LAUSR

During operation, power transformers are subjected to a variety of electrical and mechanical stresses, which can result in winding faults. The most common defects in transformers are axial displacement, radial deformation, and short circuits in the windings. It is critical to diagnose and detect faults at an early stage. The frequency response analysis (FRA) approach is well-known for detecting electrical and mechanical problems in transformers. Despite FRA’s accuracy, the interpretation of the produced frequency response traces (FRTs) is not yet standardized globally. Because the impact of faults on FRTs is uncertain, identifying efficient characteristics from such traces is required for interpreting transformer frequency response data. To ensure a reliable and better interpretation of FRTs, a novel hyperbolic fuzzy entropy (HFE) measure is introduced and employed as a clustering strategy in this article. In this work, the FRT is offered as a suitable frequency response not only to detect fault but also for fault discrimination. As a result, a two-stage method based on comparing the FRTs of healthy and problematic windings is introduced. The defect is recognized in the first step, and depending on the fault type, a suitable clustering technique is used in the second step. The maximum (minimum) HFE measure values of the normalized frequency response data are designated to the presence (absence) of transformer faults available in the suitable frequency band. It is revealed that the proposed HFE-based winding defects identification methodology has the necessary capability in providing fault information for a transformer and needs no expertise to detect and classify faults.