Parameter-Adaptive VMD and Secondary Wavelet Filtering for Cable Partial Discharge Denoising

Among the various online monitoring methods for assessing the insulation condition of high-voltage cables, partial discharge (PD) detection is among the most widely applied. However, PD signals acquired in field environments are highly susceptible to noise interference. To address this issue, this paper proposes a novel denoising approach integrating parameter-adaptive Variational Mode Decomposition (VMD) and secondary wavelet filtering. First, the Rime Optimization Algorithm (RIME) is employed to adaptively search for the optimal decomposition parameters of VMD, specifically the mode number K and penalty factor α. The parameter-adaptive VMD then decomposes the noisy PD signal into a series of intrinsic mode functions (IMFs). Next, Composite Multiscale Fuzzy Dispersion Entropy (CMFDE) is introduced to quantify the complexity of each IMF, enabling the accurate distinction between effective signal components and noise-dominated modes. Finally, an Improved Wavelet Threshold (IWT) technique is applied to the selected effective components as a secondary filtering process, effectively suppressing residual noise while preserving waveform features. Simulation experiments evaluate the signal-to-noise ratio and reconstruction accuracy, while controlled cable measurements illustrate improved PD pulse visibility. At an input signal-to-noise ratio of −4 dB, the output signal-to-noise ratio reaches 12.40 dB with a normalized correlation coefficient of 0.972. These results support online insulation monitoring of high-voltage cables.