This study proposes gas insulated switchgear (GIS) partial discharge (PD) pattern classification based on the Hilbert-Huang transform (HHT). First, this study establishes four defect types of 15 kV GIS and uses a commercial high-frequency current transformer (HFCT) sensor to measure the electrical signals caused by the PD phenomenon. The HHT can represent instantaneous frequency components through empirical mode decomposition (EMD), and then transform into a 3D Hilbert energy spectrum. Thereafter, it extracts the energy feature parameters from the 3D Hilbert spectrum by using the back-propagation neural network (BPNN) for PD recognition. This study verifies the effectiveness of the proposed method by examining the identification ability of the BPNN using 160 sets of GIS-generated PD patterns. The experiment result shows the method can classify various defect types easily. The method can also be employed by the construction unit to verify the GIS quality and determine the GIS insulation status.