Research on a noise suppression method for composite pulse signals of partial discharge

Partial discharge (PD), serving as a critical indicator of insulation degradation in power equipment, is highly susceptible to interference and often submerged in background noise during practical detection, which significantly compromises the reliability of the detection results. This study proposes an innovative noise suppression technique specifically designed for PD pulse signals in power equipment. Overcoming the limitations of conventional denoising approaches, the proposed method adaptively extracts pulse components while accounting for the composite characteristics of PD signals, achieving a synergistic optimization in both the time and frequency domains. The method first employs a PD pulse extraction (PDPE) algorithm to accurately and comprehensively isolate all valid time-domain pulse components. Subsequently, a frequency-domain resolution enhancement (FDRE) strategy is applied to efficiently eliminate residual noise. In numerical simulations, the proposed PDPE–FDRE method achieved an output SNR of 14.827 dB, an NCC of 0.970, an MSE of 0.227, and a computation time of 2.974 s. Compared with the best-performing competing method, the proposed method improved the SNR by 3.414 dB, increased the NCC by 0.117, reduced the MSE by 46.46%, and shortened the computation time by 30.14%. Furthermore, its effectiveness is validated using experimental data collected from composite PD signals, confirming its practical applicability. The technique is compatible with various types of PD detection equipment, indicating strong adaptability across different systems. Beyond improving signal clarity, the method enhances the precision of PD signal characterization, offering a powerful analytical tool for insulation diagnostics and fault classification in power systems.