Jorge Ardila-Rey

Partial discharge and noise separation by means of spectral-power clustering techniques

Partial Discharges (PDs) are one of the most important classes of ageing processes that occur within electrical insulation. The measurement of PDs is useful in the diagnosis of electrical equipment because PDs activity is related to different ageing mechanisms. Classical Phase-Resolved Partial Discharge (PRPD) patterns are able to identify PD sources when they are related to a clear degradation process and when the noise level is low compared to the amplitudes of the PDs. However, real insulation systems usually exhibit several PD sources and the noise level is high, especially if measurements are performed on-line. High-frequency (HF) sensors and advanced signal processing techniques have been successfully applied to identify these phenomena in real insulation systems. In this paper, spectral power analyses of PD pulses and the spectral power ratios at different frequencies were calculated to classify PD sources and noise by means of a graphical representation in a plane. This technique is a flexible tool for noise identification and will be useful for pulse characterization.

Separation of sources in radiofrequency measurements of partial discharges using time-power ratio maps.

Partial discharges measurement is one of the most useful tools for condition monitoring of high-voltage (HV) equipment. These phenomena can be measured on-line in radiofrequency (RF) with sensors such as the Vivaldi antenna, used in this paper, which improves the signal-to-noise ratio by rejecting FM and low-frequency TV bands. Additionally, the power ratios (PR), a signal-processing technique based on the power distribution of the incoming signals in frequency bands, are used to characterize different sources of PD and electromagnetic noise (EMN). The calculation of the time length of the pulses is introduced to separate signals where the PR alone do not give a conclusive solution. Thus, if several EM sources could be previously calibrated, it is possible to detect pulses corresponding to PD activity.

Inductive Sensor Performance in Partial Discharges and Noise Separation by Means of Spectral Power Ratios

Partial discharge (PD) detection is a standardized technique to qualify electrical insulation in machines and power cables. Several techniques that analyze the waveform of the pulses have been proposed to discriminate noise from PD activity. Among them, spectral power ratio representation shows great flexibility in the separation of the sources of PD. Mapping spectral power ratios in two-dimensional plots leads to clusters of points which group pulses with similar characteristics. The position in the map depends on the nature of the partial discharge, the setup and the frequency response of the sensors. If these clusters are clearly separated, the subsequent task of identifying the source of the discharge is straightforward so the distance between clusters can be a figure of merit to suggest the best option for PD recognition. In this paper, two inductive sensors with different frequency responses to pulsed signals, a high frequency current transformer and an inductive loop sensor, are analyzed to test their performance in detecting and separating the sources of partial discharges.

On the Use of Monopole Antennas for Determining the Effect of the Enclosure of a Power Transformer Tank in Partial Discharges Electromagnetic Propagation.

A well-defined condition-monitoring for power transformers is key to implementing a correct condition-based maintenance (CBM). In this regard, partial discharges (PD) measurement and its analysis allows to carry out on-line maintenance following the standards IEC-60270 and IEC-60076. However, new PD measurements techniques, such as acoustics or electromagnetic (EM) acquisitions using ultra-high-frequency (UHF) sensors are being taken into account, IEC-62478. PD measurements with antennas and the effect of their EM propagation in power transformer tanks is an open research topic that is considered in this paper. In this sense, an empty tank model is studied as a rectangular cavity and their resonances are calculated and compared with their measurement with a network analyser. Besides, two low cost improved monopole antennas deployed inside and outside of the tank model capture background noise and PD pulses in three different test objects (Nomex, twisted pair and insulator). The average spectrum of them are compared and can be found that mainly, the antenna frequency response, the frequency content distribution depending on the PD source and the enclosure resonances modes are the main factors to be considered in PD acquisitions with these sensors. Finally, with this set-up, it is possible to measure PD activity inside the tank from outside.

Multiple partial discharge source discrimination with multiclass support vector machines

Different types of partial discharges are created with test objects in laboratory.Their frequency content depends on the type of discharge and other external factors.An SVM extracts characteristics from the power spectral density of the pulses.Noise, corona, internal and surface discharges have different characteristics.The differences are used to classify discharges and separate them from noise. The costs of decommissioning high-voltage equipment due to insulation breakdown are associated to the substitution of the asset and to the interruption of service. They can reach millions of dollars in new equipment purchases, fines and civil lawsuits, aggravated by the negative perception of the grid utility. Thus, condition based maintenance techniques are widely applied to have information about the status of the machine or power cable readily available. Partial discharge (PD) measurements are an important tool in the diagnosis of power systems equipment. The presence of PD can accelerate the local degradation of insulation systems and generate premature failures. Conventionally, PD classification is carried out using the phase resolved partial discharge (PRPD) pattern of pulses. The PRPD is a two dimensional representation of pulses that enables visual inspection but lacks discriminative power in common scenarios found in industrial environments, such as many simultaneous PD sources and low magnitude events that can be hidden below noise. The literature shows several works that complement PRPD with machine learning detectors (neural networks and support vector machines) and with more sophisticated signal representations, like statistics captured in several modalities, wavelets and other transforms, etc. These methods improve the classification accuracy but obscure the interpretation of the results. In this paper, the use of a support vector machine (SVM) operating on the power spectrum density of signals is proposed to identify different pulses what could be used in an online tool in the maintenance decision-making of the utility. Particularly, the approach is based on an SVM endowed with a special kernel that operates in the frequency domain. The SVM is previously trained with pulses of different PD types (internal, surface and corona) and noise that are obtained with several test objects in the laboratory. The experimental results demonstrate that this technique is highly effective in identifying PD for cases where several sources are active or when the noise level is high. Thus, the early identification of critical events with this approach during normal operation of the equipment will help in the decision of decommissioning the asset with reduced costs and low impact to the grid reliability.

Automatic selection of frequency bands for the power ratios separation technique in partial discharge measurements: part I, fundamentals and noise rejection in simple test objects

Partial discharge (PD) measurements are important for the monitoring of the status of almost any type of high-voltage equipment. The use of this technique for on-line measurements involves several challenges related to noise rejection and PD source recognition. In order to face these issues, pulse shape analyses are made in high-frequency and very-high-frequency ranges. The authors proposed in a previous paper the so-called Power Ratios (PR) of the spectra as useful parameters for pulse source separation in a two-dimensional approach. This technique provides good results, but its reliability depends on the appropriate selection of two frequency intervals where these spectral power ratios are calculated. In order to improve this separation system, the authors propose an algorithm that provides the frequency intervals necessary to obtain very good separation results in several experimental setups. The explanation of the system and the results obtained are presented in two papers. The simultaneous measurements of pulse sources (PD or noise) are complemented with pulse characterization when only one single source is active in order to check that the selected frequency intervals are appropriate. In this first paper, the authors summarize the fundamentals of the proposed algorithm and its first results for PD and noise separation in simple test objects that create corona, internal and surface discharges.

A Partial Discharges acquisition and statistical analysis software

Partial Discharges (PDs) are responsible for unexpected failures in power system equipment, so their measurement is a fundamental tool for electrical equipment maintenance. In order to characterize PDs activity, some statistical magnitudes are necessary. For this purpose, PDs acquisition and processing is an important tool for critical decisions related to power systems. In this paper, the main difficulties and challenges facing PDs detection, acquisition and processing are presented. Results will be compared to a commercial PDs detection system.

Antenna selection and frequency response study for UHF detection of partial discharges

Partial Discharge (PD) detection is a widely extended technique for electrical insulation diagnosis. Classical PD detection by means of phase resolved patterns require electrical connections to the power equipment and is sensitive to many noise sources. Ultra High Frequency (UHF) detection techniques are being recently proposed to overcome these problems, and to detect partial discharges on-line. In this paper, four antennas will be tested in order to compare their response to this physical phenomenon.

On the use of Vivaldi antennas in the detection of partial discharges

Due to their impulsive nature, partial discharges in insulation systems emit in a broad band of frequencies ranging from tens of megahertz to gigahertz. Antennas can be placed at secure distances from the high-voltage source and measure partial discharges pulses on-line. The energy of partial discharges coexists with strong electromagnetic emissions such as frequency modulated radio, television broadcasting and mobile telephones. Then, the antennas should be tuned to detect the emission from partial discharges in bands where the electromagnetic noise is not so important. They should also be capable of receiving in a broad range of frequencies so resonant antennas should be avoided. Vivaldi antennas can be designed to detect emissions from a desired tuned frequency and extended to a band as wide as necessary. This makes them a valuable tool to detect partial discharges avoiding electromagnetic noise present at lower frequencies.

Partial discharge source recognition by means of clustering of spectral power ratios

Partial discharge (PD) detection can give useful information for the diagnosis of electrical apparatus, but data interpretation can be impossible if several sources are simultaneously active. Pulse characterization can be used to identify the source origin in PD measurements. The distribution of energy at different frequencies helps in distinguishing several types of discharges. The parameterization of pulses by means of spectral power ratios (PR) has been studied as a reliable technique to represent different characteristics in high-frequency current pulses in high-voltage tests. In this study, the separation of PD sources by means of PR maps is proposed. This approach has proven to be effective in the identification of PD sources when two electrical insulation systems are simultaneously subjected to discharge activity in controlled experiments where internal, surface and corona discharges were occurring. The flexibility of the system to improve cluster separation is shown, and measurements are also made on a real insulated power cable, where two simultaneous PD sources were successfully identified.