Multi-layer Topology Identification for Low-Voltage Distribution Network Using Autoencoder and Dual-CURE Cluster

Topology identification of low-voltage distribution networks is crucial for ensuring the reliable, safe, and efficient operation of power systems. With the continuous evolution of the power network, the widespread integration of distributed energy sources such as photovoltaics, energy storage device...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:2024 IEEE International Conference on Energy Internet (ICEI) S. 411 - 418
Hauptverfasser: Liu, Hui, Liu, Jiawei, Ren, Shuaihui, Hong, Haisheng, Lin, Chuang, Lu, Jinwei
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 01.11.2024
Schlagworte:
Autoencoder
Autoencoders
Clustering algorithms
Clustering Using Representatives
Distribution networks
Feature extraction
Globally Aligned Kernel
Kernel
Low voltage
Low-Voltage Distribution Network
Multi-Layer Topology Recognition
Network topology
Reliability
Topology
Transformers
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Topology identification of low-voltage distribution networks is crucial for ensuring the reliable, safe, and efficient operation of power systems. With the continuous evolution of the power network, the widespread integration of distributed energy sources such as photovoltaics, energy storage devices, and electric vehicles has made the topology of the distribution network increasingly complex. Therefore, this article focuses on the identification of multi-layer topology relationships in low-voltage distribution networks and proposes an unsupervised topology identification method based on a fusion autoencoder and an improved Clustering Using Representatives (CURE) algorithm. Firstly, incorrect topology relationships are captured through autoencoders, and then dual CURE classifiers are used to achieve multi-layer recognition of user transformer topology relationships and user phase topology relationships, using Global Aligned Kernel (GAK) as the clustering metric. Compared to other typical methods, the extracted method can effectively capture anomalous relationships and identify multi-layer topological relationships on simulated datasets.
DOI:10.1109/ICEI63732.2024.10917208