Improved K-means Algorithm for Construction of Adjustable Potential Resource Pool

As the large amount of renewable energy, energy storage system, flexible load and other multi-adjustable resource being accessed to the distribution network, it is necessary to study on the aggregation algorithm for adjustable potential of multi-adjustable resource in distribution network to make them fully utilized to satisfy the detailed requirements of big data in power system, and increase the operation efficiency of distribution network. The paper firstly put forward ADPC-kmeans(Automatically Density Peak Clustering-kmeans) algorithm based on K-means and DPC (Density Peak Clustering) aggregation algorithm. The paper then clusters and aggregates the adjustable resource based on a large quantity of history data. Next, the paper proposes a method to construct an adjustable potential resource pool in different time scales. The result of MATLAB simulation suggests that the algorithm and method proposed in the paper could clearly classify the type of adjustable resource, and makes all kinds of adjustable resource appropriately aggregate together and dispatch as well as adjust successively. The algorithm and method could provide skill support for new energy consumption, optimal scheduling of distribution network, demand side response, emergency cooperative control, ancillary services and other kinds of business.