Classification Method of Power Users Based on K-means and Support Vector Regression

According to the progress and improvement of the power bazaar, the import of demand-side has been re-recognized and users' willingness to actively participate in the grid interaction has become stronger. Effective research on power user classification is of great significance. Firstly, in view of the possible vacancy problem of load data acquisition, the cubic exponential smoothing (CES) technology is applied to complete the processing. Then, the K-means method is used to realize the load curve clustering, the contour coefficient method is used to determine the cluster category, and the hierarchical clustering method (HCM) is proposed to solve the cluster core, so as to improve the K-means clustering effect and reduce the clustering time. Support vector regression (SVR) is further used for training and learning to obtain the regression value of each input test data. Finally, the effectiveness of the suggested technology is tested based on tests of actual examples.