According to the characteristics that white noise follows 2 distribution in the S-transform domain and its average power spectrum is linearly correlated with the frequency，a partial discharge signal denoising method based on S-transform and singular value decomposition was proposed． Firstly，the time-frequency matrix of noisy partial discharge signals was obtained by S-transform． Secondly，the time and number of PD signal were determined by singular value decomposition and singular value difference spectrum theory，and the white noise area was obtained．Then the first denoising of PD signal was carried out by constructing the linear equation in proportion to frequency and dealing with the time-frequency matrix with hard threshold method． Finally，the second denoising of PD signal was done with singular value decomposition theory． Both the simulation and the measured signals show that the denoising effect of the proposed method is better than that of the wavelet threshold method，and the method is simpler and faster．