Cable partial discharge period narrow-band interference denoising method based on Gaussian scale space

Abstract

The present invention discloses a cable partial discharge period narrow-band interference denoising method based on a Gaussian scale space. The method comprises the following steps of: collecting cable particle discharge signals by an oscillatory wave system, performing Fourier transform of the collected particle discharge signals, performing smooth filtering of the Gaussian scale space for the frequency spectrum, finding period narrow-band interference on the frequency spectrum and the particle maximum value of the particle discharge signals through multiple projection, extracting and eliminating the center frequency of the period narrow-band interference according to a kurtosis, and finally discharging the signals through IFFT reconstruction. The cable partial discharge period narrow-band interference denoising method is suitable for occasions of cable particle discharge detection.

Description

technical field [0001] The invention relates to the field of cable partial discharge detection, in particular to a Gaussian scale space-based method for denoising cable partial discharge periodic narrow-band interference. Background technique [0002] Partial discharge detection is widely used to evaluate the insulation status of cables, but due to the complex on-site environment, partial discharge signals are easily "submerged" in the interference signal. There are usually three types of noise in partial discharge detection, which are white noise, random pulse interference and periodic narrow-band interference. The periodic narrow-band interference is mostly high-order harmonics of the system and radio communication signals in the environment, which cannot be avoided during the discharge signal detection process. . Periodic narrow-band interference may have larger amplitude in time domain than partial discharge signal, and appear as narrow-band discrete spectral lines in f...

AI Technical Summary

Problems solved by technology

False components and end-point effects will be generated in the empirical mode decomposition filtering process; the notch filter causes a large loss of original partial discharge energy, and when the interference frequency band changes in the field or new interference spectral lines appear, the parameters of the notch filter are difficult to determine ; It is difficult to determine the optimal wavelet base and decomposition layers in the process of wavelet denoising

[0004] The FFT threshold filtering method is currently the most commonly used method to remove periodic narrow-band interference, but there are still two problems: one is that the periodic narrow-band interference whose spectrum amplitude is close to the discharge signal cannot be filtered out; value, only a part of the interference frequency components can be set to zero or attenuated, and the remaining interference spectrum components caused by spectrum leakage will generate large oscillation signals at the beginning and end of the time domain signal

Images