Partial discharge signal extraction method and device for active protection device of transformer

Abstract

The invention discloses a partial discharge signal extraction method and device for a transformer active protection device, and the method comprises the steps: collecting a partial electric signal of a transformer, and marking a preliminary suspicious moment if the partial electric signal is greater than an electric signal threshold value; reading back local electric signals at a plurality of sampling moments before the preliminary suspicious moment, acquiring local electric signals at a plurality of sampling moments after the preliminary suspicious moment, and performing Fourier transform on the local electric signals one by one to obtain frequency spectrum signals; selecting signal amplitudes of characteristic frequency points in the frequency spectrum signal, and comparing the signal amplitudes with an amplitude threshold one by one; and if a plurality of signal amplitudes are simultaneously greater than the amplitude threshold, marking the preliminary suspicious moment as a height suspicious moment, and reporting the height suspicious moment to the active protection device. By adopting the technical scheme, the requirements on the operand and the storage capacity are effectively reduced, and the real-time operation on the transformer can be realized.

Description

Technical Field

[0001] This invention relates to the field of transformer control technology, and in particular to a method and apparatus for extracting partial discharge signals for active protection devices of transformers. Background Technology

[0002] Ultra-high voltage (UHV) transformers are core equipment in the new energy supply and consumption system, and key components for long-distance power transmission and voltage transformation. Statistical data shows that the average annual failure rate of transformers during peak periods is 0.1766 times per 100 units, with an even higher average annual defect rate.

[0003] The electrical insulation condition of a transformer directly determines the reliability of its operation. Partial discharge refers to pulsed discharges occurring in small, localized areas within or on the surface of the transformer's insulation medium due to factors such as excessively high electric field strength. The presence of partial discharge accelerates the insulation degradation process, eventually leading to the collapse of the insulation system. Therefore, studying and analyzing the characteristics of partial discharge signals generated by transformers is of great significance for in-depth research into the mechanisms of partial discharge in transformers, the laws governing insulation degradation, and for fault identification and active protection research of transformers.

[0004] Partial discharge signals are non-stationary signals, and their statistics (such as amplitude and power spectrum) are time-varying functions. Traditional time-domain and frequency-domain analyses in signal analysis and processing cannot effectively analyze such signals. Time-domain analysis methods study the signal's variation over time but cannot obtain frequency-domain information; frequency-domain analysis methods often use Fourier transform for spectral analysis, but the Fourier transform is a global transform and cannot reveal the signal's spectrum variation over time. Current technologies typically combine time and frequency domains for comprehensive analysis of partial discharges, but this often requires large amounts of computation and storage, making it unsuitable for low-cost, low-power embedded platforms with relatively limited storage and computational capabilities, thus hindering real-time operation on transformers. Summary of the Invention

[0005] Purpose of the invention: The present invention provides a method and apparatus for extracting partial discharge signals for active protection devices of transformers, aiming to address the technical problem that the extraction of partial discharge signals in the prior art requires high computational and storage capacity and is difficult to operate in real time on transformers.

[0006] Technical Solution: This invention provides a method for extracting partial discharge signals for an active protection device of a transformer, comprising: acquiring partial electrical signals of the transformer; if the acquired partial electrical signal is greater than an electrical signal threshold, marking the sampling time of the corresponding partial electrical signal as a preliminary suspicious time; performing a front-end readback step, including: reading back the partial electrical signals of multiple sampling times before the preliminary suspicious time, and performing Fourier transform on each of the front-end partial electrical signals to obtain multiple front-end spectral signals; and performing a back-end acquisition step, including: acquiring the partial electrical signals of multiple sampling times after the preliminary suspicious time, and performing Fourier transform on each of the back-end partial electrical signals to obtain multiple front-end spectral signals. Perform a Fourier transform to obtain multiple downstream spectrum signals; perform a upstream comparison step, including: for each upstream spectrum signal, select the upstream signal amplitude corresponding to a characteristic frequency point, and compare each upstream signal amplitude with an amplitude threshold; perform a downstream comparison step, including: for each downstream spectrum signal, select the downstream signal amplitude corresponding to a characteristic frequency point, and compare each downstream signal amplitude with an amplitude threshold; if multiple signal amplitudes are simultaneously greater than the amplitude threshold, mark the initial suspicious moment as a highly suspicious moment, determine that the highly suspicious moment is the moment when the partial discharge signal occurs, and report the highly suspicious moment to the active protection device.

[0007] Specifically, a window function is used to truncate the first segment of the local electrical signal along the time axis, and the truncated first segment of the local electrical signal is then subjected to a Fourier transform; similarly, a window function is used to truncate the second segment of the local electrical signal along the time axis, and the truncated second segment of the local electrical signal is then subjected to a Fourier transform.

[0008] Specifically, the local electrical signals of multiple sampling times within the first sampling period before the initial suspicious moment are read back, and the local electrical signals of multiple sampling times within the second sampling period after the initial suspicious moment are collected.

[0009] Specifically, the sum of the first sampling period and the second sampling period is a fixed total sampling period.

[0010] Specifically, the front-end readback step and the back-end acquisition step are executed synchronously in two threads.

[0011] Specifically, the front-end comparison step and the back-end comparison step are executed synchronously in two threads.

[0012] Specifically, if the amplitude of the preceding signal and the amplitude of the following signal are both greater than the amplitude threshold, the initially suspicious moment will be marked as a highly suspicious moment.

[0013] Specifically, if at least two highly suspicious moments received by the active protection device from multiple acquisition and control units point to the same partial discharge moment, the active protection device determines the partial discharge moment as the occurrence moment of the partial discharge signal.

[0014] Specifically, after the active protection device determines the partial discharge time as the occurrence time of the partial discharge signal, it filters the local electrical signals collected within the sampling period after the partial discharge time, extracts the corresponding feature values, and reports the extracted feature values ​​to the active protection device.

[0015] This invention also provides a partial discharge signal extraction device for an active protection device of a transformer, comprising: a sensor unit and a data acquisition and control unit, wherein: the sensor unit is used to acquire partial electrical signals of the transformer; the data acquisition and control unit is used to mark the sampling time of the corresponding partial electrical signal as a preliminary suspicious time if the acquired partial electrical signal is greater than an electrical signal threshold; to perform a front-end readback step, including: reading back the partial electrical signals of multiple sampling times before the preliminary suspicious time, and performing Fourier transform on each of the front-end partial electrical signals to obtain multiple front-end spectral signals; and to perform a back-end acquisition step, including: acquiring the partial electrical signals of multiple sampling times after the preliminary suspicious time. The electrical signal is subjected to Fourier transform on each of the subsequent local electrical signals to obtain multiple subsequent spectrum signals. A preceding comparison step is performed, including: for each preceding spectrum signal, selecting the preceding signal amplitude corresponding to a characteristic frequency point and comparing each preceding signal amplitude with an amplitude threshold. A subsequent comparison step is performed, including: for each subsequent spectrum signal, selecting the subsequent signal amplitude corresponding to a characteristic frequency point and comparing each subsequent signal amplitude with an amplitude threshold. If multiple signal amplitudes are simultaneously greater than the amplitude threshold, the initially suspicious moment is marked as a highly suspicious moment, and the highly suspicious moment is determined to be the moment when the partial discharge signal occurs. The highly suspicious moment is then reported to the active protection device.

[0016] Beneficial effects: Compared with the prior art, the present invention has the following significant advantages: combining time domain and frequency domain to perform dual threshold discrimination of local electrical signals, effectively distinguishing noise and partial discharge signals, and improving the accuracy of extraction results; effectively reducing the requirements for computation and storage capacity, enabling real-time operation on transformers; dual-thread synchronous calculation, effectively shortening the calculation delay. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating the partial discharge signal extraction method for an active protection device for transformers provided by the present invention. Figure 2 A schematic diagram comparing the latency of dual-thread synchronous computation provided by this invention with that of one-time data computation in the prior art; Figure 3 A schematic diagram illustrating the calculation results of the Fourier transform of the local electrical signal at multiple sampling times provided by the present invention; Figure 4This is a schematic diagram of the partial discharge signal extraction device for active protection of transformers provided by the present invention. Detailed Implementation

[0018] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0019] See Figure 1 This is a flowchart illustrating the partial discharge signal extraction method for active protection devices of transformers provided by the present invention.

[0020] In this embodiment of the invention, a local electrical signal of the transformer is acquired. If the acquired local electrical signal X... n If the signal value exceeds the electrical signal threshold (which can be set according to the actual application scenario), then the corresponding local electrical signal X will be... n The sampling time is marked as the initial suspicious time t'.

[0021] In this embodiment of the invention, the front-end readback step includes: reading back multiple sampling times (t') before the initial suspicious time t'. first The local electrical signal (X) to t') n-p X n-p+1 X n-p+2 , ..., X n-1 (where p is a fixed constant), perform Fourier transform on each of the preceding local electrical signals to obtain multiple preceding spectral signals; execute the subsequent acquisition steps, including: acquiring multiple sampling times (t' to t) after the initial suspicious time. last Local electrical signal (X) n+1 X n+2 , ..., X n+q-1 X n+q (where q is a fixed constant), perform Fourier transform on each of the local electrical signals in the latter segment to obtain multiple spectral signals in the latter segment.

[0022] See Figure 3 The figure shows a schematic diagram of the calculation results of the Fourier transform of the local electrical signal at multiple sampling times provided by the present invention. The window function is used to truncate the signal.

[0023] In this embodiment of the invention, a window function is used to truncate the local electrical signal of the preceding (following) segment along the time axis one by one, and the truncated local electrical signal of the preceding (following) segment is subjected to FFT (Fast Fourier Transform).

[0024] In this embodiment of the invention, local electrical signals from multiple sampling times within a first sampling period prior to the initial suspicious moment are read back, and local electrical signals from multiple sampling times within a second sampling period after the initial suspicious moment are collected.

[0025] In this embodiment of the invention, the sum of the first sampling period and the second sampling period is a fixed total sampling period.

[0026] In practice, the local electrical signal is read back before the initial suspicious moment and the local electrical signal is collected after the initial suspicious moment. This is generally carried out within a preset time period and time length (e.g., the first sampling period and the second sampling period). For example, the readback is performed 100ms before the initial suspicious moment t' and the collection is performed 200ms after the initial suspicious moment t'.

[0027] In practice, the collected local electrical signals are compared with the electrical signal threshold to perform time-domain analysis. It is initially determined that at the sampling time when the local electrical signal is greater than the electrical signal threshold, there is partial discharge. In order to further determine whether it is noise or an actual partial discharge signal, a subsequent Fourier transform is performed to execute frequency-domain analysis.

[0028] In practice, the first (second) spectral signal is obtained through fast Fourier transform. The spectral signal can reflect the superposition of frequencies and is a curve related to frequency points and amplitude.

[0029] In this embodiment of the invention, the front-end comparison step includes: for the front-end spectrum signal, selecting the front-end signal amplitude corresponding to the characteristic frequency point, and comparing the front-end signal amplitude one by one with the amplitude threshold; the back-end comparison step includes: for the back-end spectrum signal, selecting the back-end signal amplitude corresponding to the characteristic frequency point, and comparing the back-end signal amplitude one by one with the amplitude threshold.

[0030] In practice, the spectrum signal is a curve related to frequency points and amplitude. Among them, the characteristic frequency points are the prominent frequency points in the spectrum signal, which can effectively characterize the features of the spectrum signal. Each spectrum signal can often extract multiple characteristic frequency points. The signal amplitude corresponding to the characteristic frequency point is compared with the amplitude threshold (which can be set according to the actual application scenario) for frequency domain analysis.

[0031] In practice, the comparison method is usually as follows: the signal amplitudes of multiple characteristic frequency points of the first pre-segment (post-segment) spectrum signal are compared one by one with the amplitude threshold. Then the second pre-segment (post-segment) spectrum signal, the third pre-segment (post-segment) spectrum signal, and so on, until multiple signal amplitudes are simultaneously greater than the amplitude threshold (for example, the amplitudes of the pre-segment signal and the post-segment signal are simultaneously greater than the amplitude threshold), which means the comparison ends when the highly suspicious moment is met, or when the signal amplitudes of all characteristic frequency points are not greater than the amplitude threshold. At this point, it can be determined that the local electrical signal at the initial suspicious moment is noise.

[0032] See Figure 2This invention provides a schematic diagram comparing the latency of dual-thread synchronous computation with that of one-time data computation in the prior art.

[0033] In this embodiment of the invention, the front-end readback step and the back-end acquisition step are executed synchronously in two threads.

[0034] In this embodiment of the invention, the front-end comparison step and the back-end comparison step are executed synchronously in two threads.

[0035] In practice, the sampling rate of the acquisition device is 100 Msps. Each time, the sampled values ​​are taken from the 100ms before and 200ms after the initial suspicious time t'. If the FFT calculation is performed on these 300ms sampled values ​​all at once, the number of operations is 3 × 10^3 times. 7 ×log2(3×10 7 If a window function is used to truncate the sampled value by 1ms each time for time-frequency calculation, then the total number of operations for 300 windows within 300ms is 300 × 10^3 times; 5 ×log2(10 5 This significantly reduces the computational load of the algorithm.

[0036] In practice, after the initial suspicious moment t', the acquisition and control unit will simultaneously calculate the sampled values ​​for a period of time before and after that moment. The dual-thread synchronous calculation and the method of collecting data while calculating can effectively shorten the calculation delay.

[0037] In this embodiment of the invention, if multiple signal amplitudes are simultaneously greater than the amplitude threshold, the initial suspicious moment is marked as a highly suspicious moment, the highly suspicious moment is determined to be the moment when the partial discharge signal occurs, and the highly suspicious moment is reported to the active protection device.

[0038] In this embodiment of the invention, if the amplitude of the preceding signal and the amplitude of the following signal are both greater than the amplitude threshold, the initial suspicious moment is marked as the highly suspicious moment t''.

[0039] In practice, if the amplitude of the preceding signal and the amplitude of the following signal are both greater than the amplitude threshold, it indicates that after frequency domain analysis, the moment initially suspected is the moment when the partial discharge signal occurs. The method of dual threshold discrimination of time domain and frequency domain characteristic frequency points can effectively distinguish noise and partial discharge signals. Therefore, this moment can be marked as a highly suspicious moment and reported to the active protection device.

[0040] In this embodiment of the invention, after a highly suspicious moment is reported to the active protection device, if at least two highly suspicious moments t'' received by the active protection device from multiple acquisition and control units point to the same partial discharge moment, the active protection device determines the partial discharge moment as the occurrence moment of the partial discharge signal.

[0041] In specific implementation, if at least two highly suspicious moments point to the same partial discharge moment, specifically, if at least two highly suspicious moments are the same time point, for example, both highly suspicious moments are at 327ms, then the two highly suspicious moments are not the same partial discharge moment, for example, the first highly suspicious moment is at 320ms and the second highly suspicious moment is at 327ms.

[0042] See Figure 4 This is a schematic diagram of the partial discharge signal extraction device for transformer active protection provided by the present invention.

[0043] In practice, the acquisition of local electrical signals, time-frequency calculation, and amplitude comparison can all be performed by the acquisition control unit. There can be multiple acquisition control units on the transformer to perform the tasks of acquiring local electrical signals, time-frequency calculation, and amplitude comparison respectively.

[0044] In practical implementation, the transformer active protection device receives data reported by multiple acquisition and control units to comprehensively determine the timing of partial discharge signals, which can avoid misjudgment when a single sensor is interfered with or malfunctions.

[0045] In practical implementation, the method provided by this invention has low requirements for the performance and storage capacity of the main processor of the active protection device. The processing of signals collected by each sensor and the identification of partial discharge signals are arranged in each sensor and the acquisition and control unit, reducing the requirements for the real-time computing power and centralized data storage capacity of the main processor of the transformer active protection device; the host of the active protection device only receives the data at the moment of partial discharge signal occurrence and the processed signal characteristic values, and the compressed data communication method reduces the real-time data transmission bandwidth requirements of the sensor acquisition and control unit and the host of the protection device, enabling real-time operation on the transformer.

[0046] In practice, the active protection device can also combine the PRPD (Phase Resolved Partial Discharge) spectrum to comprehensively determine multiple highly suspicious moments received.

[0047] In this embodiment of the invention, after the active protection device determines the partial discharge time as the occurrence time of the partial discharge signal, it filters the local electrical signal collected within the sampling period after the partial discharge time (the time length of this sampling period is usually fixed, for example, the second sampling period mentioned above can be used) to filter out white noise and narrowband interference signals, thereby improving the signal-to-noise ratio of the original signal; it extracts feature values ​​from the filtered local electrical signal and reports the extracted feature values ​​to the active protection device.

[0048] This invention also provides a partial discharge signal extraction device for an active protection device of a transformer, comprising: a sensor unit and a data acquisition and control unit, wherein: the sensor unit is used to acquire partial electrical signals of the transformer; the data acquisition and control unit is used to mark the sampling time of the corresponding partial electrical signal as a preliminary suspicious time if the acquired partial electrical signal is greater than an electrical signal threshold; to perform a front-end readback step, including: reading back the partial electrical signals of multiple sampling times before the preliminary suspicious time, and performing Fourier transform on each of the front-end partial electrical signals to obtain multiple front-end spectral signals; and to perform a back-end acquisition step, including: acquiring the partial electrical signals of multiple sampling times after the preliminary suspicious time. The electrical signal is subjected to Fourier transform on each of the subsequent local electrical signals to obtain multiple subsequent spectrum signals. A preceding comparison step is performed, including: for each preceding spectrum signal, selecting the preceding signal amplitude corresponding to a characteristic frequency point and comparing each preceding signal amplitude with an amplitude threshold. A subsequent comparison step is performed, including: for each subsequent spectrum signal, selecting the subsequent signal amplitude corresponding to a characteristic frequency point and comparing each subsequent signal amplitude with an amplitude threshold. If multiple signal amplitudes are simultaneously greater than the amplitude threshold, the initially suspicious moment is marked as a highly suspicious moment, and the highly suspicious moment is determined to be the moment when the partial discharge signal occurs. The highly suspicious moment is then reported to the active protection device.

[0049] The partial discharge signal extraction device for active protection devices of transformers provided by the present invention, wherein the execution method, steps or functions performed by the execution unit can refer to the partial discharge signal extraction method for active protection devices of transformers provided by the present invention.

Claims

1. A method for extracting partial discharge signals for active protection devices of transformers, characterized in that, include: The local electrical signal of the transformer is collected. If the collected local electrical signal is greater than the electrical signal threshold, the sampling time of the corresponding local electrical signal is marked as a preliminary suspicious time. The process of performing the front-end readback step includes: reading back the local electrical signals at multiple sampling times before the initial suspicious time, and performing Fourier transform on each of the front-end local electrical signals to obtain multiple front-end spectral signals; the process of performing the back-end acquisition step includes: acquiring the local electrical signals at multiple sampling times after the initial suspicious time, and performing Fourier transform on each of the back-end local electrical signals to obtain multiple back-end spectral signals. The front-end comparison step includes: for the front-end spectrum signal, selecting the front-end signal amplitude corresponding to the characteristic frequency point, and comparing the front-end signal amplitude one by one with the amplitude threshold; the back-end comparison step includes: for the back-end spectrum signal, selecting the back-end signal amplitude corresponding to the characteristic frequency point, and comparing the back-end signal amplitude one by one with the amplitude threshold. If multiple signal amplitudes are simultaneously greater than the amplitude threshold, the initial suspicious moment is marked as a highly suspicious moment. The highly suspicious moment is determined to be the moment when the partial discharge signal occurs, and the highly suspicious moment is reported to the active protection device.

2. The method for extracting partial discharge signals for active protection devices of transformers according to claim 1, characterized in that, The step of performing Fourier transforms on each of the preceding local electrical signals includes: Using a window function, the preceding local electrical signal is truncated one by one along the time axis, and the truncated preceding local electrical signal is then subjected to a Fourier transform. The step of performing Fourier transforms on each of the subsequent local electrical signals includes: Using a window function, the local electrical signal in the latter part is truncated one by one along the time axis, and the truncated local electrical signal in the latter part is then subjected to Fourier transform.

3. The method for extracting partial discharge signals for an active protection device of a transformer according to claim 1, characterized in that, Reread the local electrical signals of multiple sampling times within the first sampling period before the initial suspicious time, and collect the local electrical signals of multiple sampling times within the second sampling period after the initial suspicious time.

4. The method for extracting partial discharge signals for active protection devices of transformers according to claim 3, characterized in that, The sum of the first sampling period and the second sampling period is a fixed total sampling period.

5. The method for extracting partial discharge signals for an active protection device of a transformer according to claim 1, characterized in that, The front-end readback step and the back-end acquisition step are executed synchronously in two threads.

6. The method for extracting partial discharge signals for an active protection device of a transformer according to claim 5, characterized in that, The front-end comparison step and the back-end comparison step are executed synchronously in two threads.

7. The method for extracting partial discharge signals for an active protection device of a transformer according to claim 1, characterized in that, If multiple signal amplitudes simultaneously exceed an amplitude threshold, the initial suspicious moment is marked as a highly suspicious moment, including: If the amplitude of the preceding signal and the amplitude of the following signal are both greater than the amplitude threshold, then the initial suspicious moment is marked as a highly suspicious moment.

8. The method for extracting partial discharge signals for an active protection device of a transformer according to claim 1, characterized in that, The process of reporting highly suspicious moments to the active protection device, followed by: If, among the multiple highly suspicious moments received by the active protection device from multiple acquisition and control units, at least two highly suspicious moments point to the same partial discharge moment, the active protection device determines the partial discharge moment as the moment when the partial discharge signal occurs.

9. The method for extracting partial discharge signals for an active protection device of a transformer according to claim 8, characterized in that, The process of reporting highly suspicious moments to the active protection device, followed by: After the active protection device determines the partial discharge time as the occurrence time of the partial discharge signal, it filters the local electrical signals collected within the sampling period after the partial discharge time, extracts the corresponding feature values, and reports the extracted feature values ​​to the active protection device.

10. A partial discharge signal extraction device for an active protection device of a transformer, characterized in that, include: Sensor unit and data acquisition and control unit, wherein: The sensor unit is used to collect local electrical signals from the transformer; The acquisition and control unit is used to mark the sampling time of the corresponding local electrical signal as a preliminary suspicious time if the acquired local electrical signal is greater than the electrical signal threshold. The process of performing the front-end readback step includes: reading back the local electrical signals at multiple sampling times before the initial suspicious time, and performing Fourier transform on each of the front-end local electrical signals to obtain multiple front-end spectral signals; the process of performing the back-end acquisition step includes: acquiring the local electrical signals at multiple sampling times after the initial suspicious time, and performing Fourier transform on each of the back-end local electrical signals to obtain multiple back-end spectral signals. The front-end comparison step includes: for the front-end spectrum signal, selecting the front-end signal amplitude corresponding to the characteristic frequency point, and comparing the front-end signal amplitude one by one with the amplitude threshold; the back-end comparison step includes: for the back-end spectrum signal, selecting the back-end signal amplitude corresponding to the characteristic frequency point, and comparing the back-end signal amplitude one by one with the amplitude threshold. If multiple signal amplitudes are simultaneously greater than the amplitude threshold, the initial suspicious moment is marked as a highly suspicious moment. The highly suspicious moment is determined to be the moment when the partial discharge signal occurs, and the highly suspicious moment is reported to the active protection device.

Images