Keywords
low-voltage station are a; phase recognition; local linear embedding; Gaussian mixture model algorithm
Abstract
In order to solve the problem of inaccurate user phase recognition in low-voltage station areas, a phase recognition method based on local linear embedding (LLE) dimensionality reduction and Gaussian mixture model (GMM) clustering algorithm based on user voltage data is proposed. In this method, the voltage data in the smart meter of the user in the station area is extracted firstly, and the principal component analysis (PCA) method is used to achieve noise reduction and redundancy. Then, the derivative dynamic time warping (DDTW) method is employed to measure the correlation between user voltages. The LLE dimensionality reduction is used to extract the voltage data features of the user, and the GMM clustering algorithm is used to perform phase recognition on the user. Finally, the simulation is verified in the actual station area, and the accuracy is as high as 100%, indicating that the proposed method can effectively solve the phase recognition problem of the user in the station area.
DOI
10.19781/j.issn.1673-9140.2025.06.011
First Page
109
Last Page
121
Recommended Citation
HONG, Jiayao; XIA, Xiangyang; LEI, Yunfei; YI, Zejian; ZHU, Hanqin; and HU, Xiaozhong
(2026)
"Phase recognition of low-voltage distribution network based on locally linear embedding and Gaussia n mixture model algorithms,"
Journal of Electric Power Science and Technology: Vol. 40:
Iss.
6, Article 11.
DOI: 10.19781/j.issn.1673-9140.2025.06.011
Available at:
https://jepst.researchcommons.org/journal/vol40/iss6/11
