Research on the Multi-User Behavior Modeling Methods Considering Demand Response

With the increasing integration of flexible resources, such as a high percentage of new energy sources connected to the grid, distributed power generation, and controllable loads, significant uncertainty arises regarding sources, loads, and grid operation trends. Consequently, the behavioral modeling of this new class of electric loads is essential. In this paper, a collection of meteorological statistical features closely related to multivariate user behavior is developed. The fuzzy C-means clustering algorithm is then applied to group the load features, resulting in clusters that align with user behavior patterns. To thoroughly capture the mapping relationships within these multivariate user behavior clusters, the XGBoost algorithm for integrated learning is introduced, enabling adaptive classification and prediction of multivariate user behavior, thereby enhancing the accuracy of the modeling process. A multivariate load dataset from a province in Southwest China is utilized, resulting in significant improvements in prediction accuracy and model generalization compared to traditional algorithms.