MÉTODO PARA RECONHECIMENTO DE DESCARGA PARCIAL EM APLICAÇÕES DE ALTA TENSÃO E UNIDADE USANDO O MÉTODO
Patent Information
- Authority / Receiving Office
- BR · BR
- Patent Type
- Patents
- Current Assignee / Owner
- HSP HOCHSPANNUNGSGERTE GMBH
- Filing Date
- 2020-07-17
- Publication Date
- 2026-08-04
Smart Images
Figure 00000016_0000 
Figure 00000016_0001
Abstract
Description
[0001] The present invention relates to a method for recognizing partial discharge in high-voltage applications and to a high-voltage unit using the method, comprising the steps of detecting a signal, transforming the time-domain signal to a frequency-domain signal, clipping frequencies above a defined limit, and retransforming the truncated frequency-domain signal to a time-domain signal.
[0002] Partial discharge is a localized dielectric breakdown of electrical insulation at high voltages. In high-voltage applications, for example, in measuring transformers, for voltages in the range of a few kilovolts to more than a thousand kilovolts, partial discharge can result in failures, measurement errors, damage, even complete demolition of the units and disruption of the electrical network.A dielectric breakdown occurs between conductors through an insulator, particularly a solid insulator such as a ceramic, silicon, and / or non-conductive composite material, or a fluid such as oil and / or SF6 or Clean Air type gas, without traversing the entire space between the conductors. In fluids such as oil, partial discharge may result in bubbles, but most of the time the discharge is not visible. In or through solid insulators, partial discharge is usually not visible until irreversible damage to the equipment occurs.
[0003] Partial discharge in or within solid insulators begins within gaps, cracks, contaminants, or inclusions, particularly at dielectric conductive interfaces, and in liquids, partial discharge begins particularly at bubbles, contaminants, or inclusions. Because partial discharge is limited to only a portion of the insulation, the discharge will only partially traverse the distance between Petition 870250074765, dated 08 / 22 / 2025, p. 10 / 46 2 / 13 the conductors. As soon as the partial discharge is initiated, high-frequency transient current pulses will appear and persist for nanoseconds to microseconds.
[0004] In this case, the current pulses disappear and reappear repeatedly as a sinusoidal wave passes through the zero crossing. The detection and measurement of partial discharges are difficult, complex, and expensive due to the high voltages and high levels of electrical background noise. Particularly in operation, partial discharges in high-voltage applications are difficult to identify and determine, and the electrical damage and / or losses can be enormous.
[0005] It is particularly difficult to distinguish between noise, for example, introduced due to corona signals in substations, crosstalk signals, electromagnetic signals, or other sources, and partial discharges, and correcting partial discharge measurements on-site or online is very complicated due to the high noise level, which obscures the real signals. Due to the short duration and nanosecond rise times of partial discharge currents, a common way to quantify the magnitude of a partial discharge is in picocoulombs. To investigate a partial discharge, an intensity versus time is displayed. For example, on an oscilloscope, a partial discharge appears as equally spaced burst events at the peak of a sine wave, while an arc and spark occur randomly.
[0006] Partial discharge measurements are performed both for quality assessment and diagnostic purposes. Partial discharge is, for example, an indicator of the status of the device under test. This measurement is performed, for example, in factory quality tests, particularly in a Faraday cage with a defined power supply, or in operation on the mains power grid with a certain amount of background noise. Especially for online testing and / or monitoring. Petition 870250074765, dated 08 / 22 / 2025, page 11 / 46 3 / 13 of high-voltage units, for example, high-voltage switches, bushings, and / or transformers, different methods have been developed. One method is based on Transient Ground Voltages, i.e., voltage peaks induced on the surface of a surrounding grounded metal.
[0007] Other methods for detecting and measuring partial discharge use ultrasonic acoustic sensors, ultra-high frequency sensors / antennas, high-frequency current transformer sensors, and / or directional couplers. Time-domain reflectometry, observing the reflected waveforms, particularly displayed in a partial discharge mapping format, allows for spatial localization of insulation irregularities. Phase-resolved partial discharge patterns, which correlate recorded signals and applied voltage, and time-frequency maps, which use the features of a recorded signal to discriminate between different sources, as well as cross-correlation analysis of recorded signals, allow for partial discharge investigation.
[0008] Particularly during on-site measurements, the methods described above provide only low sensitivity and require expensive, highly complex equipment.
[0009] One objective of the present invention is to overcome the problems described above. In particular, one objective of the present invention is to increase the sensitivity of partial discharge recognition and determination, reduce costs by enabling partial discharge measurements with less complex equipment, reduce noise, and improve measurement results.
[0010] The above objectives are achieved by a method for partial discharge recognition in high voltage applications, the method comprising the following steps: - to detect a signal; Petition 870250074765, dated 08 / 22 / 2025, page 12 / 46 4 / 13 - transform the time-domain signal into a frequency-domain signal - Cut off frequencies above a defined limit; - retransform (inverse transform) the truncated frequency domain signal to the time domain. The information content of the detected and truncated signals is determined and compared
[0011] In high voltage applications, it is very difficult to distinguish between noise and partial discharge signals. Partial discharge signals are small, but comprise information in contrast to pure noise signals. Direct detection, identification and measurement, particularly monitoring of partial discharge signals requires expensive and complex equipment and is not always possible in a reliable manner.By measuring or detecting a signal, transforming the signal to the frequency domain, clipping frequencies above a defined limit, or retransferring truncated frequency domain signals to the time domain, and by determining and comparing the information content of the detected and truncated signals, it will be possible to distinguish between noise and partial discharge with less measurement effort, greater reliability, and lower cost. A partial discharge can be reliably recognized by appearance, and measures can be initiated, for example, shutting down units and / or equipment, and / or cutting the electrical connection to the power grid, in order to prevent damage and / or failures. A partial discharge is recognized and determined with greater sensitivity, at reduced costs.The method according to the present invention enables partial discharge measurements with less complex equipment, noise reduction, and improved measurement results by identifying and distinguishing between a partial discharge and noise through analysis of the information content of the signals after cutting off the higher frequencies of the measured signals above a certain limit.
[0012] The method can understand signals with contents of Petition 870250074765, dated 08 / 22 / 2025, p. 13 / 46 5 / 13 Comparable information, particularly similar to that of the detected and truncated signal that is identified as noise, and / or signals with lower information content of the truncated signal compared to the information content of the detected signal that is identified as a signal that includes partial discharge information. This will allow distinguishing between noise and partial discharge signals, with advantages such as those described above.
[0013] The information content of the detected and truncated signals can be calculated and / or the comparison of the information content can be done by means of subtraction and / or division. Subtraction and division are easy to perform, particularly automatically and in a short time or at the same time as the measurements. The determination and / or comparison of the information content of the detected and / or truncated signals, particularly done in real time and / or automatically, can be performed with a computer and / or online. A quick and particularly automatic determination and / or comparison of the information content allows, for example, real-time monitoring of partial discharge events and a rapid reaction to protect equipment, for example, by disconnecting power lines. Maintenance may be initiated for equipment or units that show partial discharge events once and / or more frequently.An alert signal may be produced and / or transmitted, for example, locally and / or online via the internet or mobile network, or partial discharge events may be monitored online, for example, via the internet over periods of time.
[0014] Steps, particularly all steps of the method according to the present invention, may be repeated, particularly periodically over a fixed period of time, in order to increase the reliability of the results.
[0015] Determining the content of information could be Petition 870250074765, dated 08 / 22 / 2025, p. 14 / 46 6 / 13 done through a Shannon entropy calculation. Entropy is a measure of the average value of the information content of information. The information content of signals can be calculated. Claude Elwood Shannon defined the entropy of a memoryless, discrete information source, or a discrete random variable H for a given number of characters z. An information content I(z) = -log2pz is assigned to each probability p of an event z or signal value. The entropy of the character z is defined as the expected value of the information content Hi = -Zpzlog2pz with the probability pz of the character z occurring. Summations with different probabilities do not add to the sum by definition. The entropy of a signal w with a length n is Hn = Ipwlog2pw with the probability pw of the signal w occurring. The probability H is derived as a limit (“Limes”) when n ^ ^ of H = lim(n ^ ~) Hn / n.
[0016] Noise signals have no information content, since the value of the signals is random. Cutting high frequencies does not change Shannon entropy, since no information has been lost. As for partial discharge signals, cutting frequencies reduces Shannon entropy, since information content has been lost. Comparing the Shannon entropy of the as-measured signals and the Shannon entropy of the truncated signals results in the same values for pure noise, and the lower Shannon entropy values of the truncated signals compared to the Shannon entropy of the as-measured signals for signals comprising a partial discharge. The occurrence of a partial discharge can be identified, and the value of a partial discharge can be calculated, with advantages such as those described above.
[0017] A detected signal could be and / or could comprise a voltage signal and / or a current signal. For example, high-frequency transient current and / or voltage pulses could be measured, persisting for nanoseconds to microseconds, disappearing and reappearing. Petition 870250074765, dated 08 / 22 / 2025, p. 15 / 46 7 / 13 repeatedly as a sinusoidal wave passes through the zero crossing. There is no need to identify partial discharge pulses within a noisy signal using complex and expensive high-resolution equipment. With the method according to the present invention, an easy and inexpensive device can be used to detect and measure a partial discharge, or even partial discharge signals with a magnitude in the picocoulomb range at high voltages and high levels of electrical background noise. The high voltage can be in the voltage range: from a few kilovolts up to 1200 kV, particularly from 15 kV to 1200 kV.
[0018] With a recognized and / or identified partial discharge, an alarm signal may be produced and / or sent, particularly online via the internet and / or mobile network. Other measures may be taken to prevent damage and / or destruction of the units, for example, disconnecting the high voltage and / or cutting the power lines from the units to the power grid, or personnel may be sent to inspect and / or maintain the equipment or electrical units where a partial discharge occurred.
[0019] A partial discharge value can be evaluated by means of an iterative method, excluding signals identified as noise and / or determining a cutoff frequency, above which signals are identified as noise.
[0020] A high-voltage unit according to the present invention, particularly using a method such as described above, is or comprises a measuring transformer, a high-voltage switch, a surge suppressor, a bushing, and / or a medium or high-voltage transformer.
[0021] The unit may comprise an insulator, in particular an insulator housing with slots on the outer surface.
[0022] The unit may be designed for voltages in the range of Petition 870250074765, dated 08 / 22 / 2025, page 16 / 46 8 / 13 voltage: from a few kilovolts up to 1200 kV.
[0023] The advantages with respect to the high-voltage unit described according to the present invention, particularly using a method such as that described above, are similar to the advantages previously described with respect to the method for partial discharge recognition in high-voltage applications.
[0024] The present invention will be described in more detail below with reference to an embodiment illustrated in the accompanying drawings, in which:
[0025] Figure 1 schematically illustrates a measured signal transformed to the frequency domain, with a partial discharge signal, and
[0026] Figure 2 schematically shows a measured signal transformed to the frequency domain, completely composed of noise.
[0027] Figure 1 shows a measured signal from a partial discharge, transformed to the frequency domain. The signal is measured, for example, in the form of voltage and / or current over time. An alternating voltage and / or current is introduced by means of partial discharge events, for example, between two electrical poles, with an insulator in between, particularly an insulator of a high-voltage device or application / unit. As described above, the partial discharge signal is small compared to the high voltage between the poles, with the high voltage, for example, in the range of a few kilovolts up to 1200 kV, compared to the signals of a few picocoulombs from a partial discharge, i.e., to the small voltage and current signals. A partial discharge occurs, for example, through the outer surface of a solid insulator. A solid insulator, for example, consists of or comprises ceramic, silicone, and / or composite materials.The insulator is a particularly important part of a high-voltage unit, for example. Petition 870250074765, dated 08 / 22 / 2025, page 17 / 46 9 / 13 For example, a measuring transformer, a switch, a bushing, a transformer and / or a surge suppressor. The insulator is, for example, designed in the form of a housing and / or mounting structure of the high-voltage unit, for example, in a hollow cylindrical shape with slots on the outer surface. Partial discharge will also occur, for example, in fluidized insulators, particularly in the oil of a high-voltage unit.
[0028] Partial discharge is a localized dielectric breakdown of electrical insulation at high voltages between electrical poles. Partial discharge in or within the insulator begins, for example, within gaps, cracks, contaminants, or inclusions, particularly at dielectric conductive interfaces, and in liquids, partial discharge begins particularly at bubbles, contaminants, or inclusions. Partial discharge is limited to only a portion of the insulation. The discharge will only partially bridge the distance between the poles or between the electrical conductors. Once partial discharge is initiated, high-frequency transient current pulses will appear and persist for nanoseconds to microseconds. In this case, the current pulses disappear and reappear repeatedly as, for example, a sinusoidal wave passes through the zero crossing. Partial discharge signals are short-lived and exhibit current rise times in the nanosecond range.High levels of electrical background noise, for example due to corona, crosstalk, and other effects, hinder accurate measurement and the recognition or simple determination of partial discharge within a measured signal of high voltages / currents.
[0029] Measurements of a signal S, for example, a current and / or voltage signal, are performed according to time. Due to the short duration and rise times in the nanosecond range of partial discharge currents in high-voltage applications, visualization and distinction between noise and discharge becomes difficult. Petition 870250074765, dated 08 / 22 / 2025, p. 18 / 46 10 / 13 partial voltage / current versus time graphs. In order to identify a partial discharge and / or determine a partial discharge value, according to the present invention, the information content I of a signal S is investigated. The time-measured signals S(t) are transformed from the time domain to the frequency domain, for example, by Fourier Transform and / or Discrete Cosine Transform. Frequencies f above a predefined frequency limit fnm are clipped and truncated signals are obtained by removing the signal components at frequencies above fnm. The frequency limit value fnm is, for example, defined at each time at the beginning of a monitoring or measurement activity. In a step after removing frequencies f above a predefined frequency limit fnm, the signals are again transformed or retransformed from the frequency domain to the time domain.
[0030] Figure 1 shows an example of a partial discharge signal S(f) in the frequency domain, i.e., after the time domain transformation, where each component of the signal S(fi) has information h. For comparison, Figure 2 shows a signal S(fi) entirely composed of noise. Thus, no component of S(fi) will have any information, since the signal is entirely random. The signal information I is, for example, obtained as J hdf = I — ÜO, that is, by adding the information associated with each component. As observed in Figure 1 and Figure 2, partial discharge signals will contain information, while noise signals will not contain information. Noise signals are completely uncorrelated with each other. When changing from the time domain to the frequency domain, each component will have a part of the overall information L. By cutting parts of the signal above the frequency limit füm, the information will be Petition 870250074765, dated 08 / 22 / 2025, page 19 / 46 11 / 13 lost, when the measured signal comprises a partial discharge and does not consist solely of noise. According to the present invention, the method or an algorithm shall extract information from the sampled signals, for example, by calculating Shannon entropy, and compare the information content of the sampled signal with the information content of the truncated signal, derived by cutting off the signal frequencies above the flim frequency limit.
[0031] The information contents of the measured or sampled signals will be calculated, particularly by a local computer and / or in the cloud, as described above, for example, when calculating Shannon entropy, and the information contents of the truncated signal or the signal after frequency clipping will similarly be calculated, particularly by a local computer and / or in the cloud, as described above, for example, when calculating Shannon entropy. Both calculations will be, for example, performed in time with measurements or separately one after the other, or in a predefined order. After removing the frequency components, a time-domain signal for the truncated signal is obtained, for example, by an inverse integral transform, particularly by Fourier Transform and / or Discrete Cosine Transform.Determining the information content of measured and truncated signals is analogous, for example, in both cases by calculating Shannon entropy.
[0032] A comparison of the information contents of the measured and truncated signals results, according to the present invention, in a recognition and / or determination of partial discharge. A comparison is made, for example, manually and / or automatically, for example, by a computer and / or in the cloud. Methods for comparing signals include, for example, dividing and / or subtracting the information contents of the measured and truncated signals. For division, a result will be Petition 870250074765, dated 08 / 22 / 2025, page 20 / 46 12 / 13R_ I truncated signal ,,χI original signal! less than one, if a measured signal comprises a partial discharge. Without partial discharge, one has _ I truncated signal I original signal a division result, which is exactly or mostly one, since a signal consists only of noise, and no lower frequencies will be lost by cutting off the frequencies.
[0033] Partial discharge recognition and / or determination offers advantages such as those described above, for example, an alert can be triggered and / or transmitted, and other actions taken, such as voltage shutdown and / or disconnection of high-voltage units from the power grid, in order to prevent damage and / or destruction / failure of the units. Partial discharge measurements also provide quality assessment and diagnostic results. Partial discharge will, for example, be an indicator of the status of the device under test. It will be performed in factory quality tests to demonstrate proper function of the units produced. In operation on the power grid, with high background noise, online monitoring of high-voltage units will be possible, for example, high-voltage switches, bushings and / or transformers.
[0034] Signal detection is performed, for example, based on Transient Ground Voltages, i.e., voltage spikes induced on the surface of a surrounding grounded metal. Other methods for detecting and measuring a partial discharge use ultrasonic acoustic sensors, ultra-high frequency sensors / antennas, high-frequency current transformer sensors, and / or directional couplers.
[0035] The above-described embodiments of the present invention Petition 870250074765, dated 08 / 22 / 2025, page 21 / 46 13 / 13 may also be used in combination or combined with known state-of-the-art embodiments. For example, a partial discharge may be detected once by performing the steps, particularly all the steps, described above. Greater reliability will be achieved by performing the steps, particularly all the steps described above, repeatedly. An iterative method for determining a partial discharge may include determining a flim frequency limit value. An iterative method may comprise repeating the steps according to the present invention and altering the flim cutoff frequency limit until a change in Shannon entropy occurs and / or until no change in Shannon entropy occurs, with an altered flim frequency limit.
Claims
1. Method for partial discharge recognition in high voltage applications, comprising the steps of: - detecting a signal (S) comprising a voltage signal and / or a current signal; - transforming the signal (S) from time domain [S(t)] to frequency domain [S(f)]; - clipping the frequencies (f) of the signal above a defined limit; - retransforming the truncated signal (S) from frequency domain [S(f)] to time domain (S(t)); characterized in that an information content (I) of the detected signals and the truncated signal (S) is determined and compared, wherein the determination of an information content is made by a Shannon entropy calculation.
2. Method, according to claim 1, characterized in that signals (S) with comparable information contents (I), particularly equal to those of the detected and truncated signal, are identified as noise, and / or signals with lower information contents (I) than the truncated signal (S) compared to the information contents (I) of the detected signal are identified as a signal (S) that includes partial discharge information.
3. Method, according to any one of claims 1 or 2, characterized in that the information contents (I) of the detected and truncated signals (S) are calculated and / or the comparison of the information contents (I) is done by subtraction and / or division.
4. Method, according to any one of claims 1 to 3, characterized in that the determination and / or comparison of the information contents (I) of the detected and / or truncated signals (S) are made in real time, particularly automatically, Petition 870250074765, dated 22 / 08 / 2025, page 23 / 46 2 / 2 particularly automatically with a computer and / or online.
5. A method, according to any one of claims 1 to 4, characterized in that the steps are repeated particularly periodically over a fixed period of time in order to increase the reliability of the results.
6. Method, according to any one of claims 1 to 5, characterized in that an alarm signal (S) is produced and / or sent, particularly online via the internet and / or mobile network, in recognition and / or identification of a partial discharge.
7. Method, according to any one of claims 1 to 6, characterized in that a value of a partial discharge is evaluated by an iterative method, deleting the signals (S) identified as noise and / or identifying a cutoff frequency (flim), above which the signals (S) are identified as noise.
8. High-voltage unit using a method, as defined in any one of claims 1 to 7, characterized in that the unit is or comprises a measuring transformer, a high-voltage switch, a surge suppressor, a bushing, and / or a medium or high-voltage transformer.
9. High-voltage unit according to claim 8, characterized in that the unit comprises an insulator, particularly an insulator housing with slots on the outer surface.
10. High-voltage unit, according to any one of claims 8 or 9, characterized in that the unit is designed for voltages in the voltage range: from one kilovolt to 1200 kV, particularly in the range of 15 kilovolts to 1200 kV.