on this, the K-means clustering technique is enhanced to produce five large-scale output scenes, and the uncertain output curve of the scenery is obtained by combining the generation probability of different scenes with the output curve of the scene. Compared with the traditional method without cutting scenes, the power prediction curve obtained by this method reduces the influence of accidental scenes on the overall curve. It improves the rationality and accuracy of prediction.