At present, the contribution of partial discharge (PD) to power equipment insulation diagnosis is highly valued globally. However, most PD measurements are still performed in the laboratory, the main reason being that the effects of on-field noise suppression are still limited and unsatisfactory. To date, among noise suppression methods, wavelet de-noising is the most common. This study focuses on improving the effects of on-field noise suppression. The shape features (pulse equivalent bandwidth and pulse duration time) can be analyzed, and PD signals can consequently be discovered. This study applied the proposed method to on-field and laboratory experiments, and presents a discussion of the outcomes regarding the effects on noise suppression. This method can improve the effects of noise suppression.