Abstract: In order to extract characteristics of partial discharge signals, the paper proposed an adaptive threshold algorithm for processing partial discharge data by using ensemble empirical mode decomposition method for analysis of XLPE cable partial discharge signals. Firstly, partial discharge signals containing noise are decomposed into a number of independent mode function components by ensemble empirical mode decomposition method, and then adaptive threshold processing and reconstruction of the resulting mode function component were carried out to suppress noise. The algorithm overcomes mode mixing problem of empirical mode decomposition method, and simulation and measuring data verify its correctness and feasibility.