Efficient partial discharge signal denoising method via adaptive variational modal decomposition for infrared detectors

•The proposed weighted kurtosis and threshold function can adapt to signals with different characteristics.•AVMD has no modal aliasing and endpoint effect for infrared detectors, and it does not require dataset training.•The proposed AVMD can achieve the best denoising performance compared with the existing methods. Partial discharge detection is one of the effective means for judging cable insulation condition. However, subject to the operating environment of the cable, the measured signal exists a large amount of noise from infrared detectors. In this paper, we proposed an adaptive variational modal decomposition (AVMD)-based partial discharge signal denoising method, which can reduce the noise disturbance significantly. First, variational modal decomposition (VMD) is optimized by the quantum genetic algorithm, and weighted kurtosis is presented as an optimization index. Furthermore, the wavelet thresholding method is utilized to process the reconstructed signal by VMD. To verify the superity of AVMD, Gaussian white noise and periodic narrowband interference are considered in the simulation. When the input signal-to-noise ratio (SNR) is −6.5202 dB, the SNR of AVMD can reach 8.3716 dB. Compared with the existing methods, AVMD has the best denoising performance. Finally, AVMD is used to process the measured signals. Experimental results demonstrate that it can effectively separate the signal and noise and ultimately retain the signal characteristics for infrared detectors.