A Clustering Method of Electric Power Customers Based on DBSCAN Algorithm

Providing personalized services for electricity users with different needs has become an important part of deepening the reform of the electricity sales side market system, and achieving efficient and accurate electricity customer clustering is an important foundation for improving personalized services. In order to achieve efficient and accurate power customer clustering, a power customer clustering method based on DBSCAN algorithm is proposed in the paper. Firstly, the data in the power dataset is preprocessed and feature values are extracted from it. Secondly, a power customer clustering model based on the DBSCAN algorithm is constructed. Finally, the results obtained from the clustering model are visualized and analyzed. The experiment shows that when the parameter Eps of DBSCAN clustering algorithm is 0.006 and MinPts is 10, the overall clustering effect is good. At this time, the number of clusters is 5, and the profile of these 5 customer groups is obtained. This provides strong support for personalized power supply services for power enterprises and meeting the different needs of power customers.