A Method for Diagnosing Transformer Faults under Online Voltage Signals

Abstract

Transformer degradation is often accompanied by changes in electrical characteristics. Such changes may originate from insulation defects, winding abnormalities, or local dielectric damage. Detecting these variations at an early stage remains a challenge for condition monitoring. In this work, primary-side and secondary-side power-frequency voltages were analyzed through U₁–U₂ Lissajous trajectories. Ellipse parameters, including the major axis, minor axis, and inclination angle, were extracted from the trajectories and used for comparison under different operating conditions. A transformer equivalent circuit was established, and impedance variations were introduced to represent abnormal states. Trajectory geometry changed continuously as the equivalent parameters varied. Among the extracted features, the inclination angle showed the largest variation. Noticeable differences were already present when the impedance deviation reached ±3%. The obtained results indicate that voltage trajectories contain information related to internal parameter changes and may be used for condition evaluation during operation. Since only terminal voltage measurements are required, the method can be implemented without additional sensing devices.