Practical partial discharge pulse generation and location within transformer windings using regression models adjusted with simulated signals

Localisation of inter-layer partial discharges in transformer windings by logistic regression and different features extracted from current signals

“…Capacitive disturbances were performed between the taps of the winding prototype for the experimental generation of current pulses to emulate the occurrence of PDs. This approach differs from the one presented in [16], which creates discharges between the prototype taps and a grounded point. In this paper, the proposed approach is more realistic, since it considers PDs originating between winding layers, especially regarding the impedances involved in the propagation of current pulses.…”

“…In this paper, the proposed approach is more realistic, since it considers PDs originating between winding layers, especially regarding the impedances involved in the propagation of current pulses. Besides, in contrast to the several capacitance values required in [16], only one external capacitor was necessary for the experimental tests.…”

“…They are trained with a data set composed of features extracted from terminal current signals, towards an accurate and efficient PD location. Examples include use of neuro‐fuzzy networks [14], trained in an unsupervised way by orthogonal transformations of PD signals, application of a linear combination of digital filters based on wavelet coefficients and principal component analysis [15], and use of linear regression models based on statistical and energy features extracted from terminal signals [16]. Besides, deep learning techniques have also been applied for PDs pattern recognition, as performed in [17, 18].…”

Multiple Linear Regression Models for Partial Discharge Location Along Transformer Windings

“…Among the various ways of capturing the activity of the P Ds, the measurement of current pulses generated by them is highlighted. In this way, high frequency current transformers can be used, installed in the accessible terminals of the windings, i.e., in the phases and neutral/earth terminals (Gonçalves Júnior et al, 2018).…”

“…Several techniques for locating P Ds, based mainly on the high frequency current signals captured at the accessible endings of the windings, have been proposed in recent years. As examples, Nafar et al (2011) made use of correlation coefficients between waveforms to determine the point of occurrence of the P Ds; Homaei et al (2014) used unsupervised neuro-fuzzy networks, by means of orthogonal transformation of the received signals; Guillen et al (2016) employed Wavelet-Laplace functions in conjunction with Hellinger's distance calculation; Rahman et al (2016) used a combination of linear and digital filters, based on wavelet coefficients and the main components of the signals; Gonçalves Júnior et al (2018) applied regression models using statistical features from terminal currents of an experimental winding; among others. Although all these methods present significant performances, with a high percentage of hits for the P Ds location, the vast majority was evaluated for only one type of transformer windings, being the applicability in other windings structures thus unknown.…”