Partial discharge detection through ranked-ordered-wavelet based compressive sensing

Partial discharge (PD) phenomena in a power system is a precursor of an insulation breakdown or failure of power system. Equipment failure can be avoided if the PDs are diagnosed at an early stage. Some PDs are Ultra Wide Band (UWB) in nature and their frequency spectra may extend up to 5 GHz. One of the main challenges of UWB PD detection is the noises and interferences. Another challenge is the requirement of very high sampling rate ADC. A cost effective and simpler PD detection technique with lower sampling rate and strong denoising algorithm is needed for the diagnosis of UWB PD-pulses. UWB signal can be sampled at a lower sampling rate by adopting compressive sensing methods. At the same time the noisy UWB signals can be denoised if wavelet based methods are applied. This paper describes a novel PD detection technique by using ranked-ordered-wavelet based compressive sensing method. The benefits of the novel ranked-order wavelet based compressive sensing are mainly twofold: minimization of the sampling rate and denoising of the UWB signals.