Threshold adjustment method and device based on partial discharge data, equipment and medium

Abstract

This application discloses a threshold adjustment method, apparatus, device, and medium based on partial discharge data, relating to the field of partial discharge detection technology. The method includes: acquiring the current spectrum type; setting a first adjustment threshold based on the current spectrum type; acquiring effective pulse data and the corresponding number of phase windows based on the first adjustment threshold; calculating the phase window ratio based on the number of phase windows corresponding to the data; if the phase window ratio exceeds a first ratio threshold, clustering the data to obtain a DBI index; if the DBI index is less than a first index and the number of cluster centers is less than or equal to the first threshold, performing an FFT transformation on the data to obtain the frequency point with the highest energy and calculating the energy ratio of that frequency point; if the energy of that frequency point is not within a first range and the energy ratio is less than or equal to a second ratio threshold, adjusting the first adjustment threshold to obtain a second adjustment threshold. This method effectively improves the detection accuracy of partial discharge data.

Description

Technical Field

[0001] This application relates to the field of partial discharge detection technology, and in particular to a threshold adjustment method, apparatus, device and medium based on partial discharge data. Background Technology

[0002] In the field of partial discharge detection in power equipment, existing technologies have significant limitations. Current mainstream detection methods primarily rely on manual experience to set fixed thresholds or employ adaptive threshold algorithms based on simple signal amplitude statistics. In practical applications, inspectors need to preset trigger thresholds based on personal experience or repeatedly adjust parameters manually during the detection process to accurately capture partial discharge signals.

[0003] Therefore, some testing equipment manufacturers have attempted to introduce adaptive adjustment functions to improve testing efficiency. However, most of these improvements still remain at the level of simple statistical analysis of signal amplitude parameters, lacking a deep understanding of the partial discharge mechanism and signal characteristics. This one-dimensional approach results in insufficient threshold adjustment accuracy and poor performance in dealing with complex and changing field conditions, especially in complex scenarios such as strong electromagnetic interference environments and the coexistence of multiple discharge sources, making it difficult to achieve accurate and reliable signal identification.

[0004] With the increasing demands for intelligent operation and maintenance of power equipment, the limitations of traditional threshold adjustment methods are becoming increasingly apparent. On the one hand, manual intervention in detection heavily relies on the technical skills of operators, making it difficult to guarantee the consistency and comparability of detection results. On the other hand, adaptive algorithms based on simple amplitude statistics cannot effectively distinguish between real discharge signals and environmental noise, easily leading to misjudgments and missed detections. These problems directly restrict the application effectiveness of partial discharge detection technology in condition assessment and fault early warning.

[0005] Therefore, the industry urgently needs to develop a new threshold adjustment method that can achieve adaptive threshold adjustment in complex electromagnetic environments. Summary of the Invention

[0006] This application provides a threshold adjustment method, apparatus, device, and medium based on partial discharge data, which can effectively improve the detection accuracy of partial discharge data.

[0007] To achieve the above objectives, this application adopts the following technical solution:

[0008] In a first aspect, this application provides a threshold adjustment method based on partial discharge data, the method comprising:

[0009] Obtain the spectrum type of the monitored power equipment at time t, and set a first adjustment threshold based on the spectrum type at time t;

[0010] Based on the first adjustment threshold, the first effective pulse data and the number of phase windows corresponding to the first effective pulse data are collected;

[0011] The proportion of the first phase window is calculated based on the number of phase windows corresponding to the first valid pulse data.

[0012] If the proportion of the first phase window exceeds the first proportion threshold, then the first effective pulse data is clustered to obtain the first Davidson-Bolding index and the number of the first cluster centers.

[0013] If the first Davidson-Bolding index is less than the first index, and the number of the first cluster centers is less than or equal to the first threshold, then perform a Fast Fourier Transform (FFT) on the first effective pulse data to obtain the first frequency point with the largest energy in the first effective pulse data, and calculate the first energy percentage of the first frequency point.

[0014] If the energy at the first frequency point is not within the first range, or the proportion of the first energy is less than or equal to the second proportional threshold, then the first adjustment threshold is adjusted to obtain the second adjustment threshold.

[0015] In some possible implementations, the method further includes:

[0016] If the proportion of the first phase window does not exceed the first proportional threshold, the first adjustment threshold is adjusted to obtain a second adjustment threshold; wherein the second adjustment threshold is greater than the first adjustment threshold.

[0017] Based on the second adjustment threshold, the second effective pulse data and the number of phase windows corresponding to the second effective pulse data are collected;

[0018] The proportion of the second phase window is calculated based on the number of phase windows corresponding to the second effective pulse data.

[0019] If the proportion of the second phase window exceeds the first proportion threshold, then the second effective pulse data is clustered.

[0020] In some possible implementations, the method further includes:

[0021] If the first Davidson-Bauin index is greater than or equal to the first index, or the number of first cluster centers is greater than the first threshold, then the first adjustment threshold is adjusted to obtain a second adjustment threshold; wherein the second adjustment threshold is greater than the first adjustment threshold.

[0022] In some possible implementations, the method further includes:

[0023] If the energy at the first frequency point is within the first range and the proportion of the first energy is greater than the second proportional threshold, then the first effective pulse data is determined to be an effective partial discharge signal.

[0024] In some possible implementations, the method further includes:

[0025] If the first valid pulse data is determined to be a valid partial discharge signal, then based on the spectrum type at time t, the kurtosis, skewness, cross-correlation coefficient and discharge coefficient of the spectrum at time t are collected, and the first adjustment threshold is corrected to obtain the target adjustment threshold.

[0026] In some possible implementations, the map type includes partial discharge phase-resolved maps and phase-resolved pulse sequence maps;

[0027] When the spectrum type is a partial discharge phase-resolved spectrum, the first adjustment threshold is set as follows:

[0028] The partial discharge data of the phase window within the corresponding period is cached, and the number of effective pulses in the phase window is counted. If the number of effective pulses exceeds the preset pulse number threshold, the threshold is incremented. The operation is repeated until the number of effective pulses does not exceed the preset pulse number threshold. The incremented threshold is determined as the first adjustment threshold under the partial discharge phase resolution spectrum.

[0029] When the spectrum type is a phase-resolved pulse sequence spectrum, the first adjustment threshold is set as follows:

[0030] Partial discharge data of the phase window is cached in batches within the corresponding period. The number of effective pulses in each batch is counted. If the number of effective pulses exceeds the preset batch pulse number threshold, the threshold is increased. The operation is repeated until the number of effective pulses does not exceed the preset batch pulse number threshold. The increased threshold is determined as the first adjustment threshold under the phase-resolved pulse sequence spectrum.

[0031] Secondly, this application provides a threshold adjustment device based on partial discharge data, the device comprising:

[0032] The acquisition module is used to acquire the spectrum type of the monitored power equipment at time t, and set a first adjustment threshold according to the spectrum type at time t.

[0033] The first determination module is used to collect first effective pulse data and the number of phase windows corresponding to the first effective pulse data based on the first adjustment threshold; calculate the first phase window ratio based on the number of phase windows corresponding to the first effective pulse data; if the first phase window ratio exceeds the first ratio threshold, then perform clustering processing on the first effective pulse data to obtain the first Davidson-Bolding index and the number of first cluster centers.

[0034] The second determination module is used to perform a Fast Fourier Transform (FFT) on the first effective pulse data if the first Davidson-Bolding index is less than the first index and the number of first cluster centers is less than or equal to the first threshold, to obtain the first frequency point with the highest energy in the first effective pulse data, and to calculate the first energy proportion of the first frequency point; if the energy of the first frequency point is not in the first range, or the first energy proportion is less than or equal to the second proportion threshold, then the first adjustment threshold is adjusted to obtain the second adjustment threshold.

[0035] Thirdly, this application provides a computing device, including a memory and a processor;

[0036] The memory stores one or more computer programs, the one or more computer programs including instructions; when the instructions are executed by the processor, the computing device performs the method as described in any one of the first aspects.

[0037] Fourthly, this application provides a computer-readable storage medium for storing a computer program for performing the method as described in any one of the first aspects.

[0038] Fifthly, this application provides a computer program product comprising one or more computer instructions, wherein when the computer instructions are executed by a computer, the computer performs the method as described in any one of the first aspects.

[0039] As can be seen from the above technical solution, this application has at least the following beneficial effects:

[0040] In this application, the spectrum type at time t is obtained, and a first adjustment threshold is set according to the spectrum type at time t; first effective pulse data and the number of phase windows corresponding to the first effective pulse data are collected based on the first adjustment threshold; the proportion of the first phase window is calculated according to the number of phase windows corresponding to the first effective pulse data; if the proportion of the first phase window exceeds the first proportion threshold, the first effective pulse data is clustered to obtain the first Davidson-Bourdin index; if the first Davidson-Bourdin index is less than the first index, the first effective pulse data is subjected to a Fast Fourier Transform (FFT) to obtain the first frequency point with the highest energy in the first effective pulse data, and the first energy proportion of the first frequency point is calculated; if the energy of the first frequency point is not in the first range, or the first energy proportion is less than or equal to the second proportion threshold, the first adjustment threshold is adjusted to obtain the second adjustment threshold.

[0041] In existing technologies, fixed thresholds or threshold adjustments based solely on a single feature (such as pulse count) are often employed. These methods fail to adapt to the differences between various spectrum types and lack sufficient ability to distinguish noise and invalid signals, easily leading to incorrect threshold adjustments. This can result in either missing genuine partial discharge signals or misjudging noise as valid signals. Therefore, this application achieves adaptive threshold adjustment by setting an initial threshold based on spectrum type and then employing a dual criterion of phase window ratio, cluster analysis, and frequency energy analysis. This effectively filters noise and accurately identifies valid partial discharge signals under different devices and spectra, significantly improving the accuracy and reliability of partial discharge detection.

[0042] It should be understood that the descriptions of technical features, technical solutions, beneficial effects, or similar language in this application do not imply that all features and advantages can be achieved in any single embodiment. Rather, it is understood that the description of a feature or beneficial effect means that a specific technical feature, technical solution, or beneficial effect is included in at least one embodiment. Therefore, the descriptions of technical features, technical solutions, or beneficial effects in this specification do not necessarily refer to the same embodiment. Furthermore, the technical features, technical solutions, and beneficial effects described in this embodiment can be combined in any suitable manner. Those skilled in the art will understand that embodiments can be implemented without one or more specific technical features, technical solutions, or beneficial effects of a particular embodiment. In other embodiments, additional technical features and beneficial effects may be identified in specific embodiments that do not embody all embodiments. Attached Figure Description

[0043] Figure 1 A flowchart illustrating a threshold adjustment method based on partial discharge data provided in this application embodiment;

[0044] Figure 2 A schematic diagram of a threshold adjustment device based on partial discharge data provided in an embodiment of this application;

[0045] Figure 3 This is a schematic diagram of a computing device provided in an embodiment of this application. Detailed Implementation

[0046] The terms "first," "second," and "third," etc., used in this application specification and accompanying drawings are used to distinguish different objects, not to limit a specific order.

[0047] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0048] To ensure clarity and conciseness in the description of the following embodiments, a brief introduction to the related technologies is given first:

[0049] In the field of partial discharge detection for power equipment, threshold adjustment is a crucial step in ensuring detection accuracy. Currently, threshold adjustment methods in the industry have several limitations. On the one hand, most technologies employ a fixed threshold strategy. This one-size-fits-all approach completely ignores the structural differences between various power equipment and the signal distribution characteristics of different spectrum types. In practical applications, this either leads to the omission of genuine partial discharge signals due to excessively high thresholds, or the misclassification of a large amount of interference noise as valid signals due to excessively low thresholds. On the other hand, some adaptive adjustments only focus on a single feature dimension, such as adjusting the threshold based solely on the number of pulses, without combining multi-dimensional information such as phase window ratio and frequency energy distribution for comprehensive judgment. This results in a very limited ability to distinguish between valid signals and noise.

[0050] The actual operating environment of power systems is extremely complex. Partial discharge signals are affected by various factors such as electromagnetic interference and equipment operating condition fluctuations, resulting in highly dynamic and diverse signal characteristics. These shortcomings of traditional threshold adjustment methods directly hinder the improvement of the accuracy of partial discharge detection. This not only fails to provide a reliable basis for assessing the insulation status of equipment but may also lead to unnecessary equipment maintenance or allow faults to develop unchecked due to misjudgments, seriously threatening the safe and economical operation of the power system.

[0051] In view of this, embodiments of this application provide a threshold adjustment method based on partial discharge data. In this method, the spectrum type at time t is obtained, and a first adjustment threshold is set according to the spectrum type at time t; first effective pulse data and the number of phase windows corresponding to the first effective pulse data are collected based on the first adjustment threshold; a first phase window proportion is calculated based on the number of phase windows corresponding to the first effective pulse data; if the first phase window proportion exceeds a first proportion threshold, the first effective pulse data is clustered to obtain a first Davidson-Bourdin index; if the first Davidson-Bourdin index is less than a first index, a Fast Fourier Transform (FFT) is performed on the first effective pulse data to obtain a first frequency point with the highest energy in the first effective pulse data, and a first energy proportion of the first frequency point is calculated; if the energy of the first frequency point is not within a first range, or the first energy proportion is less than or equal to a second proportion threshold, the first adjustment threshold is adjusted to obtain a second adjustment threshold. As can be seen, this application achieves adaptive threshold adjustment by setting an initial threshold based on the spectrum type, and then filtering through multiple layers of criteria including phase window ratio, cluster analysis, and frequency energy analysis. This not only effectively filters noise but also accurately identifies effective partial discharge signals under different devices and different spectra, significantly improving the accuracy and reliability of partial discharge detection.

[0052] This method can be executed by a processing device. This processing device can be a terminal or a server. Terminals include, but are not limited to, smartphones, tablets, laptops, personal digital assistants, or smart wearable devices. Servers can be cloud servers, such as central servers in a central cloud computing cluster or edge servers in an edge cloud computing cluster. Alternatively, servers can be located in a local data center. A local data center refers to a data center directly controlled by the user.

[0053] To make the technical solution of this application clearer and easier to understand, a threshold adjustment method based on partial discharge data provided by an embodiment of this application will be described below with reference to the accompanying drawings. Figure 1 As shown in the figure, this is a flowchart of a threshold adjustment method based on partial discharge data provided in an embodiment of this application.

[0054] The method includes:

[0055] S101. The processing device obtains the spectrum type of the monitored power equipment at time t, and sets the first adjustment threshold according to the spectrum type at time t.

[0056] Specifically, the testing personnel remotely send an automatic threshold adjustment command via a mobile phone, which acts as the host computer; the command is received by the core processing unit (ZYNQ chip). Subsequently, the core processing unit controls the Double Data Rate Synchronous Dynamic Random Access Memory (DDR) to cache the original partial discharge data within the current power frequency cycle in a burst transmission mode, providing data support for subsequent threshold calculations.

[0057] In the partial discharge detection system, the ZYNQ chip serves as the core processing unit, forming a heterogeneous brain that works collaboratively. Its Processing System (PS) acts as the command center, responsible for receiving instructions, executing threshold adjustment logic, and making decisions; while its Programmable Logic (PL) acts as a high-speed execution engine, controlling the timing of the DDR memory to achieve high-speed caching and transfer of massive amounts of partial discharge data, thus solving the speed bottleneck of the data flow.

[0058] DDR is a widely used volatile memory characterized by high data transfer rates and low power consumption, suitable for various applications such as computers, servers, and mobile devices. DDR is a memory technology that allows data to be transferred simultaneously on both the rising and falling edges of the clock pulse, thus achieving a higher data transfer rate than traditional Synchronous Dynamic Random-Access Memory (SDRAM). The main function of DDR memory is to synchronize with the CPU frequency, improving data transfer efficiency.

[0059] Next, the core processing unit determines the spectrum type used for partial discharge detection at the current time (time t). The spectrum type is divided into partial discharge phase-resolved spectrum (PRPD) and phase-resolved power supply (PRPS). Then, different logic is used to set the initial first adjustment threshold for different spectrum types.

[0060] The map types include partial discharge phase-resolved maps and phase-resolved pulse sequence maps;

[0061] When the spectrum type is a partial discharge phase-resolved spectrum, the first adjustment threshold is set by caching the partial discharge data of the phase window within the corresponding period, counting the number of effective pulses in the phase window, and increasing the threshold if the number of effective pulses exceeds the preset pulse number threshold. The operation is repeated until the number of effective pulses does not exceed the preset pulse number threshold. The threshold at this time is the first adjustment threshold under the partial discharge phase-resolved spectrum.

[0062] When the spectrum type is a phase-resolved pulse sequence spectrum, the first adjustment threshold is set by buffering partial discharge data of the phase window in batches within the corresponding period, counting the number of effective pulses in each batch of buffers, and increasing the threshold if the number of effective pulses exceeds the preset batch pulse number threshold. This operation is repeated until the number of effective pulses does not exceed the preset batch pulse number threshold. The threshold at this time is the first adjustment threshold under the phase-resolved pulse sequence spectrum.

[0063] In some embodiments, in the partial discharge phase-resolved (PRPD) pattern mode, the double data rate synchronous dynamic random access memory buffers partial discharge data of 360 phase windows within the power frequency cycle during a single burst transmission; it traverses the 360 ​​phase windows and counts the number of effective pulses in each phase window. If the counted number of effective pulses is greater than 300, it is determined that the current threshold is too low and the signal is overwhelmed by noise, and the PS terminal sends a command to increase the threshold by 1mV; the above process is repeated until the number of effective pulses is no greater than 300, and the threshold at this point is the first adjustment threshold in the PRPD mode.

[0064] In Phase-Resolved Pulse Sequence (PRPS) pattern mode, the double data rate synchronous dynamic random access memory is transmitted in 6 bursts within one power frequency cycle, with each burst buffering 60 phase windows of data. Each transmission corresponds to 60 phase windows. The number of valid pulses in a single transmission is counted. If the number of valid pulses is greater than 50, it is determined that there is too much noise, and the threshold is increased by 1mV. The operation is repeated until the number of valid pulses is no greater than 50, and the first adjustment threshold in PRPS mode is determined.

[0065] S102, The processing device collects the first effective pulse data and the number of phase windows corresponding to the first effective pulse data based on the first adjustment threshold.

[0066] In partial discharge detection technology, the phase window is the basic unit for signal analysis and processing. Its definition is based on the periodicity of alternating current (AC). One power frequency cycle represents a complete alternating cycle of positive and negative AC voltage; for example, the cycle in a 50Hz power grid is 20 milliseconds. Within this cycle, the position on the voltage waveform at a given moment is called the phase, usually represented by angles from 0 to 360 degrees. For ease of analysis, a complete 360-degree phase cycle is divided into several equal intervals, each interval being a phase window. The specific division method depends on the detection mode. For example, in PRPD mode, each cycle is divided into 360 phase windows, each 1 degree wide; in PRPS mode, it is divided into 60 phase windows, each 6 degrees wide.

[0067] The role of phase windows is reflected in two aspects. First, statistics and location: by recording the specific phase window in which the discharge pulse occurs, the correlation between discharge activity and voltage phase can be revealed, thereby forming a characteristic spectrum that characterizes the type of insulation defect. Second, noise reduction and identification: based on the characteristic that real partial discharges often occur in specific phase windows, while noise is often randomly distributed in all phase windows, the distribution pattern of pulses in the phase window can effectively distinguish between effective signals and background interference, providing an important basis for subsequent threshold adjustment and fault diagnosis.

[0068] This step is the data acquisition and preliminary statistical stage. The first adjustment threshold is a voltage threshold determined by the system through an initial iterative process (e.g., gradually increasing from 0mV until noise is initially filtered out). Using this threshold as a criterion, the system acquires partial discharge signals in real time, and any pulse with an amplitude exceeding this threshold is recorded as the first valid pulse data. Simultaneously, the system counts the total number of phase windows within a complete power frequency cycle that capture at least one such valid pulse; this number is the corresponding number of phase windows.

[0069] Under the control of the ZYNQ chip, the DDR memory caches raw waveform data for one or more power frequency cycles. The processing device (such as the PS terminal of the ZYNQ) traverses each phase window, checks whether the maximum amplitude of the pulse within it exceeds a first adjustment threshold, and counts all phase windows containing valid pulses.

[0070] This step aims to convert a continuous analog signal into countable discrete data and quantize the spatiotemporal distribution of the signal (its breadth along the phase axis). It is the cornerstone of subsequent data quality assessment because the distribution pattern of effective pulses along the phase axis is a crucial characteristic distinguishing real discharges from random noise.

[0071] Through this step, the system transforms the macroscopic distribution characteristics of the signal into a calculable index, namely the number of phase windows, providing objective and quantitative data input for subsequent intelligent judgment and avoiding the one-sidedness of judging based solely on a single pulse amplitude.

[0072] S103. The processing device calculates the proportion of the first phase window based on the number of phase windows corresponding to the first effective pulse data.

[0073] This step standardizes the statistical results from the previous step and aims to calculate the first phase window percentage. The first phase window percentage refers to the percentage of phase windows containing valid pulses out of the total number of phase windows in one power frequency cycle.

[0074] The processing device performs a simple division operation, dividing the statistically obtained number of phase windows by the total number of phase windows in the system. For example, the number of phase windows is 360 in PRPD mode and 60 in PRPS mode. Then, multiplying by 100% gives the value of the proportion of the first phase window.

[0075] Converting absolute quantities into relative proportions allows for a unified judgment standard applicable to different spectral patterns, as the total number of phase windows varies across these patterns. The proportions more intuitively reflect the sparsity or density of the signal in the phase dimension.

[0076] The normalized proportion data eliminates the system differences caused by different detection modes, making the subsequent fixed proportion threshold judgment universal and comparable, and improving the robustness and adaptability of the algorithm.

[0077] S104. The processing device determines whether the proportion of the first phase window exceeds the first proportion threshold and obtains the first judgment result.

[0078] This step is a logical decision point. The system compares the calculated proportion of the first phase window with a preset first proportion threshold to arrive at a first judgment result of yes or no.

[0079] The purpose of this judgment is to serve as a trigger for whether to initiate computationally intensive clustering analysis. Real partial discharges are typically concentrated in specific phase regions, resulting in a low phase window percentage; while random noise is evenly distributed, leading to a high phase window percentage. In this application, setting a 40% threshold is intended to initiate a more refined analysis when noise is likely to be prevalent, while skipping it when the data appears clean to save computational resources.

[0080] This step enables intelligent data routing within the algorithm. As an efficient pre-screening mechanism, it avoids unnecessary complex clustering operations when data quality is acceptable, saving processor computing resources and improving the efficiency of the entire adjustment process.

[0081] If the first judgment result indicates that the proportion of the first phase window exceeds the first proportion threshold, then S105 is executed; if the first judgment result indicates that the proportion of the first phase window does not exceed the first proportion threshold, then S109 is executed.

[0082] S105. If the first judgment result indicates that the proportion of the first phase window exceeds the first proportion threshold, then the first effective pulse data is clustered to obtain the first Davidson-Bolding index and the number of the first cluster centers.

[0083] If the first judgment result indicates that the proportion of the first phase window exceeds the first proportion threshold, this step initiates a more in-depth data structure analysis. The system uses the K-means clustering algorithm to automatically group the first effective pulse data and identify the inherent clustering patterns in the data. After clustering is completed, the system calculates the first Davidson-Bolding index, which is a quantitative indicator for evaluating the quality of clustering. The smaller the value, the better the clustering result (compact within clusters and separated between clusters).

[0084] K-means clustering is an unsupervised learning algorithm that can automatically discover the inherent grouping structure of data. This algorithm analyzes the distribution pattern of current valid pulse data to determine whether it consists of a few clear, real discharges or a large area of ​​chaotic noise. By automatically clustering data points and outputting the number of clusters and cluster quality (DBI), the algorithm provides the system with objective evidence for whether the threshold needs to be increased, thus achieving a leap from simple amplitude filtering to intelligent pattern recognition.

[0085] The Davies-Bouldin Index (DBI) is a metric used to evaluate the performance of clustering algorithms. It measures clustering effectiveness by comparing the similarity of samples within a cluster with the dissimilarity of samples between clusters. A lower DBI value indicates better clustering, meaning higher separation between clusters. Specifically, the DBI calculation involves the distance between the center of each cluster and samples within that cluster, as well as the distance between centers of different clusters.

[0086] On the ZYNQ's PS terminal or a dedicated hardware accelerator, the K-means clustering algorithm is executed to divide the data points into K clusters based on their similarity. Subsequently, the DBI is calculated according to the definition formula of the DBI, based on the centroid of each cluster and the distribution of data points.

[0087] The aim is to distinguish real discharge signals from noise based on the inherent structure of the data. Real discharges form a few tight, separate clusters, corresponding to different discharge sources or discharge types, while noise is chaotic, either failing to form clear clusters or forming numerous loose clusters. The DBI index can quantify this structural characteristic.

[0088] Moving data quality assessment from window proportion to intrinsic structure significantly enhances the system's ability to identify and resist complex noise. DBI provides an objective and reliable mathematical tool to determine the purity of data at the current threshold, offering a scientific basis for whether the threshold needs to be further increased.

[0089] This application also obtains the number of first cluster centers, which is the number of cluster centers after clustering the first effective pulse data. This number of first cluster centers is an important result obtained by analyzing the first effective pulse data using the K-means clustering algorithm, reflecting the distribution structure characteristics of the data in the feature space. The number of cluster centers directly characterizes the number of potential discharge sources in the dataset and is one of the important indicators for evaluating data quality and signal effectiveness.

[0090] In its implementation, the system performs a K-means clustering algorithm on the first valid pulse data, automatically grouping pulse data points with similar characteristics into several clusters, each cluster corresponding to a cluster center. The number of the first cluster centers is the total number of these cluster centers. This parameter, together with the Davidson-Bolding index, constitutes a dual evaluation criterion for cluster analysis, comprehensively evaluating the clustering effect from different dimensions.

[0091] The purpose of obtaining the number of cluster centers is to further verify the validity of the data. Real partial discharge signals typically originate from a finite discharge source, therefore the number of cluster centers formed in the feature space should be relatively small. An excessive number of cluster centers often indicates the presence of significant noise interference or multiple atypical discharge modes in the data. These noise signals form multiple scattered cluster structures in the feature space, leading to an abnormally high number of cluster centers.

[0092] In practical applications, the system compares the number of first cluster centers with a preset first threshold. When the number of first cluster centers is less than or equal to the first threshold, it indicates that the data distribution conforms to the typical characteristics of a real discharge signal, and the next stage of frequency analysis can proceed. Conversely, if the number of first cluster centers exceeds the first threshold, the current data quality is deemed poor, and the threshold needs to be adjusted for data acquisition and analysis to be repeated.

[0093] This design enables the system to more accurately identify valid discharge signals, avoiding misjudgments caused by an abnormal number of cluster centers. Simultaneously, by combining the Davidson-Bolding index with a comprehensive evaluation, a more complete cluster quality judgment system is formed, improving the accuracy and reliability of threshold adjustment. By introducing the parameter of the number of first cluster centers, the system can better adapt to the detection needs under different operating conditions, ensuring accurate detection results in various complex environments.

[0094] S106. The processing device determines whether the first Davidson-Bolding index is less than the first index and whether the number of the first cluster centers is less than or equal to the first threshold, and obtains the second judgment result.

[0095] The system compares the calculated first Davidson-Bolding index with a preset first indicator to arrive at a second judgment result.

[0096] In some embodiments, the first metric can be set to 0.8, and the number of first cluster centers can be set to 3.

[0097] If the second judgment result indicates that the first Davidson-Bourdin index is less than the first indicator, then execute S107; if the second judgment result indicates that the first Davidson-Bourdin index is greater than or equal to the first indicator, then execute S110.

[0098] S107. If the second judgment result indicates that the first Davidson-Bolding index is less than the first index and the number of the first cluster centers is less than or equal to the first threshold, then perform a fast Fourier transform on the first effective pulse data to obtain the first frequency point with the largest energy in the first effective pulse data, and calculate the first energy proportion of the first frequency point.

[0099] When the first Davidson-Bolding exponent is determined to be less than the first criterion, and the number of the first cluster centers is less than or equal to the first threshold, it indicates that the time-domain distribution pattern of the current data conforms to the true discharge characteristics. This step aims to further verify the validity of the signal from the frequency domain perspective. Specifically, the system performs a Fast Fourier Transform (FFT) on the first valid pulse data that passes the clustering criteria, transforming it from the time domain to the frequency domain, thereby obtaining a frequency distribution spectrum. Subsequently, the system locates the frequency component with the strongest energy in this spectrum, namely the first frequency point, and calculates the percentage of the energy of the first frequency point in the total energy of the entire signal. The first energy percentage is set as F%.

[0100] On the PS side of the ZYNQ chip or on a dedicated hardware accelerator (PL side), the Fast Fourier Transform (FFT) calculation module is invoked to process the time-domain waveform of the first valid pulse data. After the calculation is completed, the system scans the entire spectrum to find the spectral line with the highest amplitude (representing energy), and its corresponding frequency is the first frequency point. Then, the system calculates the ratio of the energy value of the first frequency point to the total energy value of the signal, obtaining the first energy percentage F%.

[0101] Different electrical devices, due to differences in their physical structure and insulation characteristics, generate partial discharge electromagnetic waves with inherent characteristic frequency bands. Meanwhile, the energy of a genuine discharge signal is highly concentrated around a dominant frequency, while the frequency components of noise signals are usually very dispersed. Therefore, by checking whether the frequency point with the highest energy falls within the expected frequency band and whether the energy at that frequency is sufficiently concentrated (i.e., whether the F% is high enough), the authenticity of the signal can be verified from a physical mechanism perspective.

[0102] This approach extends signal analysis from the time and statistical domains to the frequency domain, creating a multi-dimensional, cross-validated discrimination system. It fully utilizes the inherent physical characteristics of partial discharge, significantly enhancing the system's anti-interference capability and identification reliability in complex electromagnetic environments, distinguishing real discharges from various frequency band noises. This is a crucial element in ensuring high detection accuracy.

[0103] S108. If the energy at the first frequency point is within the first range and the proportion of the first energy is greater than the second proportional threshold, then the first effective pulse data is determined to be an effective partial discharge signal.

[0104] If the first valid pulse data is determined to be a valid partial discharge signal, then based on the spectrum type at time t, the kurtosis, skewness, cross-correlation coefficient and discharge coefficient of the spectrum at time t are collected, and the first adjustment threshold is corrected to obtain the target adjustment threshold.

[0105] Specifically, the system compares the two results calculated by S107—the first frequency point and the first energy percentage—with preset equipment-related standards. For specific equipment such as GIS (Gas Insulated Switchgear), cable joints, and transformer windings, the first range interval (characteristic frequency band) and the second proportion threshold (minimum energy concentration) are predetermined. Only when both conditions are met simultaneously can the current data be ultimately determined as a valid partial discharge signal.

[0106] Once the determination is valid, the system does not stop immediately, but enters a more refined dynamic feedback optimization stage. It extracts multiple statistical feature parameters of the current discharge spectrum in real time, including kurtosis, skewness, cross-correlation coefficient, and discharge coefficient. It then uses these features to fine-tune the first adjustment threshold through a neural network model, thereby obtaining a more optimized target adjustment threshold.

[0107] In practice, the system compares the first frequency point with the standard characteristic frequency band of the currently detected device, and simultaneously compares the first energy percentage with the corresponding threshold. If both conditions are met, a valid judgment is triggered. Subsequently, the system calculates four time-domain feature parameters from the real-time data stream and inputs them into a pre-trained neural network. This neural network matches them with a standard typical discharge feature library and outputs a threshold fine-tuning amount to correct the first adjustment threshold.

[0108] The reason for this design is that the dual criteria ensure that the signal conforms to the characteristics of a real partial discharge in both mode and frequency. The subsequent dynamic correction is to achieve and maintain the optimal detection effect. Fixed thresholds are difficult to adapt to minute fluctuations in the signal, while the feedback mechanism based on multi-dimensional features and neural networks fine-tunes the threshold in real time according to subtle changes in the discharge morphology, thereby stably capturing the most effective and purest data during long-term monitoring.

[0109] The advantage of this process lies in the fact that the final judgment provides an authoritative conclusion for the entire dual-criteria process. Subsequent dynamic feedback optimization elevates the system's intelligence level from automatic setting to autonomous optimization, ensuring that the threshold is not only set correctly but also adaptively tracks signal changes, thereby maintaining the detection system at its optimal operating point over the long term and improving the validity of the data and the accuracy of subsequent diagnostic analysis.

[0110] S109. If the first judgment result indicates that the proportion of the first phase window does not exceed the first proportional threshold, then the first adjustment threshold is adjusted to obtain the second adjustment threshold.

[0111] Wherein, the second adjustment threshold is greater than the first adjustment threshold; based on the second adjustment threshold, second effective pulse data and the number of phase windows corresponding to the second effective pulse data are collected; the proportion of the second phase window is calculated according to the number of phase windows corresponding to the second effective pulse data; if the proportion of the second phase window exceeds the first proportion threshold, the second effective pulse data is clustered. Then return to S102 and repeat the above operation.

[0112] This step addresses another scenario for the first criterion. When the first criterion result indicates that the proportion of the first phase window does not exceed the first proportional threshold, it means that the pulses are highly concentrated in phase, and the data quality is initially good in terms of distribution, potentially eliminating the need for cluster analysis. However, the system employs a further strategy: the current threshold, i.e., the first adjustment threshold, may be too high, filtering out some real but low-amplitude discharge pulses, resulting in a low phase window proportion. Therefore, instead of directly proceeding to the frequency criterion, the system proactively lowers the threshold requirement, i.e., increases the threshold to obtain the second adjustment threshold. Then, it re-acquires data—the second effective pulse data—using this new, stricter threshold and recalculates the phase window proportion, i.e., the second phase window proportion. If the new phase window proportion exceeds the first proportional threshold, the previously skipped cluster analysis process is initiated.

[0113] In practice, the processing device increases the first adjustment threshold by a step, such as 1 millivolt, to obtain the second adjustment threshold. The system clears the original data buffer, re-acquires data for one cycle based on this new threshold, and re-executes the phase window count and phase window ratio calculation steps. If the new phase window ratio exceeds the first ratio threshold, clustering processing is performed on the new second effective pulse data. Then, the process returns to S102 and repeats.

[0114] The purpose of this step is to prevent missed triggers, that is, to avoid missing valid discharge signals due to an excessively high threshold setting. While a low phase window percentage may indicate clean data, it could also be a sign of a weak signal. The system proactively raises the threshold and re-verifies to eliminate such suspected missed detections. This is a conservative yet safe design principle that ensures detection sensitivity, preferring to perform an extra verification rather than easily overlooking potential defects.

[0115] This step adds a layer of safety redundancy to the entire automatic adjustment process, effectively reducing the risk of missed signal detection due to overly lenient threshold settings. It reflects the high importance placed on detection rate in the algorithm design, enabling the system to maintain a high detection capability even when faced with weak but critical early discharge signals, thereby improving the reliability and defect detection sensitivity of the entire method.

[0116] S110. If the second judgment result indicates that the first Davidson-Bolding index is greater than or equal to the first index, or the number of the first cluster centers is greater than the first threshold, then the first adjustment threshold is adjusted to obtain the second adjustment threshold.

[0117] When the second judgment result shows that the first Davidson-Bourdin index is greater than or equal to the first index, or the number of the first cluster centers is greater than the first threshold, the system will adjust the first adjustment threshold to obtain the second adjustment threshold. Then it will return to S102 and repeat the operation. This judgment result indicates that the clustering effect of the currently collected first valid pulse data is not ideal and fails to meet the preset clustering quality requirements. The Davidson-Bourdin index is an important indicator for evaluating clustering effect; a larger value indicates a worse clustering effect, meaning the data points are more dispersed in the feature space, and the inter-cluster separation is insufficient.

[0118] The main reason for this is that the currently set first adjustment threshold may be too low, causing a large number of noise signals to be misjudged as valid pulses. This results in a chaotic distribution of data points during cluster analysis, making it impossible to form clear cluster centers. Noise signals are usually random and dispersed, forming multiple loose clusters in the feature space or failing to form a clear cluster structure, thus leading to an excessively high Davidsonburg index.

[0119] Based on this, the system needs to take adjustment measures to appropriately increase the first adjustment threshold, resulting in a second adjustment threshold. The purpose of increasing the threshold is to filter out more noise signals and retain truly effective discharge pulses. By increasing the threshold, noise interference with low amplitude but large quantity can be eliminated, making the subsequently acquired pulse data purer and more conducive to cluster analysis to form a clear cluster structure.

[0120] This adjustment process demonstrates the system's adaptive characteristics. Instead of simply maintaining the original threshold, the system dynamically adjusts the threshold setting based on the quality assessment results of the clustering effect. When the clustering effect is found to be unsatisfactory, the system can automatically identify the problem and take targeted improvement measures. This quality feedback-based adjustment mechanism ensures that the system maintains good detection performance even in complex electromagnetic environments.

[0121] Through this step, the system achieves fine-tuning of the threshold. The setting of the second threshold adjustment is not arbitrary, but based on an objective evaluation of the data processing effectiveness of the previous stage. This data quality-oriented adjustment strategy effectively improves the quality of subsequent data acquisition, providing a more reliable data foundation for the final partial discharge determination.

[0122] S111. If the energy at the first frequency point is not within the first range, or the proportion of the first energy is less than or equal to the second proportional threshold, then the first adjustment threshold is adjusted to obtain the second adjustment threshold.

[0123] When the system detects that the energy at the first frequency point is not within the first range, or that the proportion of the first energy is less than or equal to the second proportional threshold, it will adjust the first adjustment threshold to obtain the second adjustment threshold. Then, it returns to S102 and repeats the above operation. This situation indicates that the currently acquired pulse data does not conform to the typical characteristics of a real partial discharge signal in the frequency domain, and further optimization of the threshold setting is needed.

[0124] The first range is pre-defined based on the characteristic frequency bands of partial discharge for different types of power equipment, reflecting the frequency distribution range that a normal partial discharge signal of that type of equipment should have. If the energy at the first frequency point is not within this range, it indicates that the currently detected signal may originate from other interference sources or be an atypical discharge type. Similarly, if the proportion of the first energy is less than or equal to the second proportion threshold, it indicates that the signal energy is too dispersed in the frequency domain and does not form a clear dominant frequency component, which is inconsistent with the concentrated energy characteristic of a true partial discharge signal.

[0125] The main reason for this phenomenon is that the current first adjustment threshold setting may still be too low, allowing some noise signals with specific frequency characteristics to be collected. These noise signals may be similar to real discharge signals in the time domain and statistical characteristics, but they differ significantly in the frequency domain characteristics. For example, interference signals generated by some power electronic devices may have fixed frequency components, but these frequency components are not within the typical frequency range of partial discharge.

[0126] Based on the anomalies observed in the frequency domain analysis results, the system needs to adjust the first adjustment threshold to obtain a higher second adjustment threshold. Increasing the threshold can better filter out noise signals with obvious characteristics at specific frequencies, while preserving the true partial discharge signals. This adjustment, based on further optimization of frequency domain characteristics, is an important supplement to time domain and statistical feature analysis.

[0127] This adjustment process demonstrates the superiority of the system's multi-level and multi-dimensional analysis. The system not only considers time-domain and statistical characteristics but also deeply analyzes the signal's frequency-domain properties, ensuring accurate judgment of the signal's nature through cross-validation across multiple dimensions. When any feature in any dimension fails to meet the requirements, the system can respond promptly and adjust the detection parameters.

[0128] By implementing this step, the system can effectively eliminate interference signals that resemble real discharges in the time domain but differ in the frequency domain. This multi-feature fusion-based judgment strategy greatly improves the system's anti-interference capability in complex electromagnetic environments, ensuring the accuracy and reliability of partial discharge detection. The second step, setting an adjustable threshold, allows the system to better adapt to different field environments and maintain optimal detection performance.

[0129] This application uses a comprehensive approach, combining the characteristic frequency bands and energy percentage thresholds of different types of power equipment, to identify partial discharge signals from various devices. The method sets differentiated frequency ranges and energy concentration requirements for different types of power equipment, ensuring high equipment adaptability and accuracy in the detection results.

[0130] For GIS equipment, the criterion is that if the maximum energy frequency point is within the 30-100MHz range and the energy percentage (F%) is greater than 70%, then the data collected under the current threshold is considered a valid partial discharge signal. This criterion is based on the structural characteristics and discharge physics of GIS equipment, whose partial discharge signals are usually concentrated in the ultra-high frequency band, and the energy of the actual discharge signal is highly concentrated in the dominant frequency. If the maximum energy frequency point is not within the 30-100MHz range, or the energy percentage (F%) is less than or equal to 70%, it indicates that the current threshold setting is too low, and the collected data contains a large amount of noise interference. The system will automatically increase the threshold by 1mV and re-execute the detection process until the judgment condition is met.

[0131] For cable joint equipment, the judgment criterion is as follows: if the maximum energy frequency point is within the 1-10MHz range and the energy percentage (F%) is greater than 75%, then the data below the current threshold is considered a valid partial discharge signal. Discharge signals from cable joints are typically concentrated in the mid-frequency band, and this frequency range is set to fully consider the electromagnetic wave propagation characteristics of the cable structure. If the maximum energy frequency point is not within the 1-10MHz range, or the energy percentage (F%) is less than or equal to 75%, then the current threshold is considered too low, and noise interference severely affects signal quality. The system increases the threshold by 1mV and re-detects until the signal characteristics meet the requirements.

[0132] For transformer winding equipment, the criterion is that if the maximum energy frequency point is within the 5-20MHz range and the energy percentage (F%) is greater than 80%, then the data below the current threshold is considered a valid partial discharge signal. The discharge characteristic frequency band setting for transformer windings takes into account their complex insulation structure and electromagnetic environment, requiring a higher energy concentration. If the maximum energy frequency point is not within the 5-20MHz range, or the energy percentage (F%) is less than or equal to 80%, it indicates that the current threshold setting is insufficient to effectively filter noise. The system increases the threshold by 1mV and repeats the detection process until data meeting the standard is obtained.

[0133] This device-specific differentiation method has advantages. First, by setting specific characteristic frequency bands for different devices, it fully considers the physical characteristics and discharge mechanisms of various devices, ensuring targeted detection. Second, setting different energy percentage thresholds reflects the typical characteristics of discharge signals from various devices, effectively distinguishing between real discharges and noise interference. Finally, a dynamic adjustment mechanism is adopted, gradually increasing the threshold and re-detecting to ensure that the final threshold can effectively capture real discharge signals while suppressing noise interference to the greatest extent.

[0134] This method demonstrates good adaptability in practical applications. For GIS equipment, the ultra-high frequency band effectively avoids common electromagnetic interference frequency bands; for cable joints, the selection of the intermediate frequency band conforms to their signal propagation characteristics; and for transformer windings, the requirement for a high energy ratio ensures signal reliability. This refined dual requirement of frequency and energy constitutes a complete signal validity verification system, providing reliable technical support for partial discharge detection.

[0135] By implementing this method, the system can automatically adjust the detection standards according to the characteristics of different devices, achieving a leap from general-purpose detection to specialized detection. This not only improves the accuracy and reliability of detection but also significantly reduces the false positive rate, providing more reliable data support for the insulation status assessment of power equipment. The application of this method will further promote the development of partial discharge detection technology towards intelligence and precision.

[0136] Based on the above, this application provides a threshold adjustment method based on partial discharge data. The method first obtains the spectrum type at time t and sets a first adjustment threshold accordingly. Then, based on this threshold, it collects first effective pulse data and the corresponding number of phase windows, thereby calculating the first phase window proportion. When the first phase window proportion exceeds a preset first proportion threshold, it performs clustering processing on the first effective pulse data to obtain a first Davidson-Bolding index. If this index is less than a first index, it further performs a fast Fourier transform on the data to extract the first frequency point with the highest energy and calculates its first energy proportion. Finally, when the first frequency point falls within a first range and the first energy proportion exceeds a second proportion threshold, the first effective pulse data is determined to be an effective partial discharge signal.

[0137] This application determines the initial threshold by spectrum type, and then uses a dual criterion consisting of phase window ratio analysis, cluster analysis, and frequency energy analysis to achieve adaptive threshold adjustment. This method can effectively filter out noise interference and accurately identify valid partial discharge signals under different devices and spectrum modes, thus improving the accuracy and reliability of partial discharge detection.

[0138] The above text combined Figure 1 The threshold adjustment method based on partial discharge data provided in the embodiments of this application has been described in detail. The apparatus and equipment provided in the embodiments of this application will be described below with reference to the accompanying drawings.

[0139] This application also provides a threshold adjustment device based on partial discharge data, such as... Figure 2 As shown in the figure, this figure is a schematic diagram of a threshold adjustment device based on partial discharge data provided in an embodiment of this application. The device includes: an acquisition module 201, a first determination module 202, and a second determination module 203.

[0140] The acquisition module 201 is used to acquire the spectrum type of the monitored power equipment at time t, and set a first adjustment threshold according to the spectrum type at time t.

[0141] The first determination module 202 is used to collect first effective pulse data and the number of phase windows corresponding to the first effective pulse data based on the first adjustment threshold; calculate the first phase window ratio according to the number of phase windows corresponding to the first effective pulse data; if the first phase window ratio exceeds the first ratio threshold, then perform clustering processing on the first effective pulse data to obtain the first Davidson-Bolding index and the number of first cluster centers.

[0142] The second determination module 203 is used to perform a Fast Fourier Transform (FFT) on the first effective pulse data if the first Davidson-Bolding index is less than the first index and the number of first cluster centers is less than or equal to the first threshold, to obtain the first frequency point with the largest energy in the first effective pulse data, and to calculate the first energy proportion of the first frequency point; if the energy of the first frequency point is not in the first range, or the first energy proportion is less than or equal to the second proportion threshold, then the first adjustment threshold is adjusted to obtain the second adjustment threshold.

[0143] In some possible implementations, the device further includes:

[0144] The first determination module 202 is further configured to adjust the first adjustment threshold to obtain a second adjustment threshold if the proportion of the first phase window does not exceed the first proportional threshold; wherein the second adjustment threshold is greater than the first adjustment threshold.

[0145] Based on the second adjustment threshold, the second effective pulse data and the number of phase windows corresponding to the second effective pulse data are collected;

[0146] The proportion of the second phase window is calculated based on the number of phase windows corresponding to the second effective pulse data.

[0147] If the proportion of the second phase window exceeds the first proportion threshold, then the second effective pulse data is clustered.

[0148] In some possible implementations, the device further includes:

[0149] The second determination module 203 is further configured to adjust the first adjustment threshold to obtain a second adjustment threshold if the first Davidson-Bauer index is greater than or equal to the first index, or the number of the first cluster centers is greater than the first threshold; wherein the second adjustment threshold is greater than the first adjustment threshold.

[0150] In some possible implementations, the device further includes:

[0151] The second determination module 203 is further configured to determine the first effective pulse data as an effective partial discharge signal if the energy of the first frequency point is in the first range and the proportion of the first energy is greater than the second proportional threshold.

[0152] In some possible implementations, the device further includes:

[0153] The second determination module 203 is further configured to, if the first effective pulse data is determined to be an effective partial discharge signal, collect the kurtosis, skewness, cross-correlation coefficient and discharge coefficient of the spectrum at time t based on the spectrum type at time t, and correct the first adjustment threshold to obtain the target adjustment threshold.

[0154] In some possible implementations, the map type includes partial discharge phase-resolved maps and phase-resolved pulse sequence maps;

[0155] When the spectrum type is a partial discharge phase-resolved spectrum, the first adjustment threshold is set as follows:

[0156] The partial discharge data of the phase window within the corresponding period is cached, and the number of effective pulses in the phase window is counted. If the number of effective pulses exceeds the preset pulse number threshold, the threshold is incremented. The operation is repeated until the number of effective pulses does not exceed the preset pulse number threshold. The incremented threshold is determined as the first adjustment threshold under the partial discharge phase resolution spectrum.

[0157] When the spectrum type is a phase-resolved pulse sequence spectrum, the first adjustment threshold is set as follows:

[0158] Partial discharge data of the phase window is cached in batches within the corresponding period. The number of effective pulses in each batch is counted. If the number of effective pulses exceeds the preset batch pulse number threshold, the threshold is increased. The operation is repeated until the number of effective pulses does not exceed the preset batch pulse number threshold. The increased threshold is determined as the first adjustment threshold under the phase-resolved pulse sequence spectrum.

[0159] The threshold adjustment device based on partial discharge data according to the embodiments of this application can correspond to the execution of the method described in the embodiments of this application, and the other operations and / or functions of each module / unit of the threshold adjustment device based on partial discharge data are respectively for implementing Figure 1 For the sake of brevity, the corresponding processes of each method in the illustrated embodiments will not be described in detail here.

[0160] This application also provides a computing device. For example... Figure 3 As shown in the figure, this is a schematic diagram of a computing device provided in an embodiment of this application. The computing device 400 includes a bus 401, a processor 402, a communication interface 403, and a memory 404. The processor 402, the memory 404, and the communication interface 403 communicate with each other via the bus 401.

[0161] Bus 401 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, Figure 3 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0162] Processor 402 can be any one or more of the following processors: central processing unit (CPU), graphics processing unit (GPU), microprocessor (MP), or digital signal processor (DSP).

[0163] The communication interface 403 is used for communication with external devices. For example, if the computing device is a first switch, the communication interface 403 can be used for communication between the first switch and a first user terminal, or for communication between the first switch and a second switch.

[0164] Memory 404 may include volatile memory, such as random access memory (RAM). Memory 404 may also include non-volatile memory, such as read-only memory (ROM), flash memory, hard disk drive (HDD), or solid state drive (SSD).

[0165] The memory 404 stores executable code, and the processor 402 executes the executable code to perform the aforementioned threshold adjustment method based on partial discharge data.

[0166] Specifically, in achieving Figure 2 In the case of the illustrated embodiment, and Figure 2 When the modules or units of the threshold adjustment device based on partial discharge data described in the embodiments are implemented by software, the following steps are performed: Figure 2 The software or program code required for the functions of each module / unit can be partially or entirely stored in memory 404. Processor 402 executes the program code corresponding to each unit stored in memory 404, and executes the aforementioned threshold adjustment method based on partial discharge data.

[0167] This application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that a computing device can store, or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state drive). The computer-readable storage medium includes instructions that instruct a computing device to execute the aforementioned threshold adjustment method based on partial discharge data.

[0168] This application also provides a computer program product comprising one or more computer instructions. When the computer instructions are loaded and executed on a computing device, all or part of the processes or functions described in this application are generated.

[0169] The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, or data center to another website, computer, or data center via wired (e.g., coaxial cable, fiber optic) or wireless (e.g., infrared, wireless, microwave, etc.) means.

[0170] When the computer program product is executed by a computer, the computer performs any of the aforementioned threshold adjustment methods based on partial discharge data. The computer program product can be a software installation package; when any of the aforementioned threshold adjustment methods based on partial discharge data needs to be used, the computer program product can be downloaded and executed on the computer.

[0171] The descriptions of the processes or structures corresponding to the above figures each have their own emphasis. For parts of a process or structure that are not described in detail, please refer to the relevant descriptions of other processes or structures.

[0172] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. A method for adjusting a threshold value based on partial discharge data, characterized in that, The method comprises: acquiring a graph type of partial discharge data of a monitored power equipment at a t time, and setting a first adjustment threshold according to the graph type at the t time; collecting first effective pulse data and a corresponding phase window number of the first effective pulse data based on the first adjustment threshold; calculating a first phase window proportion according to the phase window number corresponding to the first effective pulse data; if the first phase window proportion exceeds a first proportion threshold, performing clustering processing on the first effective pulse data to obtain a first Davies-Bouldin index and a first cluster center number; if the first Davies-Bouldin index is less than a first index and the first cluster center number is less than or equal to a first threshold, performing fast Fourier transform (FFT) on the first effective pulse data to obtain a first frequency point with maximum energy in the first effective pulse data, and calculating a first energy proportion of the first frequency point; if the energy of the first frequency point is not in a first range interval or the first energy proportion is less than or equal to a second proportion threshold, adjusting the first adjustment threshold to obtain a second adjustment threshold.

2. The method of claim 1, wherein, The method further comprises: if the first phase window proportion does not exceed the first proportion threshold, adjusting the first adjustment threshold to obtain a second adjustment threshold; wherein the second adjustment threshold is greater than the first adjustment threshold; collecting second effective pulse data and a corresponding phase window number of the second effective pulse data based on the second adjustment threshold; calculating a second phase window proportion according to the phase window number corresponding to the second effective pulse data; if the second phase window proportion exceeds the first proportion threshold, performing clustering processing on the second effective pulse data.

3. The method of claim 1, wherein, The method further comprises: if the first Davies-Bouldin index is greater than or equal to the first index or the first cluster center number is greater than the first threshold, adjusting the first adjustment threshold to obtain a second adjustment threshold; wherein the second adjustment threshold is greater than the first adjustment threshold.

4. The method of claim 1, wherein, The method further comprises: if the energy of the first frequency point is in the first range interval and the first energy proportion is greater than the second proportion threshold, determining that the first effective pulse data is an effective partial discharge signal.

5. The method of claim 4, wherein, The method further comprises: if it is determined that the first effective pulse data is an effective partial discharge signal, collecting a graph kurtosis, skewness, cross-correlation coefficient and discharge coefficient at the t time based on the graph type at the t time, and correcting the first adjustment threshold to obtain a target adjustment threshold.

6. The method of claim 1, wherein, The graph type comprises a partial discharge phase resolution graph and a phase resolution pulse sequence graph; when the graph type is the partial discharge phase resolution graph, the first adjustment threshold is set in the following manner: buffering phase window partial discharge data in a corresponding period, counting the number of effective pulses in the phase window, and if the number of effective pulses exceeds a preset pulse number threshold, increasing the threshold, repeating the operation until the number of effective pulses does not exceed the preset pulse number threshold, and determining the increased threshold as the first adjustment threshold under the partial discharge phase resolution graph; when the graph type is the phase resolution pulse sequence graph, the first adjustment threshold is set in the following manner: In the corresponding period, the partial discharge data of the phase window is cached in batches, the number of effective pulses in each batch is counted, and if the number of effective pulses exceeds a preset batch pulse threshold, the threshold is incremented, and the operation is repeated until the number of effective pulses does not exceed the preset batch pulse threshold, and the incremented threshold is determined as the first adjustment threshold under the phase resolution pulse sequence atlas.

7. A threshold adjustment device based on partial discharge data, characterized by The device comprises: An acquisition module is configured to acquire a graph type of partial discharge data of a monitored power equipment at a t-th moment, and set a first adjustment threshold according to the graph type at the t-th moment. A first determination module is configured to collect first effective pulse data and a number of phase windows corresponding to the first effective pulse data based on the first adjustment threshold, calculate a first phase window proportion according to the number of phase windows corresponding to the first effective pulse data, and perform clustering processing on the first effective pulse data if the first phase window proportion exceeds a first proportion threshold, to obtain a first Davies-Bouldin index and a first number of clustering centers. A second determination module is configured to perform fast Fourier transform (FFT) on the first effective pulse data if the first Davies-Bouldin index is less than a first index and the first number of clustering centers is less than or equal to a first threshold, to obtain a first frequency point with the maximum energy in the first effective pulse data and calculate a first energy proportion of the first frequency point, and adjust the first adjustment threshold to obtain a second adjustment threshold if the energy of the first frequency point is not in a first range interval or the first energy proportion is less than or equal to a second proportion threshold.

8. The apparatus of claim 7, wherein, The first determination module is further configured to adjust the first adjustment threshold to obtain a second adjustment threshold if the first phase window proportion does not exceed the first proportion threshold, wherein the second adjustment threshold is greater than the first adjustment threshold. Second effective pulse data and a number of phase windows corresponding to the second effective pulse data are collected based on the second adjustment threshold. A second phase window proportion is calculated according to the number of phase windows corresponding to the second effective pulse data. The second effective pulse data is processed by clustering if the second phase window proportion exceeds the first proportion threshold.

9. A computing device, comprising: A memory and a processor are included; One or more computer programs are stored in the memory, and the one or more computer programs include instructions; when the instructions are executed by the processor, the computing device performs the method of any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium is used to store a computer program for executing the method of any one of claims 1 to 6.

Images