Dynamic Clustering Method for Distribution Networks with Massive Controllable Power Resources based on Improved K-means++

With the continuous growth of large number of the distributed controllable power resources in distribution networks, using traditional methods to directly participate in distribution network regulation will bring new challenges such as explosion of dimension in decision variables or difficulty in obtaining solution results, etc. Focused on increasing the regulation efficiency of massive controllable power resources in participating in active power optimal dispatching in distribution networks, this paper proposes an improved K-means++ clustering algorithm based on dynamic weight to the clustering of massive controllable power resources, and builds a dynamic clustering model for participating in the regulation of distribution networks with a high proportion of controllable power resources. Firstly, a resource clustering index system considering multiple factors is established; Secondly, through combining subjective and objective factors with real-time dispatching requirements of distribution networks, the dynamic weighting of index is realized; Finally, the improved K-means++ algorithm is used to realize the dynamic clustering of massive controllable power resources. The effectiveness and advantages of the proposed method are verified by simulation.