Partitioning Active Distribution Networks by Using Spectral Clustering

With the increasing intermittent renewable sources integrated into active distribution networks (ADNs), distributed management should be adopted to improve the control and management level of ADNs effectively. Accurate partitioning is a vital step to manage distributed resources. A partitioning method of ADNs by using spectral clustering algorithm is proposed in this paper to avoid the curse of dimensionality and accurately obtain the partitioning scheme of the ADNs. First, the graph model of the ADN is established to simplify the distribution network structure. Then the normalized symmetric Laplacian matrix of the distribution network is calculated according to the graph model. Finally, the eigenvectors of the normalized Laplacian matrix are solved and dimensionally reduced by using spectral clustering. The k-means algorithm is used to cluster the newly generated node data to obtain the partitions of the ADN. Case studies on the IEEE 13-node test feeder and IOWA 240-node distribution test system demonstrate the feasibility, adaptability, and efficacy of the proposed partitioning method.