Research Of Periodic Narrow-band Noise Suppression Method In PD Signal

The key of on-line monitoring partial discharge(PD) in power equipment is toextract the partial discharge signals effectively. Due to the presence of a variety ofon-site equipment running interference signals, to extract partial discharge signalscaused great difficulties, of which the most serious is continuous periodic narrowbandinterference.Based on the study of existing partial discharge narrow-band interferencesuppression methods, this paper targeted proposed narrowband interferencesuppression methods effectively, and simulation and measured partial dischargesignal denoising research.1. In view of the empirical mode decomposition in the mode existing in partialdischarge narrow-band noise suppression of the aliasing phenomenon, this paperproposes a set of empirical mode decomposition and kernel independent componentanalysis method of periodic narrow-band noise suppression. In partial dischargesignals in the superposition of different frequency and amplitude of narrow-bandinterference conditions, this method can extract the discharge signal, effectivelyimprove the signal-to-noise ratio.2. The traditional method based on frequency band partition or basis function ofthe signal processing has its application limitations, in this paper, a periodicnarrow-band noise suppression method of reverse separation was put forward. Itfirstly estimates the frequency of periodic narrowband noise using the matrix pencilalgorithm. Then by constructing a virtual narrowband observed signal as an inputvector, separated by ICA to extract the partial discharge signals after narrowbandinterference suppression. Effectively suppress narrowband noise, while retaining thebetter PD pulses of high-frequency components and the low-frequency component ofthe discharge information. 3. The use of the characteristics is that there is no correlation between partialdischarge signals and the periodic narrowband interference and the concentratedenergy narrowband interference signal, and the partial discharge signal energy isscattered, and then put forward a kind of periodic narrowband noise suppressionmethod of subspace refactoring. This method does not need to predict the exact valueof the narrow-band noise frequency, which is suitable for practical application.4. Discussed a kind of partial discharge periodic narrowband noise suppressionresonance sparse decomposition method. Using narrowband noise signals and partialdischarge signals quality factor is different, on the basis of the resonance based onsignal sparse decomposition theory, achieve effective separation of narrow-bandnoise and discharge pulse, which has a good application prospect.