现代电子技术

在低压配电网中，台区的拓扑关系对于配电网运行和管理至关重要。由于台区结构复杂、数据量庞大，且受到用户数据相关性高及噪声的影响，准确地辨识台区之间的拓扑关系具有一定挑战性。为此，提出一种基于稀疏自编码器降维和改进BIRCH（Balanced Iterative Reducing and Clustering using Hierarchies）聚类算法的拓扑关系辨识方法。首先通过对稀疏自编码器进行训练，获取数据重要的线性及非线性特征，从而对原数据进行降维；然后，利用改进BIRCH聚类算法对降维后的数据进行聚类，辨识出低压配电网中台区之间的拓扑关系。通过算例结果验证以及与其他方法进行对比，证明该方法较之传统方法能够更高效地辨识低压配电网台区的拓扑关系，为后续数据分析和决策支持提供了有益的工具。

In low-voltage distribution networks， the topological relationship of the substations is crucial for the operation and management of the distribution network. Due to the complex structure of the substations， large data volume， and the influence of high user data correlation and noise， accurately identifying the topological relationship between substations is challenging. A topology identification method based on sparse autoencoder dimensionality reduction and an improved BIRCH (balanced iterative reducing and clustering using hierarchies) clustering algorithm is proposed. The sparse autoencoder is trained to obtain the important linear and nonlinear features of the data， thereby reducing the dimensionality of the original data. Then， the improved BIRCH clustering algorithm is used to cluster the dimensionality-reduced data to identify the topological relationships between substations in the low-voltage distribution network. The method is verified by example results and compared with other methods. It can identify the topology of substations in the low-voltage distribution network more efficiently than traditional methods， providing a beneficial tool for subsequent data analysis and decision support.