Comparisons of digital filter, matched filter and wavelet transform in PD detection

C. Zhou, X. Zhou, B.G. Stewart, A. Nesbitt, D.M. Hepburn, D. Guo, M. Michel

Research output: Contribution to conferenceOther

4 Citations (Scopus)


Analysis of Partial Discharge (PD) measurement is often hampered by electrical noise. To remove this noise it is necessary to apply sophisticated digital signal processing techniques. In the past, various digital filtering techniques based on the Fast Fourier Transform (FFT) were applied. However, their results have not been up to expectation. More recently, the recent application of the matched filters and the Wavelet Transform (WT) have shown great promise, in terms of the effectiveness of denoising and enhancing the SNR (signal-noise ratio). This paper presents the authors' experience in applying three different types of techniques, i.e. matched filter, wavelet transform and the traditional digital filter, to PD data denoising in laboratory and industrial PD testing. The paper starts with an introduction to the background of the three methods. The second section of the paper presents the results of the application to, and the effects of the three methods on analysis of, PD data from laboratory experiments, practical PD testing on a 400/275kV power transformer using IEC60270 detection systems and PD testing on utility cables using high frequency current transformers (HFCTs). Finally the conclusions are presented with comparisons of the three methods in terms of effectiveness, computation time and their advantages and disadvantages. The results revealed that both the matched filter and WT techniques are superior to the traditional digital filters in PD detection and analysis, in terms of SNR and the number of PD pulses that can be discriminated from noise. The matched filter is the most effective when the waveform of the signal to be detected is perfectly known and when the only interference present is white noise. The more flexible and robust technique of WT can be applied to on-site testing, where severe and more complex noise interference is present. Another important advantage of the WT over the matched filter is that WT allows unambiguous reconstruction of the original pulse, which is extremely useful given that the distinctive shapes of PD pulses may be used to identify signal sources, for diagnosis or validation purpose.
Original languageUndefined/Unknown
Publication statusPublished - 2006
Event41st International Conference on Large High Voltage Electric Systems - Paris, France
Duration: 27 Aug 20061 Sep 2006


Conference41st International Conference on Large High Voltage Electric Systems
Abbreviated titleCIGRE 2006


  • de-noising
  • digital filtering techniques
  • Digital signal processing techniques
  • high frequency current transformers
  • laboratory experiments
  • partial discharge measurements
  • signal-noise ratio
  • unambiguous reconstruction, Fast Fourier transforms
  • matched filters
  • partial discharges
  • power transformers
  • signal detection
  • signal to noise ratio
  • wavelet transforms
  • white noise
  • digital filters

Cite this