Topology Identification Method for Low-Voltage Distribution Node Networks Based on Density Clustering Using Smart Meter Real-Time Measurement Data

Abstract

Correctly identifying the topology of a low-voltage distribution network aids in its management by power companies. However, the low collection rate and poor quality of the customer data collected by smart meters create difficulties in the application of data analysis algorithms. To address this, a new data-driven approach is proposed for topology identification in low-voltage distribution networks. First, a linear fitting method is used to approximate the noise-containing data. Next, the t-distributed stochastic neighbor embedding (t-SNE) method is used to reduce the dimensionality of the high-dimensional time-series voltage data. Then, the density-based spatial clustering of applications with noise (DBSCAN) method is applied to cluster the reduced dimensional data to perform topology identification. Finally, the simulated data and the actual data of a low-voltage distribution network in Dongguan, Guangdong Province are analyzed, and the results show the effectiveness and practicality of the proposed method.

Keywords

• Low-voltage distribution network
• topology identification
• noise-containing customer data
• t-distributed stochastic neighbor embedding (t-SNE)
• density clustering