Abstract: Different types of partial discharge (PD) in gas insulated switchgear (GIS) cause different damage to

GIS insulation. Correctly identifying the type of partial discharge is very important to evaluate the insulation status

of GIS. In order to simplify the process of feature extraction and improve the recognition rate of PD types，the deep

forest algorithm was introduced into GIS PD pattern recognition，and a deep forest model for PD pattern recognition was constructed. 252 kV GIS PD detection experiment platform was set up and a typical defect model was designed，and the partial discharge of four kinds of typical insulation defect models was detected using the ultra-high frequency method; the collected discharge waveforms were normalized by graying and bilinear interpolation，which were used as input of the deep forest model; the multi-grained scanning structure was used to extract the adaptive features of the PD gray-scale image to avoid the subjective influence of feature selection; the cascade forest structure was used as a classifier to classify the types of PD. The recognition results show that the comprehensive recognition rate of this method is as high as 99%，which can effectively identify the PD type of GIS.

      Key words: gas insulated switchgear (GIS)；partial discharge（PD）；pattern recognition；deep forest