The interactive grid is the basic model of the modern global power grids, analysis of users’ interaction power behavior is a core task. This paper firstly uses self-organizing map SOM neural network training and artificially separated methods to optimize the initial clustering center of K-means algorithm. Then, under the background of peak-to-valley time-of-use electricity price, the adjustment potential index based on user psychology is constructed, and the users’ electricity consumption behavior based on load data and adjustment potential index is analyzed. Finally, the clustering results of the two improved algorithms and the clustering results of classical K-means algorithm are compared. By comparing the advantages of K-means++ algorithm in accurate identification and clustering of users’ electricity consumption behavior, the effectiveness of K-means++ clustering center selection process in significantly shortening clustering time is analyzed.
Keywords electricity consumption behavior, cluster center optimization, load data, adjustment potential index, clustering analysis.