Abstract

Power transformer is a significant transmission and distribution equipment. Irreversible minor mechanical failures frequently occur in the transformer winding, which is difficult to detect in the early latent stage. Frequency response analysis (FRA) is widely used to detect transformer winding faults. However, the current interpretations of Frequency response analysis pay more attention to amplitude-frequency characteristics without phase-frequency characteristics, which is inaccurate in identifying minor flaws, detecting fault states, and quantifying fault severity. To fill these issues, this paper proposes a transformer winding fault diagnosis method based on the polar plot. The ratio of excitation-response is put forward for calculating the relative factor to detect the minor fault. And then, the fault states are diagnosed by the distribution of the polar plot, which combines amplitude-frequency and phase-frequency characteristics. Finally, the Minkowski distance of the frequency sampling points quantifies the extent of the failure. The experiments are implemented in a 10KV power transformer and COMSOL platform simulation. The fault status contains axial displacement, axial deformation, radial displacement, and radial deformation. The experimental results show that different states greatly influence the distribution of sampling points in polar plot. The proposed method effectively detects the minor flaw of transformer winding and performs well in judging the degree of fault severity and location.