基于CatBoost算法的配电网分区拓扑辨识

含多分布式电源配电网的拓扑结构具有多样性与多变性，影响拓扑辨识的实时性和准确性。提出一种基于CatBoost算法的配电网分区拓扑辨识方法。构建结合拓扑分区的配电网拓扑辨识框架，采用区域开关状态矩阵描述拓扑结构，以进行物理上的辨识降维；提出基于CatBoost算法的特征选择与拓扑辨识方法，通过分区并行离线训练得到历史拓扑和未知拓扑的区域拓扑辨识CatBoost模型，通过在线应用得到实时的区域开关状态矩阵标签，形成配电网开关状态矩阵，实现系统拓扑辨识。配电网算例系统测试结果验证了所提方法的有效性。

Partition topology identification of distribution network based on CatBoost algorithm

The topology structure of distribution network with multiple distributed generations has diversity and variability, which affects the real-time and accuracy of topology identification. A partition topology identification method of distribution network based on CatBoost algorithm is proposed. The topology identification framework of distribution network combined with topology partition is constructed, and the region switch state matrix is used to describe the topology structure for dimensionality reduction of physical identification. The feature selection and topology identification methods based on CatBoost algorithm are proposed, the regional topology identification CatBoost models of historical topology and unknown topology are obtained by partition parallel off-line training, the real-time regional switch state matrix labels are obtained by online application, and the switch state matrix of distribution network is formed to realize system topology identification. The effectiveness of the proposed method is verified by the test of distribution network examples.