Method for partial discharge detection in medium and high voltage equipment

Abstract

A method (100) for detecting partial discharge in a medium-voltage equipment (1) based on measurement data (3) recorded by at least one sensor (2), the method (100) comprising the steps of: determining (110) whether the measurement data (3) contains at least one partial discharge contribution (4), which is identified as being generated by a partial discharge pulse occurring at a location in the equipment (1) and propagating the generated signal to at least one sensor (2) and being statistically repetitive, wherein determining whether a contribution can be reasonably identified as a partial discharge contribution (4) comprises: determining (112) that the pulses are partial discharge pulses from the same source in response to determining (111) that a plurality of pulses have a consistent pulse pattern determined by the propagation or transmission of the signal from its source to at least one sensor (2); and determining (120) that a partial discharge is occurring in the equipment (1) if at least one such partial discharge contribution (4) exists.

Description

Technical Field

[0001] This invention relates to monitoring partial discharge in medium and high voltage equipment under voltage stress. Background Technology

[0002] When a dielectric material is subjected to voltage stress between two electrodes, it may experience localized dielectric breakdown, which bridges only a portion of the distance between the electrodes. These localized dielectric breakdowns are called partial discharges. Partial discharges typically begin in voids, inclusions, or other defects within the dielectric material. Contamination of the dielectric material, such as the surface of an insulator, is another common cause of partial discharges. If partial discharges persist for too long, the integrity of the dielectric material may be eroded over time until the insulation function completely fails, resulting in a short circuit. Therefore, monitoring for partial discharges in medium and high voltage equipment is essential.

[0003] For partial discharge detection in switchgear or other air-insulated or gas-insulated equipment, various partial discharge sensors or detection systems are employed: capacitive sensors, high-frequency current sensors, ultra-high frequency (UHF) sensors based on radio waves, transient voltage-to-ground (TEV) sensors, or acoustic sensors. Retrofitting capacitive and high-frequency current sensors is typically costly and can introduce points of failure, such as due to dielectric insulation failure. Similar to TEV sensors, they can pinpoint partial discharge sources to specific electrical connections, phases, or compartments or areas within compartments, but they cannot pinpoint their spatial location. Acoustic and UHF sensors are commonly used for this type of location. Typically, multiple sensors are installed using a time-of-flight method. This is not possible with a single sensor; furthermore, such systems cannot handle signal reflections or other distortions that rely on direct time-of-flight methods. They also cannot handle situations with multiple partial discharge sources because they superimpose signals from two sources, a common problem with single-sensor partial discharge systems. This introduces the risk that a strong but harmless partial discharge signal may mask a weaker, more damaging signal. For solutions using multiple sensors to overcome the above problems, see, for example, US 10,488,487B2, which proposes a dedicated calibration procedure for ultra-high frequency applications. Summary of the Invention

[0004] Therefore, the purpose of this invention is to achieve more accurate and reliable detection of partial discharge events in medium and high voltage equipment with appropriate sensor equipment investment.

[0005] This objective is achieved by the method described in the independent claim. Further advantageous embodiments are described in detail in the dependent claims.

[0006] This invention provides a method for detecting partial discharge in medium- and high-voltage equipment based on measurement data recorded by at least one sensor. The measurement data may, for example, include time-series data of measurement values ​​recorded by at least one sensor.

[0007] During the execution of this method, partial discharge contribution analysis is performed on the measurement data. That is, it is determined whether the measurement data contains at least one partial discharge contribution. If at least one such partial discharge contribution exists, it is determined that partial discharge is occurring in the device.

[0008] In this paper, the partial discharge contribution in the measurement data refers to any contribution that satisfies the following conditions:

[0009] - By any suitable criterion, it is reasonably determined that the event was caused by the occurrence of a partial discharge pulse at some point in the device and the resulting signal propagating to at least one sensor, and

[0010] - It appears repeatedly in a statistical manner.

[0011] In particular, the statistical repetition of partial discharge events is a characteristic property. When a device is under AC voltage, the electric field strength rises and falls periodically. Each time the local electric field exceeds the field strength required to trigger a local breakdown, a new partial discharge event is triggered. Here, the exact location of the new partial discharge event at the fault point and the magnitude of the voltage required to trigger it will vary slightly. One reason for this is that each partial discharge event causes a slight difference in the spatial distribution of charge and the electric field strength within the material. Therefore, precise repetition of partial discharge events reflected in the contribution of the measurement data cannot be expected. However, it can be anticipated that each partial discharge event occurring at the same fault point will produce a substantially similar signature in the measurement data. In particular, repetitive signals from the same point are expected to have substantially similar shapes. Specifically, this may mean that the changes in measurement data over time caused by repetitive partial discharge events from the same point are significantly correlated with each other. For example, this can be quantified by the correlation coefficient between such time variations associated with different events being higher than a given threshold.

[0012] The propagation process from the fault point to the sensor (such as a microphone or electromagnetic sensor) can be described by a complex transfer function (here, "complex" means "composite," not necessarily "complex"). For example, an acoustic signal may reflect multiple times within the enclosure housing the switchgear. However, each time the signal propagates from a specific fault point to the sensor, it is processed by the same transfer function because the geometry remains unchanged. Therefore, a detailed understanding of the complex transfer function is not necessary to detect only the presence of a partial discharge contribution.

[0013] Furthermore, and importantly, firstly, the contribution of partial discharge is reasonably assumed to be generated by partial discharge pulses. Not every signal contribution that recurs periodically under AC voltage is generated by partial discharge. For example, even without a partial discharge event, a transformer will emit an audible hum due to electromagnetic forces. However, partial dielectric failures are characterized by the generation of brief discharge pulses. That is, once a breakdown occurs, all available local charge flows through the breakdown point.

[0014] Therefore, in the execution of this method, determining whether a contribution is reasonably identified as a partial discharge contribution includes: in response to determining that multiple pulses have a consistent pulse pattern imparted by a signal propagating from its source or transmitted to at least one sensor, determining that these pulses are partial discharge pulses from the same source.

[0015] In other words, during the execution of this method, for each partial discharge contribution from the same source, the characteristic pulse shape (or "pulse response") of each pulse arriving at the sensor can be learned. This pulse shape is determined by the fault itself and the transfer function propagating to the sensor.

[0016] The inventors discovered that, in this way, signals indicating partial discharge events occurring at multiple locations within a device can be extracted from measurement data picked up by a single sensor. This means that monitoring using a single sensor may be sufficient to detect the occurrence of partial discharge events within a device.

[0017] The mere presence of partial discharge contributions can serve as a basis for triggering maintenance work. In particular, the new appearance of specific partial discharge contributions can be used as an indication of equipment degradation. For example, if measurement data from high-voltage transmission line monitoring previously only contained partial discharge contributions from unavoidable, harmless corona discharges, the presence of strong new partial discharge contributions may indicate that the solid insulator has begun to fail. Therefore, maintenance actions can be initiated on the monitored equipment in response to the appearance of new partial discharge contributions.

[0018] In practical examples, measurement data can be acquired at sampling rates higher than 10*c / D, more preferably higher than 100*c / D, and most preferably higher than 1000*c / D. Here, c is the propagation speed of the measured signal / attribute (such as sound waves or electromagnetic waves), and D is the geometric distance determining the propagation from the fault point to the sensor used. For example, D may correspond to the housing size of the device, or the distance from the sensor used to the farthest part of the device where partial discharge may occur. The sampling time interval for measurement data should be at least T = D / c, preferably T = 10*D / c. In particular, the time interval T can be longer than 20 ms, and preferably longer than 100 ms.

[0019] This method can be used to determine the time and relative intensity of partial discharge events with extremely high temporal resolution on the order of 1–10 µs. The output can be generated sequentially according to the location of the partial discharge event, thereby allowing the determination of multiple combinations of time and intensity.

[0020] Optionally, a "fingerprint" (i.e., the impulse response at the location of each identified partial discharge event) can be determined. This can be stored in a database using some other method to determine whether a partial discharge event at a particular location is: a) newly occurring, b) existing, or c) eliminated after a maintenance visit. For existing cases (case b), an intensity trend can be calculated to determine whether the equipment is in a stable state or is deteriorating.

[0021] The resolution of partial discharge events can be far superior to traditional methods (such as peak detection algorithms). For example, events with apparent charge below 100 pC have been shown to be distinguishable in typical medium-voltage applications. Even higher sensitivity is expected in high-voltage applications. The distinguishable intensity is as low as 20% of the strongest partial discharge events, and often even as low as 10%.

[0022] In a particularly advantageous embodiment, the presence of at least one partial discharge contribution is determined in response to identifying a substantially common pulse pattern, or a portion thereof, that is statistically distributed at different times in the measurement signal. That is, the presence of recurring pulses in the measurement data can be checked without prior knowledge of the specific pulse pattern. This is achieved using a so-called "detector." As previously described, once a certain pulse pattern is identified as recurring (an indicator factor), that pulse pattern can be learned (as a transfer function). In one example, a substantially common pulse pattern can be determined if the time signals of the pulses are highly correlated with each other and / or have high pattern consistency. However, recurring patterns can also be found in other ways.

[0023] That is, in another particularly advantageous embodiment, the presence of a consistent or substantially common pulse pattern is determined by calculating at least one correlation between measurement data or a portion thereof within a certain time period and a version of measurement data or a portion thereof shifted to another time period. Multiple combinations of time periods and time shifts can be tested.

[0024] In another particularly advantageous embodiment, the transfer function of the partial discharge pulse signal propagating to at least one sensor is determined by at least one partial discharge contribution. This facilitates the spatial localization of the fault point (e.g., within a cabinet or other enclosure). That is, even with only a single sensor present, some estimation of the fault point is possible. This saves time in detailed fault tracing.

[0025] In another particularly advantageous embodiment, the pulse shape of at least one partial discharge pulse can be determined, at least in part, based on the transfer function. The pulse shape provides clues about the specific physical nature of the failure, which can further assist in fault location and more accurate problem identification, thereby accelerating repairs. Determining the pulse shape is especially advantageous if the pulse contains electromagnetic waves (e.g., UHF signals).

[0026] In another particularly advantageous embodiment, the determination of the transfer function is based at least in part on prior knowledge about the shape of the partial discharge pulse. That is, if such prior knowledge exists, it is easier to separate the original pulse shape from the correction of that pulse shape by the transfer function. For example, the pulse can be assumed to be a Delta function pulse. However, as mentioned earlier, the simple detection of a partial discharge event does not depend on any such prior knowledge because the pulse shape is self-learning.

[0027] In particular, the starting point and / or selection of the relevant time periods considered for the study timeframe and / or scale can be determined at least partially based on prior knowledge. Detection is faster and more accurate if the content being sought is known to at least some extent. Some examples of time scales have been given above.

[0028] In another particularly advantageous embodiment, the transfer function is determined based on a propagation model that incorporates the reflection of a signal or wave (such as a sound wave) on a wall surface and / or the diffraction and / or deflection of the signal on an obstacle and / or the attenuation of the signal due to its propagation through a medium and / or the distance traveled. That is, if the geometry of the device (such as the arrangement of busbars and insulators within the enclosure) is known, at least an approximate understanding of the transfer function can be obtained. The clearer the understanding of the transfer function, the more accurate the determination of the presence of partial discharge, and the more precise the location of the fault point.

[0029] In another particularly advantageous embodiment, determining the source-specific signature of a partial discharge pulse from a single source in the measurement data comprises: coherently averaging the signatures of multiple partial discharge pulses from the same source in the measurement data. In this manner, statistical variations in the pulses are eliminated by averaging, and the signature better represents the true characteristics of the partial discharge transfer function. The improved signature can then be used as an estimate of the true transfer function.

[0030] In another particularly advantageous embodiment, determining at least one partial discharge contribution includes:

[0031] - Obtain the signature generated by at least one partial discharge pulse in the measurement data; and

[0032] - Search the measurement data for one or more instances of the signature.

[0033] The initial signature can be obtained from any suitable source. Specifically, if a signature has been self-learned from the measurement data, further instances of that signature can be searched to obtain the complete partial discharge contribution of a specific location to the measurement data. However, signatures can also be determined by prior knowledge or by methods operating in a "detector" manner. An example is the "robust detector," which will be discussed later. Obtaining a signature once and then searching for more instances saves considerable effort, as signature acquisition can be costly.

[0034] Specifically, the time period for searching for instances of one or more signature occurrences can be determined at least in part based on the signature and / or the location of that signature in the measurement data indicated by a "detector" (e.g., a "robust detector" as described later). In this way, self-learned knowledge (and / or prior knowledge) is immediately used to search for further instances of signature occurrence, thereby expanding a single signature to the complete partial discharge contribution of a given location to the measurement data.

[0035] In another particularly advantageous embodiment, in addition to the measurement data and / or signature, the frequency representation and / or time-frequency representation of the measurement data may be examined alternatively or in combination to determine the presence of at least one partial discharge contribution. That is, the measurement data is examined in the frequency domain. For example, this representation can be obtained using Fourier transform, short-time Fourier transform, wavelet transform, similar function analysis or integral transform, and / or spectrograms. Examining this representation means searching for partial discharge contributions in the manner described above. The signature of certain partial discharge events may be more apparent in the frequency domain than in the time domain.

[0036] In another particularly advantageous embodiment, determining whether the measurement data contains at least one partial discharge contribution includes detecting the signal energy and / or power introduced into the air or space by a partial discharge event and reaching the detection unit. Expressing the signal in terms of energy and / or power facilitates the calculation of gain and loss in decibels (dB). Furthermore, the partial discharge event may be associated with a positive introduction of energy and / or power.

[0037] Specifically, the detection of energy or power may include: evaluating the power spectral density of the measured data within a predetermined or automatically adjusted frequency range. This helps to detect the positive introduction of power.

[0038] That is, in another particularly advantageous embodiment, the detection of energy or power may include:

[0039] - Calculate the root mean square value of the introduced power and / or the average value of the energy from the measurement data;

[0040] - Calculate the expected signal attenuation due to wall reflection, inverse distance law, or medium attenuation;

[0041] - Remove or correct for the effect of this attenuation from the root mean square value to obtain the corrected power; and

[0042] - In response to determining that the corrected power as a function of time contains at least one positive change, it is determined that the positive change corresponds to at least one partial discharge pulse.

[0043] This means that a simple reflection of a signal on a wall surface will not be misinterpreted as a partial discharge event at a specific location.

[0044] The determination and evaluation of frequency representation, as well as the detection of signal energy and / or power, can be performed using analog electronic components such as bandpass filters. This has the advantage of saving computation time.

[0045] In another particularly advantageous embodiment, the precise moment of occurrence of the partial discharge event, or the precise phase angle relative to the AC voltage, is determined from measurement data and a transfer function known in any way (e.g., known a priori, or determined during the execution of this method). This allows the moment to be determined with an accuracy better than 1 ms, preferably better than 1 µs. This can be preferably or advantageously used to provide a phase-resolved partial discharge (PRPD) profile, which is crucial for analyzing the type of partial discharge and thus assessing its severity.

[0046] In another particularly advantageous embodiment, the (relative) intensity of the partial discharge is determined by at least one contribution from the partial discharge. This allows for the assessment of the severity of the partial discharge. In particular, it can be determined whether partial discharge events at a particular location maintain the same intensity or whether their intensity increases due to the worsening of the fault that caused them.

[0047] In another particularly advantageous embodiment, determining the partial discharge intensity includes:

[0048] - Obtain calibration information based on at least one signature of partial discharge; and

[0049] - The intensity of the partial discharge is calculated based at least in part on the intensity contributed by the partial discharge and calibration information.

[0050] That is, the signature indicates the strength of the contribution of the partial discharge to the measurement data, given the intensity of the partial discharge event in units of transferred charge.

[0051] In another particularly advantageous embodiment, multiple partial discharge contributions are determined from measurement data. In this way, partial discharge events occurring at different locations can be distinguished from each other. As previously mentioned, this helps to differentiate, for example, harmless corona discharges from discharges that indicate impending catastrophic insulation failure.

[0052] In another particularly advantageous embodiment, in response to determining that the measurement data contains at least one partial discharge contribution,

[0053] - Remove the partial discharge contribution from the measurement data to obtain the remaining measurement data for analysis; and

[0054] - Determine whether the remaining measurement data to be analyzed contains at least one further contribution from partial discharge.

[0055] That is, the measurement data can be analyzed iteratively. In particular, strong partial discharge contributions can be removed from the measurement data first, which may reveal weaker partial discharge contributions. It would be much more difficult to directly attempt to separate partial discharge contributions with large differences in intensity in a single step.

[0056] In another particularly advantageous embodiment, the evaluation is performed by the transfer function:

[0057] - The distance between the location of the partial discharge event and at least one sensor, and / or

[0058] - The path between the location of the partial discharge event and at least one sensor, and / or

[0059] - Does the partial discharge event occur at only one location or at multiple locations?

[0060] As mentioned earlier, this information helps to locate the fault that causes partial discharge and to assess the severity of the fault.

[0061] Therefore, in another particularly advantageous embodiment, the method further includes determining a severity rating of the partial discharge pulse for normal equipment operation, based at least in part on the transfer function and / or pulse shape and / or the temporal or phase distribution of partial discharge events at that location. In particular, this severity rating can be used to prioritize maintenance actions. For example, equipment closer to catastrophic insulation failure is more urgently required to undergo maintenance.

[0062] In another particularly advantageous embodiment, the severity rating depends at least on whether at least one partial discharge contribution is present throughout the entire operation of the device or appears at a later point in the operation. As previously mentioned, a partial discharge contribution appearing at a later point in time indicates that the device health is deteriorating.

[0063] In another particularly advantageous embodiment, the measurement is based on acoustic signals or measurement principles. A single acoustic sensor may be sufficient to monitor the entire switch cabinet, or even multiple cabinets.

[0064] As an alternative to or in combination with the above methods, measurements can be based on electromagnetic wave signals or measurement principles, such as ultra-high frequency (UHF). The advantage of electromagnetic wave signals is that they can detect internal partial discharges because electromagnetic waves can propagate in both solid and gaseous substances.

[0065] Since the present method may be implemented entirely or at least partially by a computer, the present method may be embodied in the form of software. Therefore, the present invention further relates to a computer program comprising machine-readable instructions which, when executed on one or more computers and / or computing instances, cause the one or more computers and / or computing instances to perform the above method. Examples of computing instances include virtual machines, containers or serverless execution environments in the cloud. The present invention further relates to a machine-readable data carrier and / or a download product having the computer program. A download product is a digital product having the computer program, for example, which can be sold in an online store for immediate fulfillment and download to one or more computers. The present invention further relates to one or more computers and / or computing instances having the computer program, and / or computers and / or computing instances having the machine-readable data carrier and / or the download product.

[0066] An exemplary implementation of the method is as follows:

[0067] In the first step, band-pass filtering is performed on the input signal, and the time-resolved average sound power within the time window T1 in the frequency band BP1 is determined. Typically, 100µs < T1 < 20ms is adopted. The band-pass selects an ultrasonic frequency band from 10kHz to 400kHz. This is to take into account the short temporal structure of the generated acoustic signal and reduce the sensitivity of the sensor to other acoustic events within the housing. The average sound power is calculated <bp1>time derivative This is used by the detector. A positive value of the derivative exceeding a certain threshold indicates that the acoustic signal of a partial discharge (PD) event may have reached the sensor. Average acoustic power. <bp1>The averaging time should be appropriately chosen to cover the complete first reflection of the acoustic pulse received from the wall, so that each PD event has only one peak of derivative D, while still being able to distinguish consecutive PD events. This detector has been used to report PD activity within the enclosure because the signal is roughly proportional to the PD discharge intensity. However, the correlation between the signal amplitude and PD intensity (such as apparent charge) is poor. More seriously, the temporal resolution of this method is limited by the average window width, which itself is limited by the first reflection interval and thus by the enclosure size. Therefore, a temporal resolution better than 0.5 ms is extremely difficult to achieve. Furthermore, weaker PD events are difficult to detect, especially those immediately following stronger partial discharges.

[0068] The second step utilizes the fact that the PD generates repeatable acoustic pulses at the source (i.e., the time shape of the local sound pressure level). The shape of the acoustic signal at the receiver is the convolution of the sensor impulse response, the housing impulse response at a specific source position S, and the source impulse response at the receiver position M.

[0069] For a given PD source S, the algorithm identifies and separates the impulse response or "signature" corresponding to each PD source without requiring an additional reference signal. In this identification process, the algorithm first searches for the signature only at the time position of the positive signal displayed by the robust detector in step 1 (corresponding to a strong signal that is very likely to belong to the PD signal).

[0070] Unlike detectors described in existing technologies that search for fixed patterns, this algorithm does not rely on fixed pre-computed patterns but instead identifies signatures (such as acoustic signatures) from signals. Therefore, it is agnostic and adaptive to variations in the physical location and size of the housing, as well as different microphone characteristics and positions.

[0071] The signature is then further post-processed, specifically normalized, to facilitate searching within the signal. The search is typically performed by multiplying the desired signal shape by the signal at different time delays and searching for matches corresponding to the high values ​​of the sum of the products.

[0072] The method of this search step is designed to report not only the relative time of PD occurrence in the (acoustic) signal, but also the relative amplitude of PD.

[0073] The algorithm is unique in that it can determine the relative time of each PD event to an accuracy far below 100µs. When the acoustic propagation time between the sound source and the microphone is known by other means, the absolute time of the event can be determined. However, since the PD sources are from the same source, the relative time of each source is often sufficient. Accurate timing is crucial for phase-resolved analysis, even when the absolute phase cannot be determined.

[0074] Another unique feature of this algorithm is its ability to distinguish between PDs generated at different locations S1, S2, etc., within the casing. This is because the pattern of the reflected signal (such as sound waves) is a function of both the source and sensor positions. The algorithm subsequently needs to be extended to identify not just one, but multiple typical signal forms.

[0075] The sensitivity of this method can be enhanced by properly arranging the sensors relative to possible PD sources and ensuring that each expected relevant PD location presents a unique (acoustic) signature, thus distinguishing them from one another.

[0076] In one possible embodiment of the invention, the algorithm reports the complete signature or some kind of unique identifier for each PD source, as well as the time and intensity of the corresponding PD occurring at positions S1, S2, ...

[0077] In another specific embodiment of the invention, the signature of each source is multiplied by a suitably calculated relative amplitude to completely remove the sound received from position S1 from the signal. In the resulting differential signal, the sound from the second PD position S2 can be identified; otherwise, these sounds would be masked by the large amplitude of the acoustic signal from S1.

[0078] Therefore, understanding the acoustic impulse response (signature) of each PD source is crucial and beneficial for high-sensitivity PD detection systems.

[0079] In laboratory tests using only one microphone and real geometry containing obstacles that block the direct wave and first reflected wave of the acoustic path between the source and the sensor, the algorithm has successfully detected three types of partial discharges with apparent PD intensity lower limits better than 100 pC: corona PD, surface PD, and levitation PD.

[0080] In another specific embodiment of the invention, the signature of each source and the amplitude of its corresponding occurrence frequency are analyzed to calculate the relative PD intensity equivalent to the apparent charge (IEC 60270). This analysis may include calculating the peak amplitude, root mean square (RMS) amplitude, time peak interval, and envelope of the signature.

[0081] The main advantage of this method is that it uses only a single sensor, but it is obviously easy to extend to multiple sensors, and its accuracy or sensitivity can be further improved. When using multiple sensors, the time delay from the source to each sensor is generally different; therefore, the algorithm for detecting the presence of potential partial discharges and determining the transfer function needs to be extended, especially in terms of the time period being analyzed, to accommodate this situation. Attached Figure Description

[0082] The invention will now be described with reference to the accompanying drawings, which are not intended to limit the scope of the invention. The drawings show:

[0083] Figure 1 An exemplary embodiment of the method 100 for detecting partial discharge is shown;

[0084] Figure 2 Exemplary propagation of partial discharge acoustic pulses in enclosures with and without obstructions is illustrated;

[0085] Figure 3 Exemplary acoustic measurement data and the partial discharge contribution evaluated therefrom are shown 4;

[0086] Figure 4 Exemplary signatures of partial discharge events from a specific location are shown in both obstructed and unobstructed acoustic propagation paths. Specific Implementation

[0087] Figure 1 This is a schematic flowchart of one embodiment of a method 100 for detecting partial discharge in medium- and high-voltage equipment 1. Method 100 begins with the equipment 1 being monitored by at least one sensor 2 and measurement data 3 being recorded.

[0088] In step 110, it is determined whether the measurement data 3 contains at least one partial discharge contribution 4, which is:

[0089] - It is reasonably determined that the partial discharge pulse occurring at some point in device 1 and the resulting signal being transmitted to at least one sensor 2, and

[0090] - It appears repeatedly in a statistical manner.

[0091] Specifically, according to block 111, it is determined whether multiple pulses have a consistent pulse pattern imparted by a signal propagating from its source or transmitted to at least one sensor 2. If this is the case (truth value 1), then according to block 112, it is determined that the pulse is a partial discharge pulse from the same source.

[0092] According to box 113, it can be determined whether the fundamentally common pulse pattern, or a portion thereof, is statistically distributed at different times in the measured signal. If this is the case (truth value 1), then according to box 114, it can be determined that at least one partial discharge contribution 4 exists.

[0093] According to box 115, determining the signature of a partial discharge pulse from a source in the measurement data 3 may include: coherently averaging the signatures of multiple partial discharge pulses from the same source in the measurement data 3.

[0094] According to box 116, a signature generated by at least one partial discharge pulse in measurement data 3 can be obtained. According to box 117, one or more instances of this signature can be searched in the measurement data.

[0095] According to box 118, determining whether the measurement data 3 of 110 contains at least one partial discharge contribution 4 may include: detecting signal energy and / or power introduced into the air or space by a partial discharge event or absorbed by the detected unit.

[0096] Specifically, according to box 118a, the detection of energy or power may include: evaluating the power spectral density of measurement data 3 within a predetermined or automatically adjusted frequency range.

[0097] According to box 118b, the root mean square (RMS) value of the introduced power and / or the average energy can be calculated from the measurement data 3. According to box 118c, the expected signal attenuation due to wall reflection can then be calculated. According to box 118d, the effect of this attenuation can be removed or corrected from the RMS value. This yields the corrected power. According to box 118e, it can then be determined whether the corrected power contains at least one positive change as a function of time. If so (truth value 1), then according to box 118f, it can be determined that this positive change corresponds to at least one partial discharge pulse.

[0098] If at least one partial discharge contribution 4 (true value 1) is determined in step 110, then in step 120 it is determined that a partial discharge is occurring in device 1.

[0099] In step 130, the transfer function 5 for the partial discharge pulse signal to propagate to at least one sensor 2 can be determined by at least one partial discharge contribution 4.

[0100] According to box 131, the determination of the transfer function 5 can be performed at least in part based on prior knowledge about the shape of the partial discharge pulse. In particular, according to box 131a, the selection of the starting point and / or the relevant time considered for the length and / or scale of the time period under study can be determined at least in part based on prior knowledge.

[0101] According to box 132, the transfer function 5 can be determined based on a propagation model that includes the reflection of (acoustic) signals or waves on the wall and / or the diffraction and / or deflection of signals on obstacles and / or the attenuation of signals as they travel through the medium and / or the distance traveled.

[0102] In step 140, the pulse shape 6 of at least one partial discharge pulse may be determined at least in part based on the transfer function 5.

[0103] In step 150, the precise moment 7a of the partial discharge event or the precise phase angle 7b relative to the AC voltage can be determined from the measurement data 3 and the known transfer function 5.

[0104] In step 160, the (relative) intensity 9 of the partial discharge can be determined by at least one partial discharge contribution 4.

[0105] According to box 161, calibration information can be obtained based on at least one signature of the partial discharge. According to box 162, the intensity 9 of the partial discharge can then be calculated at least in part based on the intensity contributed by the partial discharge and the calibration information.

[0106] In step 170, identified partial discharge contributions 4 can be removed from the measurement data. This produces remaining measurement data 3' to be analyzed. In step 180, it can then be determined whether the remaining measurement data 3' to be analyzed contains at least one further partial discharge contribution 4. That is, the remaining measurement data 3' to be analyzed can be fed back to step 110 to identify more, possibly weaker, partial discharge contributions 4.

[0107] In step 190, the result can be evaluated using transfer function 5:

[0108] - The distance between the location of the partial discharge event and at least one sensor 2, and / or

[0109] - The path between the location of the partial discharge event and at least one sensor 2, and / or

[0110] - Whether the partial discharge event occurs at only one location or at multiple locations.

[0111] In step 200, the severity rating 8 of the partial discharge pulse in affecting the normal operation of device 1 may be determined, at least in part, based on the transfer function 5 and / or the pulse shape and the temporal or phase distribution of the partial discharge event at that location.

[0112] Specifically, according to box 201, the severity rating 8 may depend at least on whether at least one partial discharge contribution 4 is present throughout the entire operation of the device 1 or appears at a later point in the operation.

[0113] Figure 2 This illustrates how a Delta function shape pulse from an internal source S of device 1 is transformed into acoustic measurement data 3 according to the geometry of device 1.

[0114] Figure 2 Figure a illustrates a simplified scenario where source S and at least one sensor 2 are within direct line of sight. In this way, the direct signal D can propagate from source S to sensor 2 in a straight line. Simultaneously, the pulse will also propagate through reflections R1, ..., R2. N Propagation occurs along indirect paths.

[0115] Figure 2 b shows the amplitude A of the measured signal 3. The direct signal D is attenuated compared to the original pulse S. The reflected signals R1, ..., R... N The attenuation is greater, and the arrival time varies, depending on the length of each propagation path.

[0116] Figure 2 c illustrates a slightly more complex scenario where there is an additional obstacle O within device 1 inside the housing. This means that the direct signal D is diffracted by obstacle O.

[0117] Figure 2 Figure d illustrates the effect of this diffraction on the amplitude A of the measured signal. (Compared to...) Figure 2 Compared to b, the direct signal D attenuates more, and is therefore weaker than the reflected signal.

[0118] Figure 3 The acoustic power P, derived from acoustic measurement data 3, is shown to determine the partial discharge contribution 4 and its signature. This acoustic power P is the signal used for the "robust detector." As a reference, Figure 3 The diagram also shows an electrical signal E, whose pulse indicates a partial discharge event. The pulse in the electrical signal E generates a delayed pulse in the acoustic power P, the delay being the time required for the signal to propagate from the fault point to sensor 2.

[0119] Figure 3 At least three statistically recurring partial discharge contributions 4, 4', and 4'' are visible. These partial discharge contributions 4, 4', and 4'' can be easily identified from the acoustic power P using a robust detector.

[0120] Figure 4 The markers left by partial discharge events at specific locations in measurement data 3 are shown (represented by signatures and / or transfer functions). The amplitude A of measurement signal 3 is plotted over time, averaged over more than 20 partial discharge events. Figure 4 a corresponds to Figure 2 The geometry of device 1 is shown in figure a (with no obstruction O between source S and sensor 2), while Figure 4 b corresponds to Figure 2 c shows the geometry of device 1 (with an obstacle O between source S and sensor 2). Different geometries exhibit different signatures. The signature is delayed by an acoustic propagation time Δt relative to the partial discharge event.

[0121] List of reference numerals

[0122] 1 Equipment

[0123] 2 Sensors

[0124] 3 Measurement Data

[0125] 3' Remaining measurement data to be analyzed

[0126] 4, 4', 4'' Partial discharge contribution

[0127] 5. Transfer Function

[0128] 6. Pulse shape of partial discharge pulse

[0129] 7a Timing of partial discharge pulse

[0130] 7b AC phase timing of partial discharge pulse

[0131] 8. Severity rating of partial discharge events

[0132] 9. Intensity of partial discharge events

[0133] 100 Methods for detecting partial discharge

[0134] 110 Determine if partial discharge contributes.

[0135] 111 Determine whether the pulses have a consistent pulse shape.

[0136] 112. The pulse was determined to be a partial discharge pulse.

[0137] 113. The common pulse morphology is determined to be statistically distributed.

[0138] 113a Calculate correlation

[0139] 114. Partial discharge contribution confirmed.

[0140] 115. Signature determined by coherent averaging

[0141] 116 Obtain the signature of the partial discharge pulse

[0142] 117 Search for signatures in measurement data 3

[0143] 117a Determining search parameters based on signature

[0144] 118 Detect signal energy and / or power

[0145] 118a Evaluation of power spectral density

[0146] 118b Calculate the root mean square value

[0147] 118c Calculation of expected decay

[0148] 118d Removal or correction of attenuation effect

[0149] 118e Determines whether a positive change exists.

[0150] 118f determines that a positive change corresponds to a partial discharge pulse.

[0151] 120 Confirms partial discharge is occurring.

[0152] 130 Determine the transfer function 5

[0153] 131. Utilizing prior knowledge about pulse shape

[0154] 131a Setting the starting point of parameters based on prior knowledge

[0155] 132 Using the propagation / diffraction / deflection / attenuation model

[0156] 140 Determining the pulse shape based on the transfer function 5 6

[0157] 160 Determining the relative intensity of partial discharge events 9

[0158] 161 Determine calibration information

[0159] 162. Determine the intensity using calibration information. 9

[0160] 170 Remove partial discharge contribution 4 from measurement data 3

[0161] 180 Analysis of the partial discharge contribution of the remaining measurement data 3' 4

[0162] 190 Further evaluation of the transfer function 5

[0163] 200 Determine the severity rating of a partial discharge event 8

[0164] 201 utilizes the later occurrence of the event in severity rating 8.

[0165] A signal amplitude

[0166] D Direct signal

[0167] O Obstacles

[0168] R1-R N Reflected signal

[0169] S Source of partial discharge events

[0170] t time

[0171] Δt Acoustic propagation time

Claims

1. A method (100) for detecting partial discharge in a medium- or high-voltage device (1) based on measurement data (3) recorded by at least one sensor (2), the method (100) comprising the following steps: - Determine whether the (110) measurement data (3) contains at least one partial discharge contribution (4), which contribution - It can be reasonably determined that the partial discharge pulse occurred at a location in the device (1) and was generated by propagating the generated signal to the at least one sensor (2), and - Repeated occurrences in a statistical manner, Determining whether a contribution can be reasonably identified as a partial discharge contribution (4) includes: in response to determining (111) that the plurality of pulses have a consistent pulse pattern determined by the propagation or transmission of a signal from its source to the at least one sensor (2), determining (112) that the pulses are partial discharge pulses from the same source; and - In the presence of at least one such partial discharge contribution (4), it is determined that a partial discharge is occurring in device (1) (120).

2. The method (100) according to claim 1, wherein, In response to determining (113) that the basic common pulse pattern or a portion thereof is statistically distributed at different times in the measured signal, it is determined (114) that the at least one partial discharge contribution (4) exists.

3. The method (100) according to any one of claims 1 and 2, wherein, The existence of a consistent or substantially common pulse pattern (113a) is determined by calculating at least one correlation between a measurement data (3) or a portion thereof within a specific time period and a version of the measurement data (3) or a portion thereof that has been time-shifted to another time period.

4. The method (100) according to any one of claims 1 to 3, further comprising: The transfer function (5) that propagates the partial discharge pulse signal to the at least one sensor (2) is determined by at least one partial discharge contribution (4).

5. The method (100) according to claim 4, further comprising: The pulse shape (6) of at least one partial discharge pulse is determined (140) based at least in part on the transfer function (5).

6. The method (100) according to claim 4 or 5, wherein, The determination of the transfer function (5) is performed at least in part based on prior knowledge about the shape of the partial discharge pulse (131).

7. The method (100) according to claim 6, wherein, The starting point of the length and / or scale of the time period under study and / or the selection of the relevant time periods under consideration are determined at least in part based on prior knowledge (131a).

8. The method (100) according to any one of claims 4 to 7, wherein, The determination of the transfer function (5) (130) is based on the (132) propagation model, which includes the reflection of the signal or wave on the wall, and / or the diffraction and / or deflection of the signal on the obstacle, and / or the attenuation of the signal due to the medium through which it propagates and / or the propagation distance.

9. The method (100) according to any one of claims 1 to 8, wherein, Determining the signature of a partial discharge pulse from a source in the measurement data (3) includes: coherent averaging (115) of the signatures of multiple partial discharge pulses from the same source in the measurement data (3).

10. The method (100) according to any one of claims 1 to 9, wherein, Determining (110) at least one partial discharge contribution (4) includes: - Obtain (116) the signature generated in the measurement data (3) by at least one partial discharge pulse; and - Search in the measurement data (3) for one or more occurrences of the signature (117).

11. The method (100) according to claim 10, wherein, The time period used to search for one or more occurrence instances of the signature is determined at least in part based on the signature and / or the location of the signature in the measurement data (3) (117a).

12. The method (100) according to any one of claims 1 to 11, wherein, Instead of the measurement data (3), examine the frequency representation and / or time-frequency representation of the measurement data (3) in conjunction with the measurement data (3) to determine whether there is at least one partial discharge contribution (4).

13. The method (100) according to any one of claims 1 to 12, wherein, Determining (110) whether the measurement data (3) contains at least one partial discharge contribution (4) includes: detecting (118) the energy and / or power of the signal introduced into the air or space by a partial discharge event and reaching the detection unit.

14. The method (100) according to claim 13, wherein, The detection of energy or power (118) includes: evaluating (118a) the power spectral density of the measurement data (3) within a predetermined or automatically adjusted frequency range.

15. The method (100) according to claim 13 or 14, wherein, The detection of energy or power (118) includes: - Calculate the root mean square value of the power introduced by (118b) and / or the average value of the energy from the measurement data (3); - Calculate (118c) the expected signal attenuation due to wall reflection, inverse distance law, or attenuation caused by the medium; - Remove or correct for the effect of this attenuation (118d) from the root mean square value to obtain the corrected power; and - In response to determining (118e) that the corrected power contains at least one positive change as a function of time, it is determined (118f) that the positive change corresponds to at least one partial discharge pulse.

16. The method (100) according to any one of claims 1 to 15, further comprising: The precise moment (7a) or precise phase angle (7b) of the partial discharge event is determined (150) by the measured data (3) and the known transfer function (5).

17. The method (100) according to any one of claims 1 to 16, further comprising: The relative intensity (9) of the partial discharge is determined by the contribution (4) of the at least one partial discharge.

18. The method (100) according to claim 17, wherein, Determining the intensity of partial discharge includes: - Obtain calibration information (161) based on at least one signature of partial discharge; and - The intensity of the partial discharge (9) is calculated (162) based at least in part on the intensity contributed by the partial discharge and calibration information.

19. The method (100) according to any one of claims 1 to 18, wherein, Multiple partial discharge contributions (4) are determined from the measurement data (3).

20. The method (100) according to claim 19, wherein, In response to determining that the measurement data (3) contains at least one partial discharge contribution (4). - Remove (170) this partial discharge contribution (4) from the measurement data (3) to obtain the remaining measurement data (3') to be analyzed; and - Determine (180) whether the remaining measurement data (3') to be analyzed contains at least one additional partial discharge contribution (4).

21. The method (100) according to any one of claims 1 to 20, further comprising: Evaluation of (190) by transfer function (5): - The distance between the location of the partial discharge event and the at least one sensor (2), and / or - The path between the location of the partial discharge event and the at least one sensor (2), and / or - Does the partial discharge event occur at only one location or at multiple locations? 22. The method (100) according to any one of claims 1 to 21, further comprising: The severity rating (8) of the partial discharge pulse in relation to the normal operation of the device (1) is determined (200) based at least in part on the transfer function (5) and / or pulse shape and / or the temporal or phase distribution of the partial discharge event from that location.

23. The method (100) according to claim 22, wherein, The severity rating (8) depends at least on whether at least one partial discharge contribution (4) is present throughout the entire operation of the device (1) or occurs at a later point in the operation.

24. The method (100) according to any one of claims 1 to 23, wherein, The measurement is based on acoustic signals or measurement principles.

25. The method (100) according to any one of claims 1 to 24, wherein, Measurements are based on electromagnetic wave signals or measurement principles, such as ultra-high frequency (UHF).

26. A computer program comprising machine-readable instructions that, when executed on one or more computers and / or computing instances, cause the one or more computers and / or computing instances to perform the method (100) according to any one of claims 1 to 25.

27. A non-transitory machine-readable data carrier and / or downloadable product having the computer program according to claim 26.

28. One or more computers and / or computing instances having a computer program according to claim 26, and / or having a non-transitory machine-readable data carrier and / or download product according to claim 27.

Images