Hydro-generator partial discharge pulse adaptive extraction method, device and medium

Abstract

The present application relates to a kind of water turbine generator partial discharge pulse adaptive extraction method, equipment and medium, the method is by from the data of one power frequency cycle in the partial discharge signal of acquisition and preprocessing, respectively calculate pulse rough positioning threshold and slice energy threshold, utilize pulse rough positioning threshold when traversing signal preliminary mark pulse occurrence position, again with the position as starting point based on slice energy threshold by sliding window method accurately locate the start and end of pulse, finally according to pulse duration, reject interference pulse, the method does not need artificial configuration parameter, the double threshold value obtained by adaptive calculation can effectively overcome the fuzzy influence of strong background noise to pulse edge, accurately locate the start and end of partial discharge pulse, solve the problem that existing method is difficult to accurately locate pulse edge in strong noise environment, parameter setting relies on prior knowledge and poor adaptability.

Description

Technical Field

[0001] This invention relates to the field of partial discharge detection in generators, specifically to an adaptive extraction method, equipment, and medium for partial discharge pulses in hydro-generators. Background Technology

[0002] Partial discharge monitoring detects partial discharge pulses within the generator stator windings to obtain information on insulation defects and faults. Combined with pattern recognition and fault diagnosis techniques, it enables fault diagnosis and accident prevention in stator winding insulation, making it a crucial means to improve the condition monitoring level and safety stability of hydro-generators. Partial discharge is closely related to the insulation degradation process; it is both a significant cause of insulation degradation and a key characteristic reflecting it. Online measurement of partial discharge under actual operating stresses not only allows for timely monitoring of insulation health, effective prevention of insulation accidents, and provision of reference for condition-based maintenance, but also identifies the impact characteristics of various operating stresses on insulation health, detects events that threaten insulation, and provides a basis for fault diagnosis.

[0003] Extensive research has been conducted both domestically and internationally on generator partial discharge monitoring, yielding fruitful application results. However, with the deepening of engineering practice, problems such as incomplete monitoring by partial discharge monitoring systems and low accuracy in partial discharge pulse extraction have become increasingly prominent, posing new challenges to the partial discharge monitoring project.

[0004] Due to the presence of background noise, the signals acquired by online partial discharge monitoring contain a large amount of noise, making the edges of individual partial discharge pulse signals very indistinct, which brings difficulties to the subsequent analysis and identification of partial discharge pulse signal characteristics. Therefore, determining the edges of partial discharge pulse signals and accurately extracting the partial discharge pulse waveform from background noise has become one of the key issues in partial discharge detection technology for hydro-generators. The extraction of partial discharge pulse signals, as a crucial step in the preprocessing stage of partial discharge measurement signals, is the foundation for in-depth analysis of partial discharge signals.

[0005] Currently, there are various methods for analyzing and processing partial discharge signals, such as wavelet thresholding denoising and empirical mode decomposition denoising. Some literature proposes using complex wavelet transform to denoise UHF partial discharge signals, but it is difficult to simultaneously optimize the selection of the complex wavelet, the determination of the wavelet decomposition order, and the selection of the threshold algorithm. Other literature proposes using spectral kurtosis to design an adaptive bandpass Wiener filter for filtering before wavelet smoothing denoising. However, since the parameters of the Wiener filter are fixed, achieving optimal filtering results requires prior knowledge of the statistical characteristics of the signal and noise, which is often difficult to predict in practical engineering. Summary of the Invention

[0006] This invention provides an adaptive extraction method, device, and medium for partial discharge pulses from a hydro-generator. The purpose is to solve the problems of existing partial discharge pulse extraction methods, such as difficulty in accurately locating pulse edges under strong background noise, reliance on prior knowledge for parameter settings, and poor adaptability.

[0007] To achieve the above objectives, the first aspect of the present invention provides an adaptive extraction method for partial discharge pulses in a hydro-generator, comprising the following steps: Acquire partial discharge signals at the neutral point of the hydro-generator; The acquired partial discharge signals are preprocessed to obtain preprocessed partial discharge signals; Extract a power frequency cycle from the preprocessed partial discharge signal and calculate the pulse coarse localization threshold and slice energy threshold; The preprocessed partial discharge signals are traversed, the pulse occurrence position is determined according to the pulse coarse localization threshold, and the pulse start and end positions are determined according to the slice energy threshold. The pulse duration is calculated based on the pulse start and end positions. Interference pulses are eliminated, and the remaining pulses are identified as partial discharge pulses.

[0008] Furthermore, the method for calculating the pulse coarse localization threshold by extracting a power frequency cycle from the preprocessed partial discharge signal includes: Extract a power frequency cycle from the preprocessed partial discharge signal to obtain the power frequency cycle data segment; The power frequency data segment is evenly divided into multiple sub-segments; Obtain the maximum amplitude of the signal in each segment, sort the maximum amplitudes of all segments in ascending order, and take the average of the first preset maximum values ​​as the reference value. Three times the reference value is determined as the pulse coarse positioning threshold.

[0009] Furthermore, methods for calculating the slice energy threshold by extracting a power frequency cycle from the preprocessed partial discharge signal include: The power frequency frequency data segment is processed using a sliding window method. The sum of squares of the signal amplitude is calculated within each sliding window to obtain a set of slice energy values. Determine the maximum value among all slice energy values, and take half of that maximum value as a reference value; The reference value is divided into multiple equal parts to obtain multiple incremental candidate thresholds; For each candidate threshold, the portion of the slice energy value that is greater than or equal to the candidate threshold and the portion that is less than or equal to the candidate threshold are counted, and the proportion of each part in the total slice energy value and their respective average values ​​are calculated. Calculate the inter-class variance corresponding to each candidate threshold based on the stated proportion and average value; The candidate threshold with the largest inter-class variance is selected as the slice energy threshold.

[0010] Furthermore, the method of traversing the preprocessed partial discharge signals, determining the pulse occurrence location based on the pulse coarse localization threshold, and then determining the pulse start and end locations based on the slice energy threshold includes: The preprocessed partial discharge signal is traversed, and the amplitude of each sampling point is compared with the pulse coarse positioning threshold. If the amplitude of a sampling point is greater than the pulse coarse positioning threshold, then the sampling point is marked as the pulse occurrence location; Starting from the pulse occurrence position, the slice energy within each sliding window is calculated by sliding left and right using a preset window length. When sliding to the left, the starting point of the sliding window where the energy of the first slice is less than the energy threshold of the slice is determined as the pulse start point; When sliding to the right, the starting point of the sliding window where the first slice energy is less than the slice energy threshold is determined as the pulse end point; Record the positions of the pulse start and pulse end points.

[0011] Furthermore, the process of eliminating interference pulses includes: calculating the pulse duration based on the pulse start and end points, and eliminating pulses with a duration greater than 50 microseconds.

[0012] Furthermore, the acquisition of the partial discharge signal at the neutral point of the hydro-generator includes: using a capacitive coupling sensor to sense the partial discharge signal at the neutral point of the hydro-generator, and acquiring the partial discharge signal through an acquisition module with a sampling frequency of not less than 100MHz.

[0013] Furthermore, the step of uniformly dividing the power frequency data segment into multiple sub-segments includes: uniformly dividing the power frequency data segment into more than 2000 sub-segments; the step of taking the first preset maximum value to calculate the average value includes: taking the first 1000 maximum amplitude values ​​after sorting to calculate the average value.

[0014] Furthermore, the sliding window method for processing the power frequency frequency data segment includes: performing sliding window processing on the power frequency frequency data segment with a window length of 20, and calculating the sum of squares of the signal amplitude within each sliding window; dividing the reference value into multiple parts includes: dividing the reference value into no less than ten parts to obtain multiple incremental candidate thresholds.

[0015] To achieve the above objectives, a second aspect of the present invention provides an electronic device including a memory and a processor, the memory being used to store a program that supports the processor in executing the adaptive extraction method for partial discharge pulses of a hydro-generator, and the processor being configured to execute the program stored in the memory.

[0016] To achieve the above objectives, a third aspect of the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, performs the steps of the adaptive extraction method for partial discharge pulses of a hydro-generator.

[0017] The beneficial effects of this invention are: Compared with existing technologies, the present invention provides an adaptive extraction method, device, and medium for partial discharge pulses in hydro-generators. This method extracts a power frequency cycle from the acquired and preprocessed partial discharge signal, calculates a coarse pulse location threshold and a slice energy threshold, uses the coarse pulse location threshold to initially mark the pulse occurrence position while traversing the signal, and then uses this position as the starting point to accurately locate the pulse's start and end points using a sliding window based on the slice energy threshold. Finally, interfering pulses are removed based on the pulse duration. This method does not require manual parameter configuration, and the adaptively calculated dual thresholds effectively overcome the blurring effect of strong background noise on the pulse edge, accurately locating the start and end positions of the partial discharge pulse. It solves the problems of existing methods, such as difficulty in accurately locating pulse edges in strong noise environments, parameter settings relying on prior knowledge, and poor adaptability. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below.

[0019] Figure 1 This is a flowchart of an adaptive extraction method for partial discharge pulses of a hydro-generator disclosed in an embodiment of the present invention.

[0020] Figure 2 This is a diagram showing the result of extracting a standard pulse signal using the method proposed in this invention, as disclosed in an embodiment of the present invention.

[0021] Figure 3 This is a magnified image of a pulse extracted according to an embodiment of the present invention. Detailed Implementation

[0022] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0023] According to embodiments of the present invention, it should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the following methods, in some cases the steps shown or described may be executed in a different order than that shown here.

[0024] like Figure 1 As shown, the present invention provides an adaptive extraction method for partial discharge pulses in hydro-generators, comprising the following steps: Step S100: Acquire the partial discharge signal of the neutral point of the hydro-generator; In this step, the partial discharge signal at the neutral point of the hydro-generator is first detected using a capacitive coupling sensor. The capacitive coupling sensor is a non-invasive measuring device with advantages such as high sensitivity, wide bandwidth, and easy installation.

[0025] The neutral point is the common point when the three-phase windings of a generator are connected in a star configuration. Installing a sensor at this location can simultaneously monitor the partial discharge activity of the three-phase windings. Furthermore, the power frequency voltage at the neutral point is relatively low, which facilitates the safe and stable operation of the sensor.

[0026] The acquisition module uses high-speed data acquisition equipment with a sampling frequency of The sampling rate is set to at least 100MHz to ensure complete capture of the rapid rise edge and duration of partial discharge pulses, avoiding signal distortion. During acquisition, the analog signal output by the sensor is transmitted to the acquisition module via a coaxial cable, where it undergoes analog-to-digital conversion and is stored as a digital signal for subsequent processing. Partial discharge refers to the phenomenon where only a portion of an insulation system discharges without forming a penetrating breakdown. Its pulse signal typically has a rise time on the order of nanoseconds and a duration on the order of microseconds, thus requiring a high sampling rate to accurately reconstruct the waveform characteristics.

[0027] Step S200: Preprocess the acquired partial discharge signal to obtain the preprocessed partial discharge signal; In this step, the acquired raw partial discharge signal is preprocessed to suppress various background noises present in the field environment and improve signal quality, resulting in a preprocessed partial discharge signal, denoted as . .

[0028] Background noise refers to non-discharge interference signals introduced during partial discharge online monitoring by factors such as on-site electromagnetic interference, power electronic device operation, broadcast communication signals, power frequency and its harmonic components. These noises are usually superimposed on the partial discharge pulse signal, resulting in blurred pulse edges and reduced signal-to-noise ratio, which seriously affects the accuracy of subsequent pulse extraction and feature analysis.

[0029] A high-pass filter is a filtering circuit or algorithm that allows high-frequency signals to pass while attenuating or suppressing low-frequency signals. Its working principle is to set a cutoff frequency, significantly attenuating signal components with frequencies below this cutoff frequency, while allowing signal components with frequencies above this cutoff frequency to pass with their original amplitude. Since partial discharge pulse signals have nanosecond-level rise times and microsecond-level pulse widths, their energy is mainly concentrated in the high-frequency band, while field interference such as 50Hz power frequency and its harmonics, and electromechanical vibrations are usually located in the low-frequency band. Therefore, a high-pass filter can effectively filter out low-frequency interference components and retain high-frequency partial discharge pulse information. In practical engineering applications, the cutoff frequency of the high-pass filter can be reasonably selected according to the spectral characteristics of the partial discharge signal and the field noise distribution, usually set between tens of kilohertz and several megahertz, to ensure that the main energy of the pulse signal is not lost while effectively suppressing noise. After high-pass filtering preprocessing, the obtained partial discharge signal... Background noise is significantly suppressed, and the rising edge and peak characteristics of the partial discharge pulse are clearer.

[0030] It should be noted that the preprocessed signal middle Represented by a positive integer, indicating the index of the discrete sampling point. The signal length should meet the requirements of subsequent analysis. The range of values ​​is not less than ,in Sampling frequency, The frequency is the power frequency to ensure that the data length is sufficient to cover multiple power frequency cycles.

[0031] Step S300: Extract a power frequency cycle from the preprocessed partial discharge signal and calculate the pulse coarse positioning threshold and slice energy threshold; In this step, the preprocessed partial discharge signal is first analyzed. A complete power frequency cycle is extracted from the data for subsequent threshold calculation. Power frequency cycle length. Based on sampling frequency and power frequency The decision was made jointly, and the calculation formula is as follows:

[0032] in, This represents the floor function. Set to at least 100MHz. It is 50Hz, therefore This corresponds to the number of sampling points within one power frequency cycle. Extracting data from one power frequency cycle is to statistically reflect the distribution characteristics of partial discharge pulses and background noise, while avoiding excessive computation due to excessively long data.

[0033] Next, the pulse coarse positioning threshold is calculated based on the extracted power frequency cycle data segments. The pulse coarse localization threshold is used to initially identify the approximate location where the pulse occurs. Its calculation process is as follows: First, the length is The data segments are evenly divided into Each segment The value is a positive integer greater than 2000 to ensure that each segment is short enough to capture local maxima; then, the maximum amplitude of the signal within each segment is extracted and denoted as . Then, sort these maximum values ​​in ascending order; take the first 1000 maximum values ​​after sorting (i.e., the smallest 1000 maximum values) and calculate their arithmetic mean, denoted as . This approach is based on the assumption that these smaller maxima originate primarily from background noise rather than partial discharge pulses; finally, Multiply by 3 to get the pulse coarse positioning threshold. ,Right now:

[0034] This threshold is used in subsequent steps to determine whether the signal amplitude may belong to a partial discharge pulse.

[0035] Subsequently, the slice energy threshold was calculated based on the same power frequency frequency data segment. Slice energy reflects the degree of energy concentration of a signal within a local window, enabling more sensitive detection of pulse start and end boundaries. During calculation, the sliding window length is first set. , much smaller In this embodiment Take 20. Use a sliding window approach to traverse the entire data segment, calculating the sum of squares of the signal amplitude within each window to obtain a set of slice energy values, denoted as SliceEnergy, with a length of... .

[0036] Then, find the maximum value in SliceEnergy and take half of it as a reference value. ,Right now:

[0037] Will Divided into equal parts share, The thresholds are positive integers not less than 10, thus yielding a series of increasing candidate thresholds. ,in:

[0038] For each candidate threshold , in SliceEnergy greater than or equal to The data is divided into set C1, less than or equal to The data is divided into set C1 and set C2. The proportion of elements in C1 and C2 to the total number of elements in SliceEnergy is calculated, denoted as [missing information]. and And calculate the mean of C1 and C2, denoted as . and Based on this, the inter-class variance is calculated. Its formula is:

[0039] Inter-class variance is a metric used in pattern recognition to measure the degree of separation between two classes of data. The larger the value, the more likely it is to be expressed as follows: The more significant the difference between high-energy and low-energy data when using a threshold to divide the data, the more effectively the threshold can distinguish between the pulse energy region and the noise energy region. Finally, a threshold is selected that... When the maximum value is reached, the corresponding As the energy threshold of the slice The energy threshold of this slice will be used in subsequent steps to accurately locate the start and end points of the pulse. The threshold determined in this adaptive manner does not require manual setting, can adapt to different on-site noise levels and pulse intensities, and has strong robustness.

[0040] Step S400: Traverse the preprocessed partial discharge signals, determine the pulse occurrence position according to the pulse coarse positioning threshold, and then determine the pulse start position and end position according to the slice energy threshold. Based on the calculated pulse coarse localization threshold and slice energy threshold, the preprocessed partial discharge signal The process involves iterating through the data to achieve preliminary location of the pulse occurrence and precise location of the pulse start and end points.

[0041] First, traverse the preprocessed partial discharge signals. The amplitude of each sampling point is compared with the pulse coarse localization threshold. It should be noted that although step S300 calculates both the pulse coarse localization threshold and the slice energy threshold, during the actual traversal, the existence of a pulse is first determined based on the pulse coarse localization threshold. If the amplitude of a sampling point is greater than the pulse coarse localization threshold, that sampling point is marked as the pulse occurrence location, indicating that this point may be part of a partial discharge pulse. The pulse occurrence location is a rough location marker, indicating the presence of partial discharge activity near the sampling point, but this point itself is not necessarily the true start or end point of the pulse; further precise definition of the pulse boundary is needed.

[0042] After marking the pulse occurrence location, a preset sliding window length is used as the starting point. Perform a sliding search to the left and right respectively, calculating the slice energy within each sliding window. Sliding window length. In this embodiment, the window length used in calculating the slice energy threshold is consistent with that used in step S300. The value is set to 20, meaning each window contains 20 consecutive sampling points. The method for calculating the slice energy is the same as in step S300, which involves summing the squares of the amplitudes of all sampling points within the window to obtain the slice energy value corresponding to that window.

[0043] When sliding to the left, starting from the pulse occurrence position, move one sampling point to the left each time, and calculate the slice energy of each window in turn until the first window with a slice energy less than the slice energy threshold is found. The starting point of this window is then determined as the pulse start point.

[0044] When sliding to the right, starting from the pulse occurrence position, move one sampling point to the right each time, and calculate the slice energy of each window in turn until the first window with a slice energy less than the slice energy threshold is found. The starting point of this window is then determined as the pulse end point.

[0045] This boundary search method based on sliding windows and slice energy can effectively overcome the positioning error caused by amplitude fluctuations at individual sampling points. This is because slice energy reflects the energy accumulation in a local area and has a smoothing effect on noise interference, thus more accurately reflecting the actual start and end positions of the pulse. After determining the pulse start and end points, the sampling point numbers corresponding to these two positions are recorded for use in subsequent steps.

[0046] It should be noted that the determination of the pulse start and end points follows the principle of searching outwards from the pulse origin, ensuring that the extracted pulse waveform completely encompasses the main energy region of the pulse, while avoiding the mistaken inclusion of noise regions far from the pulse within the pulse range. In this way, each marked pulse origin location will eventually correspond to a set of start and end point coordinates, forming a complete partial discharge pulse waveform segment. If a signal boundary (such as a data start or end point) is encountered during the search, the boundary is used as the default pulse start or end point. However, in practical engineering applications, it is usually ensured that the data length is sufficient to cover the complete pulse waveform to avoid boundary truncation.

[0047] Step S500: Calculate the pulse duration based on the pulse start and end positions, eliminate interference pulses, and identify the remaining pulses as partial discharge pulses.

[0048] For each pulse extracted in step S400, the pulse duration is calculated based on its start and end positions. The pulse duration refers to the time elapsed from the pulse's start to its end, and it is calculated by dividing the difference in sampling point indices between the start and end points by the sampling frequency. That is, duration = (end position - start position) / Due to sampling frequency The frequency is typically not less than 100MHz, so the duration of the pulse can be calculated precisely.

[0049] Preferably, after calculating the duration of each pulse, pulses with a duration greater than 50 microseconds are discarded.

[0050] After removing interfering pulses, the remaining pulses are identified as valid partial discharge pulses. These pulses will serve as the basis for subsequent partial discharge characteristic analysis, pattern recognition, and fault diagnosis. It should be noted that this step uses only duration as the criterion for removing interfering pulses. However, in practical applications, combining other characteristic parameters (such as pulse amplitude, rise time, fall time, and spectral characteristics) for comprehensive judgment can further improve the accuracy of pulse identification. Nevertheless, the duration-based screening method is already effective in removing most common interferences, and it is computationally simple, has good real-time performance, and is suitable for high-speed acquisition and real-time processing applications in engineering fields.

[0051] Through the above five steps, this invention realizes a complete process of partial discharge pulse extraction, from raw signal acquisition, preprocessing, adaptive threshold calculation, precise pulse start and end point positioning to interference pulse removal. The entire process requires no manual intervention or parameter configuration and can adaptively adapt to different on-site noise levels and pulse intensity changes, providing an efficient, accurate, and robust pulse extraction method for online monitoring of partial discharge in hydro-generators.

[0052] To verify the effectiveness of the method proposed in this invention in practical applications, a standard pulse signal was used for testing, and the extraction results are as follows: Figure 2 As shown in the figure, the partial discharge pulse waveform extracted from the noisy signal is clearly displayed. It can be seen that the method of the present invention can effectively suppress background noise, completely preserve the main energy region of the pulse, and accurately restore the morphological characteristics of the pulse.

[0053] The extracted pulses were observed under local magnification, such as... Figure 3 As shown in the figure, the red * indicates the extracted pulse signal. The magnified image provides a clearer view of the pulse's start and end points; the pulse's rising and falling edges remain steep, and waveform details are fully preserved, demonstrating the excellent performance of the method in accurately locating pulse boundaries. Figure 2 and Figure 3The results together verify the accuracy and reliability of the method of the present invention in adaptive extraction of partial discharge pulses.

[0054] Compared with existing technologies, the invention lies in its ability to accurately extract partial discharge pulses by adaptively calculating the pulse coarse localization threshold and the slice energy threshold, and combining a two-step strategy of coarse and precise pulse localization. The pulse coarse localization threshold is adaptively determined based on the statistical characteristics of the maximum amplitude of each segment within the power frequency cycle, used to initially identify the pulse's location. The slice energy threshold is adaptively selected through sliding window energy analysis and the inter-class variance maximization criterion, used to precisely define the pulse's start and end points. This scheme does not rely on prior statistical knowledge of the signal and noise, avoiding the difficulties in parameter selection inherent in traditional methods such as complex wavelet transform and the dependence of Wiener filters on prior information. It effectively overcomes the blurring effect of strong background noise on the pulse edges, accurately locates the pulse's start and end points, and features low computational complexity and strong adaptability, making it widely applicable to the engineering extraction of partial discharge signals from different hydro-generators.

[0055] According to another aspect of the embodiments of this application, an electronic device is also provided, including a processor and a memory, wherein the processor is configured to implement the steps of the method when executing a computer program stored in the memory.

[0056] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0057] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0058] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0059] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0060] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. An adaptive extraction method for partial discharge pulses in a hydro-generator, characterized in that, Includes the following steps: Acquire partial discharge signals at the neutral point of the hydro-generator; The acquired partial discharge signals are preprocessed to obtain preprocessed partial discharge signals; Extract a power frequency cycle from the preprocessed partial discharge signal and calculate the pulse coarse localization threshold and slice energy threshold; The preprocessed partial discharge signals are traversed, the pulse occurrence position is determined according to the pulse coarse localization threshold, and the pulse start and end positions are determined according to the slice energy threshold. The pulse duration is calculated based on the pulse start and end positions. Interference pulses are eliminated, and the remaining pulses are identified as partial discharge pulses.

2. The adaptive extraction method for partial discharge pulses of a hydro-generator as described in claim 1, characterized in that, Methods for calculating the pulse coarse localization threshold by extracting a power frequency cycle from the preprocessed partial discharge signal include: Extract a power frequency cycle from the preprocessed partial discharge signal to obtain the power frequency cycle data segment; The power frequency data segment is evenly divided into multiple sub-segments; Obtain the maximum amplitude of the signal in each segment, sort the maximum amplitudes of all segments in ascending order, and take the average of the first preset maximum values ​​as the reference value. Three times the reference value is determined as the pulse coarse positioning threshold.

3. The adaptive extraction method for partial discharge pulses of a hydro-generator as described in claim 2, characterized in that, Methods for calculating the slice energy threshold by extracting a power frequency cycle from the preprocessed partial discharge signal include: The power frequency frequency data segment is processed using a sliding window method. The sum of squares of the signal amplitude is calculated within each sliding window to obtain a set of slice energy values. Determine the maximum value among all slice energy values, and take half of that maximum value as a reference value; The reference value is divided into multiple equal parts to obtain multiple incremental candidate thresholds; For each candidate threshold, the portion of the slice energy value that is greater than or equal to the candidate threshold and the portion that is less than or equal to the candidate threshold are counted, and the proportion of each part in the total slice energy value and their respective average values ​​are calculated. Calculate the inter-class variance corresponding to each candidate threshold based on the stated proportion and average value; The candidate threshold with the largest inter-class variance is selected as the slice energy threshold.

4. The adaptive extraction method for partial discharge pulses of a hydro-generator as described in claim 1, characterized in that, The method for traversing the preprocessed partial discharge signals, determining the pulse occurrence location based on the pulse coarse localization threshold, and then determining the pulse start and end locations based on the slice energy threshold includes: The preprocessed partial discharge signal is traversed, and the amplitude of each sampling point is compared with the pulse coarse positioning threshold. If the amplitude of a sampling point is greater than the pulse coarse positioning threshold, then the sampling point is marked as the pulse occurrence location; Starting from the pulse occurrence position, the slice energy within each sliding window is calculated by sliding left and right using a preset window length. When sliding to the left, the starting point of the sliding window where the energy of the first slice is less than the energy threshold of the slice is determined as the pulse start point; When sliding to the right, the starting point of the sliding window where the first slice energy is less than the slice energy threshold is determined as the pulse end point; Record the positions of the pulse start and pulse end points.

5. The adaptive extraction method for partial discharge pulses of a hydro-generator as described in claim 1, characterized in that, The process of eliminating interference pulses includes: calculating the pulse duration based on the pulse start and end points, and eliminating pulses with a duration greater than 50 microseconds.

6. The adaptive extraction method for partial discharge pulses of a hydro-generator as described in claim 1, characterized in that, The process of acquiring the partial discharge signal of the neutral point of the hydro-generator includes: using a capacitive coupling sensor to sense the partial discharge signal of the neutral point of the hydro-generator, and acquiring the partial discharge signal through an acquisition module with a sampling frequency of not less than 100MHz.

7. The adaptive extraction method for partial discharge pulses of a hydro-generator as described in claim 2, characterized in that, The step of uniformly dividing the power frequency frequency data segment into multiple sub-segments includes: uniformly dividing the power frequency frequency data segment into more than 2000 sub-segments; the step of taking the first preset maximum value to calculate the average value includes: taking the first 1000 maximum amplitude values ​​after sorting to calculate the average value.

8. The adaptive extraction method for partial discharge pulses of a hydro-generator as described in claim 3, characterized in that, Processing the power frequency frequency data segment using a sliding window method includes: performing sliding window processing on the power frequency frequency data segment with a window length of 20, and calculating the sum of squares of the signal amplitude within each sliding window; dividing the reference value into multiple equal parts includes: dividing the reference value into no less than ten equal parts to obtain multiple incremental candidate thresholds.

9. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports the processor in executing the adaptive extraction method for partial discharge pulses of the hydro-generator according to any one of claims 1-8, and the processor is configured to execute the program stored in the memory.

10. A computer-readable storage medium storing a computer program thereon, characterized in that, When the computer program is run by the processor, it executes the steps of the adaptive extraction method for partial discharge pulses of a hydro-generator according to any one of claims 1-8.

Images