Research on clustering algorithm of user electricity behavior for identification of typical should scene

With the construction of smart grid, advanced power equipment based on multiple types is increasing. More and more power data can be obtained by sensing devices and sent to edge devices and cloud computing servers through communication networks, supporting a variety of advanced applications. Among them, there are a lot of data-driven ARTIFICIAL intelligence computing analysis, are based on the power big data for computing analysis. However, for massive power big data, it is impossible to obtain more internal relevant information only by observing the data from the surface. To a certain extent, it directly affects the upper level of advanced applications. To solve this problem, this paper studies a clustering algorithm of electricity consumption behavior for typical scene identification based on DBSCAN density clustering algorithm. By clustering user load curve of DBSCAN, the classification of electricity behaviors is realized. AND cluster dispersion (CDI) and the ratio of sum of squares within clusters to differences between clusters (WCBCR) are selected as the indexes to evaluate the clustering quality. In addition, measurement noise can also be identified. The accuracy of the proposed clustering method is verified by experiments on actual user load data.