基于密度的改进K均值聚类算法在配网区块划分中的应用

在已知城市中压配电网的变电站位置、数量和容量的前提下,提出一种基于密度的改进K均值聚类算法,从初始聚类中心的选择和最佳聚类数K的确定两方面进行改进,并提出基于类间差异度和类内差异度的评价函数,对聚类结果的质量进行评估。将配电网划分为大小合适的配电网格,距离相近的变电站划分在同一网格内,每一网格独立供电,避免了距离过远的变电站之间的联络,为后续配电网络的优化规划提供了支撑。算例分析结果验证了该方法的有效性。

Application of improved K-means clustering algorithm based on density in distribution network block partitioning

Based on the position, number and capacity of the electric substations in the urban medium voltage distribution network, an improved K-means clustering algorithm based on density was proposed. The two aspects in the selection of the initial cluster centers and the optimal cluster number K were improved. And the evaluation function based on intra-cluster variation and inter-cluster variation was proposed to evaluate the quality of clustering results. The distribution network was divided into some suitable distribution grids. The substations that were close in distance were divided into the same grid, and each grid was independent of power-supplying, which avoided the contact between the substations that were too far away and provided support for the optimization of the network structure in the distribution network. The results of calculated example showed the effectiveness of the proposed method.