Research On Noise Recognition And Suppression In Partial Discharge On-line Monitoring For Transformer

On-line Partial Discharge(PD) monitoring for transformer is meaningful in precise interpretation of insulating condition and accident prevention .One of the most important problems in on-line PD monitoring for transformer is how to detect the weak stochastic PD pulses from the strong background noises. PD signal is a kind of non-stationary time-variable signal. Frequency Spectrum analysis technology is a traditional method for signal detecting and analysising. This method is effective when signal is stationary and spectrum is different from noise. But in fact, what we face with usually is nonstationary signal .It is necessary that every frequency component in every moment need to be analyzed. So the traditional method have some disadvantages and it is difficult to detect the PD signal from the interference effectively. The wavelet analysis method has simultaneous time-domain and frequency-domain resolution capability and it can be used to process the singular signal especially. To study the characteristics of PD pulses and interferences that may appear in the on-line monitoring for transformer and explore their time-frequency distribution more deeply, the wavelet analysis is applied and lots of literatures are referenced. Based On the different pulses and interference's finger-print map,the pulses and interferences simulation models are established. After analyzing materials and previous work,some disadvantages are found in processing the no calm wave signal of PD using the wavelet analysis method at the present time。A new adaptive algorithm based on wavelet analysis to suppress narrow bandwidth noise is proposed. The usual adaptive filter is one of the best algorithm in suppressing sinusoidal noises in PD signal processing. But it is difficult in setting the parameters of the adaptive filter in PD on-line monitoring due to wide frequency range of narrow bandwidth noise, and it may be unstable sometimes when the parameters set improperly. Study shows that the new algorithm has better performance and stability compared to usual adaptive filters. When a denoised process is performed on a signal with wavelet transform modulus maximum principle,how to reconstruct a satisfactory signal from the remained modulus maxima is an important subject. In this paper,an analysis is made on the relationship between the modulus maxima and wavelet coefficients. Then the fact that modulus maxima are actually discrete samples of wavelet cofficients in a specific sense is obtained. By preprocessing the modulus maxima we get a new set of pseudo modulus maximum sequence with which a new piecewise cubic spline interpolating algorithm to reconstruct the wavelet coefficients is presented.Compared with the alternate projection method,this algorithm is simple and easy to implement and can get higher reconstruction signalSNR gain and smaller RMSE than the altermate projection method,so it is a practical and efficient algorithm. The processing results of numerical simulation demonstrate that the proposed method has tremendous potential for extracting partial discharge signals from the noisy background, thus providing reliable signals, on the basis of which, further study can be explored.