Considering Topology Recognition for Pseudo-Measurement Generation in Low-Voltage Distribution Networks

• Published in: 2024 Second International Conference on Cyber-Energy Systems and Intelligent Energy (ICCSIE) pp. 1 - 6
• Main Authors: Wang, Biqing, Ju, Yuntao, Hu, Shiyao, Feng, Xichun, Jiang, Jing
• Format: Conference Proceeding
• Language: English
• Published: IEEE 17.05.2024
• Subjects: Accuracy; Distribution networks; Electric potential; Graph neural network; Graph neural networks; Low voltage; Low voltage distribution network; Network topology; Pseudo-measurement generation; Refining; Smart meters; State-estimation; Topology; Transformers
• DOI: 10.1109/ICCSIE61360.2024.10698001

To improve the precision of pseudo-measurement generation in low-voltage distribution networks, this paper considers the use of graph neural networks that utilize topological relationships. However, issues such as the unavailability of topology data in some areas and incorrect topology configurations impact the quality of the generated pseudo-measurement data. To address this challenge, this paper proposes a pseudo-measurement generation method that considers topology identification for low-voltage distribution networks. The proposed method leverages graph neural networks, which are capable of aggregating information and dependencies from key voltage and power measurements. Additionally, we integrate recurrent neural networks to capture the temporal dynamics of the distribution networks, thereby enhancing the model's ability to handle data that varies over time. The effectiveness of our method is demonstrated through simulations on a 14-node low-voltage distribution network, showcasing its potential in practical applications.