At present, the contribution of partial discharge (PD) to power equipment insulation diagnosis is highly valued all over the world. However, most PD measurements are still carried out in the laboratory. The main reason for this is that the effects of on-field noise suppression are still limited and unsatisfactory. In recent years, among noise suppression methods, wavelet de-noising is the most common. The paper centers on improving the effects of on-field noise suppression. According to the shape features (pulse equivalent bandwidth, pulse duration time, and pulse polarity) can be analyzed and PD signals can then be discovered. This method is applied to on-field and laboratory experiments and the outcomes are discussed in terms of the effects on noise suppression. This method can obviously improve the effects of noise suppression.