Abstract: It is blind that the signal process of actual partial discharge（PD） signals always is first denoised and then analyzed,and this paper proposed the idea that the signal-to-noise ratio（SNR） estimation of noise interference degree on PD source signals is first implemented and then the subsequent signal processing is carried on.2-order statistical theory was employed to construct autocorrelation matrix of PD source signals for the singular value decomposition,to build up the rules that reflect the energy of PD signal for signal subspace and noise subspace for the estimation of the evaluation parameter SNR to actual PD source signals.By the comparative analysis from simulative PD signals and actual PD signals,the noise interference degree on PD source signals can be well estimated and evaluated,which is the theory evidence for the subsequent denoise of PD source signals.