Online condition assessment of traction transformer windings based on load-current wave transform

The non-stationary nature of electrical signals presents a major obstacle to online condition assessment for traction transformer windings under impact loads. In response, this paper presents a specialized signal processing technique: the Load-current Wave Transform. The method engineers a custom mother wavelet, the Load-current Wave, whose mathematical form is derived from impact load signatures. This dedicated wavelet converts one-dimensional current data into a 2D time-frequency representation, thereby revealing subtle variations indicative of early winding deformation. Validation on a fault simulation platform confirms the method's effectiveness in isolating fault-related features, offering a viable new strategy for monitoring transformer health.