Research on User-transformer Relationship Recognition Method of Low Voltage Distribution Network Area Based on Multi Dimensional Feature Fusion

Abstract

The accuracy of low-voltage distribution network feeder area topology—comprising user-transformer relationships, branch connections, and phase attribution—is a fundamental prerequisite for refined distribution network management. To address the limitations of conventional approaches, such as low efficiency of manual inspection and poor robustness of single-quantity-based identification, this paper proposes a low-voltage feeder area user-transformer relationship identification method based on multi-dimensional feature fusion. By integrating five weighted indicators, including event correlation and power reconstruction, the method enables accurate user–transformer matching. In the case study, a 110 kV/10 kV/0.4 kV three-level distribution network simulation model was established in pandapower. A collaborative load simulation strategy combining "basic load + peak load + group event load" was designed, and nodal voltage time series were calculated considering line impedance characteristics. Using 15-minute interval measurement data for validation, the proposed method achieved an identification accuracy of 95.7%. The results demonstrate that the method requires no additional hardware investment and relies solely on smart meter measurements, achieving precise topology identification with strong anti-interference capability and engineering applicability. It can be directly applied to daily operation and maintenance tasks such as line loss calculation and three-phase imbalance mitigation.