Research on Dynamic Clustering Strategies of Flexible Resources in Building Complexes for Multi-Interaction Scenarios

A dynamic stratified classification clustering strategy is proposed to improve the precision and efficiency of clustering flexible resources in building complexes for multi-interaction scenarios, including frequency regulation, peak shaving, voltage control, congestion management, and economic dispatch. First, the regulation characteristics of flexible resources are analyzed to establish a stratified classification method. A rolling dynamic clustering process based on hierarchical clustering is then introduced, utilizing a sliding time window to update clusters every 5 min over a 15 min horizon. To meet the diverse requirements of different interaction scenarios, a dynamic weighting strategy for resource characteristic parameters is developed. An improved median distance method is also proposed, using the cluster center and maximum dispersion point as similarity measures between clusters, thereby reducing computational complexity while maintaining accuracy. A case study demonstrates that the proposed strategy enhances the silhouette coefficient and clustering quality, offering an effective clustering scheme for regulating flexible resources in building complexes under multi-interaction scenarios.