Differentiating between partial discharge sources using envelope comparison of ultra-high-frequency signals

T. Pinpart, M.D. Judd

Research output: Contribution to journalArticlepeer-review

59 Citations (Scopus)

Abstract

Partial discharges (PD) in high-voltage equipment cause damage to insulation. One method for detecting them is the ultra-high-frequency (UHF) method. Subsequent analysis using phase-resolved patterns or time-of-flight measurements can identify PD type or locate the defect. This study presents a method for comparing UHF signal envelopes to differentiate between multiple PD sources. Experimental investigations are conducted to identify an approach that strikes a compromise between competing aims of minimising signal sampling rates while retaining sufficient ability to discriminate between PD sources in different locations. The envelope comparison method is efficient and has potential for incorporation within automated systems. Results clearly demonstrate the ability to distinguish between PD sources in different positions, which offers a means to determine the number of PD sources present. Since each PD pulse can then be associated with a particular subset of data, subsequent PD interpretation could be applied to separate PD date streams, rather than to mixed data. An radio frequency (RF) detector circuit was designed and used to validate the approach by means of experiments carried out in a model transformer tank. The technique should be applicable to other condition monitoring systems that are required to distinguish between complex transient signals.
Original languageEnglish
Pages (from-to)256-267
Number of pages12
JournalIEE Proceedings Science Measurement and Technology
Volume4
Issue number5
DOIs
Publication statusPublished - Sept 2010

Keywords

  • partial discharge sources
  • comparison
  • UHF signals

Fingerprint

Dive into the research topics of 'Differentiating between partial discharge sources using envelope comparison of ultra-high-frequency signals'. Together they form a unique fingerprint.

Cite this