LAUSR

This paper proposes a novel fault interpretation method for transformer winding using features extracted from equivalent gradient areas of frequency responses. Firstly, frequency responses are pre-processed using binary morphology, aiming to eliminate the influence of stochastic factors such as noises and measurement configurations. Then pre-processed curves are divided into several frequency sub-bands, each of which reflects the influence of faults. Then fault features based on equivalent gradient areas are derived to quantify shape variations of frequency response curves with respect to specific fault types. Finally, real case studies are used to verify the capability of the proposed method, which can perform automated winding fault diagnosis.