Early insulation defects and hidden dangers in gas insulated switchgear (GIS) can be found by partial discharge (PD) detection of GIS, and then the insulation accidents can be prevented. In this paper, the complex wavelet transform (CWT) is used to process the ultra-high frequency partial discharge (UHF PD) signal in GIS at different scales. The trend curves of CWT energy entropy (CWT-EE) under different decomposition scales are analyzed, and it is found that the PD feature information mainly distributed in the scales, in which the gradient of CWT-EE are big. Besides, The CWT-EE characteristics and their scales are extracted to the structure characteristic pairs for PD type identification, which contained not only the PD signals energy feature information, but also the wavelet scale information of UHF PD signals. Finally, the support vector machine (SVM) method is used to classify four typical defects UHF PD signals in GIS. The recognition results show that the characteristic pair can effectively identify four typical defects in GIS and obviously reduce the decomposition scales of UHF PD and the feature dimension.