Faults in the windings are one of the major sources of transformer failures. The present work is aimed at developing a simple Sweep Frequency Response (SFR) based methodology for detecting, quantifying and locating the short circuit faults in transformer windings using statistical parameter interpretations. By imposing short circuit faults along a 22 kV continuous disc model winding, the faulty SFRs are measured. Using healthy and faulty SFRs, Min–Max ratio (MM) characteristics across the winding are derived. As the analysis with the entire frequency spectrum is not useful in locating faults, the sliding window technique is employed to identify the fault sensitive sub-band. The first resonance frequency based sub-band is found to be successful in locating faults. The MM characteristics are found to be parabolic for different percentages of faults and are used in locating the fault with respect to the centre of the winding. In practice, as the measurement of SFRs is not possible across the entire winding, parabolic derivative method is proposed to predict the MM characteristics with only two SFRs measured at any existing tappings. The proposed methodology is checked for interleaved and layer model windings and also on a 33 kV winding of a practical transformer.

绕组中的故障是变压器故障的主要来源之一。本工作旨在开发一种基于简单扫频响应（SFR）的方法，用于使用统计参数解释来检测，量化和定位变压器绕组中的短路故障。通过沿22 kV连续圆盘模型绕组施加短路故障，可以测量出故障的SFR。使用正常的和有故障的SFR，可以得出整个绕组的最小-最大比率（MM）特性。由于对整个频谱的分析在故障定位中无用，因此采用滑动窗口技术来识别故障敏感子带。发现基于第一共振频率的子带在定位故障中是成功的。发现MM特性对于不同百分比的故障是抛物线形的，并用于相对于绕组中心定位故障。在实践中，由于不可能在整个绕组上测量SFR，因此提出了抛物线导数法来预测在任何现有抽头处仅测量两个SFR的MM特性。检查所提出的方法是否适用于交错和分层模型绕组，以及是否适用于实际变压器的33 kV绕组。