Abstract

The acquisition of discharges data is usually affected by external interference leading to erroneous interpretations. In the viewpoint of improving computation accuracy, the discrete wavelet transform (DWT) has been used as an effective tool for de-noising purposes. In this study, creeping discharges waveforms have been investigated under the positive lightning impulse voltage over various materials. The used polymers belong to three distinct families (i) thermoplastics (namely polyamide 6 and polyamide 66), (ii) one filled cycloaliphatic epoxy resin and (iii) one unfilled ethylene propylene diene monomer (EPDM). Two approaches namely one-dimensional (1D) and 2D DWT algorithms have been adopted for de-noising signals and images, respectively. Both techniques were applied for simulated and laboratory data using various wavelet families (Daubechies, Symlet and Coiflet). The comparison of de-noising performance for signals has been achieved through the calculation of signal-to-noise ratio, normalised correlation coefficient (Norm_corr) and root mean squared errors. Moreover, the computation of power spectral density of lightning current transients using a periodogram estimator is also analysed. For images, mean square error and peak-signal-to-noise ratio assess the quality of filtering regarding the visualisation of discharges patterns. Results show that Coiflet wavelets demonstrate their effectiveness to reduce background noise and preserve the signals shape and images edges.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call