The suppression of white noise interference is one of the key techniques of on-line monitoring of partial discharge. This paper proposes a de-noising method based on particle swarm optimization adaptive wavelet threshold estimation. The wavelet de-noising algorithm is based on an optimum and adaptive shrinkage scheme. When choosing the threshold, the generalized cross validation criterion is established and is used as fitness function. By using the particle swarm optimization algorithm, the optimum threshold of every decomposition scale is adaptively determined. The threshold selection method which does not rely on any prior knowledge is an adaptive method. The de-noising results of simulation signals and field PD signal show that compared with the standard threshold estimation method, the method proposed in this paper can remove the white noise in PD signals more effectively.