A household-transformer relationships and phase identification method of low-voltage distribution networks based on graph transformation and transfer learning

Abstract

A method based on graph transformation and transfer learning is proposed to further enhance the accuracy of household-transformer relationships and phase identification in low-voltage distribution networks. Firstly， a graph transformation method based on Gramian angular field （GAF） is introduced to convert electricity consumption data into a two-dimensional representation， facilitating the identification of differences in one-dimensional time-series electricity consumption data. Next， to address challenges such as sparse user data in low-voltage distribution networks， limited data acquisition methods， and a scarcity of samples， a deep learning model suitable for household-transformer relationship and phase identification is constructed using transfer learning and leveraging pre-trained parameter weights. Experimental validation demonstrates that the proposed model outperforms mainstream methods in both household-transformer relationship and phase identification， exhibiting improved accuracy and stability.

Keywords

• low-voltage distribution networks
• household-transformer relationship and phase recognition
• gramian angular summation field
• transfer learning
• deep learning model