Employing the mathematical morphology filters for de-noising Partial Discharge signals

Partial Discharge (PD) detection is mainly used for quality assessment and fault identification in high voltage power apparatus. Study of PD pattern reveals the nature and type of defects in the insulation system. Nowadays online and onsite measured PD signals are highly polluted with external interferences and severe noise. Accordingly the paper is broadly classified into two parts. In first part, it reveals about the modeling of PD mechanism for understanding the characteristics using simulation. The second part describes about the external interferences which are artificially created and mixed with simulated PD signals and denoising by using various techniques. In this proposed work, a feature oriented noise reduction technique, based on Mathematical Morphological Filters (MMF), is used. The efficiency of the denoising technique is proved by comparing with different Wavelet transform (WT) denoising techniques.