Ardila Rey, Jorge Alfredo

Partial discharge (PD) monitoring is one of the most used tools for diagnosing the condition of electrical equipment and machines that operate normally at high voltage levels. Ideally, PD identification can be easily done if there is a single source acting over the electrical asset during the measurement. However, in industrial environments, it is common to find the presence of multiple sources acting simultaneously, which hinders the identification process, due to sources of greater amplitude hiding the presence of other types of sources of lesser amplitude that could eventually be much more harmful to the insulation system. In this sense, the separation of PD through the use of clustering techniques allows individual source recognition once they have been clearly separated. This article describes the main clustering techniques that have been used over time to separate PD sources and electrical noise. The results obtained by the different authors in the utilization of each technique demonstrates good performance in terms of separation.

One of the main tools for monitoring the condition of high voltage equipment is the measurement of partial discharges (PD). The electromagnetic (EM) radiation originated from this degradation phenomenon can be captured by various types of ultra-high frequency (UHF) antennas carefully designed and optimised for specific frequency bands. However, the presence of environmental noise may limit the use of this technique. Different types of monopole antennas normally used in UHF PD measurement have been evaluated in order to validate the performance of a novel separation technique of EM sources. Accordingly, a new separation technique based on the energy ratio of the captured signals was developed, considering noise interferences. The results revealed that the new technique allows an adequate separation, even when three sources act simultaneously.

Pulsed plasma discharges, such as the plasma focus, are a source of pulsed X rays, therefore it is desirable to understand the relationship between this fast transient phenomena and the electrical variables of the discharge. Parameters from the electrical diagnostic signals are typically used to characterize the plasma focus discharge and for the correlations with X rays measurements via scatter plots. To further evaluate relevant information in the electrical signals, besides the characteristic parameters, an implementation of different types of machine learning algorithms, that included deep learning, was performed. A classification of pulses associated with an X rays measurement, in terms of the electrical signals data as input, was carried out. Two approaches were compared: the selection of the characteristic parameters and the use of the entire signals so the algorithms could find additional information for the classification task. The electrical diagnostic signals corresponded to: the voltage at the electrodes of the discharge chamber measured with a resistive voltage divider; time variation of the circuit current measured with a Rogowski coil and an inductive loop sensor; and the electromagnetic burst from the circuit measured with a Vivaldi antenna. The X rays measurement corresponded to the signal obtained from a scintillator-photomultiplier. In terms of the performance of the algorithms models in this classification problem, the results indicated that there is no significative improvements when using the entire signal or the selection of characteristic parameters. The best results were obtained when the following parameters were used: voltage at time of gas breakdown, voltage at time of pinch, current at time of pinch, time derivative of current at time of pinch, time from breakdown to pinch, and the Fast Fourier Transform of the part of the Vivaldi antenna signal related to the pinch event.

Electrical trees are one of the main mechanisms of degradation in solid polymeric insulation leading to the failure of high voltage equipment. They are interconnected networks of hollow tubules typically characterized from two-dimensional (2D) projections of their physical manifestation. It is shown that complete characterization requires a three-dimensional (3D) imaging technique such as X-ray computed tomography (XCT). We present a comprehensive set of parameters, quantitatively characterizing two types of tree topology, conventionally known as bush- and branchtype. Fractal dimensions are determined from 3D models and from 2D projections, and a simple quantitative relationship is established between the two for all but dense bush trees. Parameters such as number of nodes, segment length, tortuosity and branch angle are determined from tree skeletons. The parameters most strongly indicative of the differences in the structure are the number of branches, individual channel size, channel tortuosity, nodes per unit length and fractal dimension. Studying two stages of a bush tree's development showed that channels grew in width, while macroscopic parameters such as the fractal dimension and tortuosity were unchanged. These parameters provide a basis for tree growth models, and can shed light on growth mechanisms.