Dynamic equivalence split-level semi-supervised spectral clustering algorithm based on wind farm

The present invention discloses a wind power generation of the wind farm dynamic equivalent spectral clustering based Semi-supervised hierarchical division. The method based on the measured wind speed in the wind farm of wind turbines of all data, the configuration may be embodied in a spatial structure of the original wind speed data and can provide more useful information for clustering the feature vector space. When the space division clustering groups of samples, using the a priori information of a small sample set, using cluster split top-down strategy, the sample group is divided semi-supervised clustering, cluster division result obtained WTG . For each wind turbine within the same group by equating a wind turbine, wind turbine equivalent parameters calculated by weighting capacity, thereby establishing a multi-machine model characterizing the wind farm dynamic equivalent. This method improves the clustering effect, the equivalent dynamic model of wind electric field created can more accurately reflect the dynamic response of wind farms.