A correlated empirical mode decomposition method for partial discharge signal denoising

Ya Wen Tang, Cheng Chi Tai, Ching Chau Su, Chien Yi Chen, Jiann Fuh Chen

Research output: Contribution to journalArticle

31 Citations (Scopus)

Abstract

Empirical mode decomposition (EMD) is a signal processing method used to extract intrinsic mode functions (IMFs) from a complicated signal. For a measurement with two or more correlated inputs, finding and capturing the correlated IMFs is a critical challenge that must be confronted. In this paper, a new correlated EMD method is proposed. The cross-correlation method was employed to determine dependence between the IMFs. To verify feasibility, an analysis was performed on simulated test signals and practically measured partial discharge (PD) signals collected from several acoustic emission sensors. At the surface of the gas-insulated transmission line, the PD signal arrived at the AE sensors with varying time delays and unique mechanism vibrations. Following an abnormal detection using the standard-deviation variation, the PD signal and the background signal of each sensor were applied using the correlated-EMD method. A twice correlated-EMD calculation was applied to the signals for the purpose of noise elimination. In addition, the unwanted low-frequency IMFs induced from the EMD calculations were excluded. The experimental results reveal that the correlated-EMD method performs well on both selecting and denoising the correlated IMFs. The results further provide analysis on correlated-input applications with a precise signal completely induced from the disturbance.

Original languageEnglish
Article number085106
JournalMeasurement Science and Technology
Volume21
Issue number8
DOIs
Publication statusPublished - 2010 Aug

All Science Journal Classification (ASJC) codes

  • Instrumentation
  • Engineering (miscellaneous)
  • Applied Mathematics

Fingerprint Dive into the research topics of 'A correlated empirical mode decomposition method for partial discharge signal denoising'. Together they form a unique fingerprint.

  • Cite this