Feature-oriented de-noising of partial discharge signals employing mathematical morphology filters

Abstract

Noise reduction is one of the important tasks of cable insulation assessment based on partial discharge. This paper has developed a novel attitude in noise reduction, i.e. feature-oriented de-noising. Implementing this idea, a new de-noising filter based on Mathematical Morphology Filters (MMF) is presented where the process of removing white noise mixed with partial discharge signal is executed by employing several simple morphological filters. Comparing with the de-noising process based on Wavelet-transform, in terms of extracted features after de-noising, the proposed method preserves the features of PD pulses better. The performance evaluations and comparisons are examined by simulated and field measured signals.