Study On Denoising Methods With Wavelet Packet And Pattern Recognition By Statistical Features For On-line Transformer Partial Discharge Monitoring

After analyzing the correlative technology of the on-line PD monitoring and pattern recognition for the power transformer, the author studies the propagation characteristics , the pick-up of the eigenvalue, the statistical way and the way of restraining disturbance using wavelet packets transform by the numbers. The main contents and the creationary outcomes of this paper are described as follows:(1) The transient models of transformer windings and bus-connected networks are set up. The author also studies the propagation of PD signals in the system. The influences of bus-connected networks, different monitoring points, low voltage windings and types of neutral point grounding on the frequency responses of PD signals are analyzed which give suggestions for the selection of monitoring band and monitoring points and PD localization.(2) Based on Wavelet Packet Decomposition(WPD), Entropy Threshold Method (ETM) has been put forward to reject the Discrete Spectrum Interference (DSI) in on-line PD monitoring. With the frequency division of WPD, ETM uses Shannon entropy as the criterion of determining whether or not DSI exist in certain WPD tree nodes and interference are suppressed successfully.(3) Lots of simulation data, lab data and on-site data have indicated that ETM works with good efficiency, without pre-knowing of DSI information, extracts the phase of PD pulses accurately and can calibrate quantity of single type discharge. We also study the influencing factors including sampling frequency, PD waveform, decline time, time intervals between PD pulses, mother wavelet, center frequency of interference, entropy threshold value and SNR. (4) The wavelet packets transform based on the restrict can restrain the white noise disturbance of PD. The precondition that this method can inspect the PD signal is . The key point that this method restrain the distortion of the white noise disturbance is the undee parameter.(5) The PD parameter extracting methods of the fingerprint charts were researched, and a new method was put forward that used Weibull model to analyze PD pulse amplitude distribution spectrum. Weibull parameter scale parameter and shape parameter were used as new character.After comparing the parameter estimate methods, this paper suggested that use the(6) least square methods to calculate the double-parameter Weibull distribution and use the optimized model to calculate the mixed Weibull distribution. Particular analysis was done according to the single PD resource and mixed PD resources, and Weibull parameters' distributions corresponding PD types were summarized.(7) PD pulse phase distribution spectrum was compared with spectrum, and 95% fiducial intervals of statistic operators were also summed up.