A Dynamic Clustering Method Considering User Behavior and Quantity Evolution

As one of the important means of “source-load interaction” in the new power system, load aggregation service can help to improve the regulation and control capability of massive loads by orderly clustering of massive user load resources. Existing user clustering methods are mostly based on clustering methods, which are overly dependent on the historical static characteristics of users, ignoring the continuous changes in the number of users in the future and the evolution of electricity consumption behavior, which is not conducive to the mining of the laws of user electricity consumption. To address this problem, this paper analyzes the adverse effects of the evolution of user behavior and quantity on user clustering results and modeling, and proposes a user dynamic clustering algorithm based on the evolutionary clustering, which ensures the consistency and reasonableness of the user clustering results in each time step. The validity of the proposed method is verified by using several real data sets, and the simulation analysis results show that the proposed algorithm can realize a reasonable, smooth and flexible user clustering method, and help to improve the accuracy of cluster load forecasting.