Research and Application of User Profiling Methods Based on User Behavior Characteristics

In the context of the deepening development of smart grids, modeling the behavior of electricity users has become crucial for realizing personalized services and precise scheduling. This study proposes a multi-level user profiling method based on electricity behavior characteristics, integrating electricity behavior mapping and Graph-Embedded Spectral Clustering (GESC) technology to comprehensively characterize user electricity consumption patterns and behavior modes. Using smart meter data from 100,000 users collected between 2022 and 2023, 31-dimensional feature vectors are constructed, and behavior graphs are generated. The experimental results show that the GESC algorithm achieves a Silhouette Score of 0.72 and a Modularity Q of 0.59, with user profiling label accuracy reaching 91.4%, significantly outperforming traditional methods. In short-term load forecasting, the MAPE decreases from 19.7% to 7.8%, and the forecast accuracy improves to 92.5%. The study verifies the practicality and efficiency of the proposed method in user identification, prediction optimization, and strategy recommendation.