基于有限关键节点及Wasserstein距离的配网拓扑识别

Distribution Network Topology Identification Based on Finite Key Nodes and Wasserstein Distance

明确配电网结构是配电网最优潮流、安全评估、网络重建、故障定位的基础。针对现有配电网拓扑识别方法缺乏结合现有网络结构参数和潮流信息,仅通过量测数据来进行拓扑识别效率低的问题,提出一种基于有限关键节点及Wasserstein距离的配电网拓扑识别方法。首先,利用子空间扰动模型证明配网拓扑变化时,可以利用有限的关键节点来进行拓扑识别,基于熵值法的混合K-Shell算法引入影响度概念,通过影响度与节点电气距离得出节点的重要度,确定配电网拓扑结构中的关键节点。其次,基于密度的噪声应用聚类算法通过电压、电流、有功、无功等4个特征来进行节点的聚类,将其他节点与关键节点进行类别归属,再结合Wasserstein距离得出节点间的连接关系从而得出配电网的拓扑结构。最后,通过IEEE 33节点算例和某小区实例,验证该方法的有效性。该方法极大地提高配电网拓扑识别效率与正确率,实现了配网拓扑结构的动态识别。

Defining the distribution network structure is the basis for optimal power flow,safety assessment,network reconstruction and fault location of the distribution network.Aiming at the problem that the existing distribution network topology identification methods are poor in efficiency due to their topology identification achieved only through measurement data without combination of existing network structure parameters and power flow information,a distribution network topology identification method based on finite key nodes and Wasserstein distance is proposed.Firstly,the finite key nodes can be used to identify the topology when the subspace perturbation model is used to prove the topology change of the distribution network,and the concept of influence degree is introduced through the entropy method based hybrid K-Shell algorithm and the importance of nodes is obtained by the influence degree and the electrical distance between the nodes,thus determining the key nodes in the distribution network topology.Secondly,the nodes are clustered with the density-based noise application clustering algorithm through four characteristics of voltage,current,active and reactive power,and other nodes and key nodes are classified into nodes and key nodes.And then,the connection relationship between nodes is obtained with the Wasserstein distance,consequently obtaining the topology of distribution network.Finally,a case study of the IEEE 33 node and a residential area has verified the effectiveness of the proposed method.This method greatly improves the identification efficiency and accuracy of distribution network topology,and realizes the dynamic identification of distribution network topology.