When the existing frequency response method is used for detecting transformer winding deformation faults

only the variation of amplitude-frequency characteristic signature is analyzed and processed

and the information of phase-frequency characteristic signature is neglected. In order to make up for the above shortcoming

we propose a method for diagnosing transformer winding deformation faults based on frequency response binary image. The basic principle of this method is introduced

and the diagnostic process of winding deformation fault is provided. The test platform of emulating transformer winding deformation is then established to perform 3 kinds of typical fault experiments

the amplitude-frequency characteristic and phase-frequency characteristic of measured frequency response are converted into polar plot

and the digital image processing technique is used to obtain the image correlation diagnostic indicator and criterion. Finally

the measurement cases of large power transformer are analyzed. The test result shows that the distribution of diagnostic indicators corresponding to the typical faults show classifying and clustering characteristics in a three-dimensional space

which can be used for winding fault identification. Besides

diagnostic criteria can be used to diagnose the winding faults of power transformers. The analysis conclusion indicates that the proposed method has an advantage over the common frequency response analysis method in aspect of diagnostic sensitivity.