Partial discharge signal identification method and device based on multi-dimensional features, and medium

Abstract

The invention discloses a partial discharge signal identification method and device based on multi-dimensional features and a medium, and the method comprises the steps: collecting a partial discharge signal and a random interference pulse generated in the operation process of a transformer substation, and constructing an original data set according to the partial discharge signal and the random interference pulse; expanding the original data set through a data enhancement technology to obtain a sample data set after sample expansion; extracting a multi-dimensional feature corresponding to each signal sample from the sample data set; wherein the multi-dimensional features comprise a time domain feature, a frequency domain feature and a time-frequency domain feature; and based on the multi-dimensional features, training through a support vector machine technology to obtain a signal classification model so as to classify the partial discharge signals and the random interference pulses through the signal classification model.

Description

Technical Field

[0001] This application relates to the field of signal recognition technology, specifically to a method, device, and medium for identifying partial discharge signals based on multi-dimensional features. Background Technology

[0002] Partial discharge is a key factor leading to insulation faults in transformers. Partial discharge detection can effectively assess the insulation status of equipment and ensure its safe and stable operation. However, substations often experience severe electromagnetic interference. The operation of switching equipment and discharges from lines or nearby equipment can generate numerous random pulse interferences. These pulse interferences may resemble real partial discharge signals in amplitude and waveform, interfering with accurate partial discharge detection. Existing typical noise suppression methods are mainly based on signal decomposition techniques. These methods separate the partial discharge signal from the noise signal components through transformation, then remove the noise components to restore the partial discharge signal. However, in practical applications, removing noise components often results in the loss of discharge signal components, leading to attenuation of the amplitude and waveform distortion of the partial discharge signal after denoising. This results in problems such as loss of discharge information and reduced accuracy of partial discharge detection.

[0003] With the development of artificial intelligence technology, data-driven techniques are commonly used to completely separate discharge signal components from noise components. These techniques directly identify and classify signals based on waveform characteristics, filtering out and suppressing interference pulses. However, the performance of data-driven methods heavily relies on large-scale, high-quality labeled datasets. For partial discharge detection in transformers, acquiring and labeling massive amounts of field-measured data is extremely difficult, and the scarcity of datasets severely restricts the generalization ability and robustness of intelligent diagnostic models. Furthermore, the waveforms of partial discharge signals and random interference pulses are complex and diverse. Existing signal recognition models based on single-dimensional features struggle to effectively identify complex interference pulses, limiting the accuracy of signal separation. Summary of the Invention

[0004] To address the aforementioned problems, this application proposes a partial discharge signal identification method based on multi-dimensional features, comprising: Collect partial discharge signals and random interference pulses generated during the operation of the substation, and construct an original dataset based on the partial discharge signals and random interference pulses; The original dataset is augmented using data augmentation techniques to obtain an augmented sample dataset. From the sample dataset, extract the multi-dimensional features corresponding to each signal sample; wherein, the multi-dimensional features include time-domain features, frequency-domain features, and time-frequency-domain features; Based on the aforementioned multi-dimensional features, a signal classification model is trained using support vector machine technology to classify partial discharge signals and random interference pulses.

[0005] In one implementation of this application, the data enhancement techniques include signal length normalization, temporal perturbation, and noise superposition based on signal-to-noise ratio.

[0006] In one implementation of this application, the original dataset is enhanced using data augmentation techniques, specifically including: The signal length standardization process standardizes the signal length of the signal samples in the original dataset so that the signal length of the processed signal samples is consistent with the signal input length required by the signal classification model.

[0007] In one implementation of this application, the signal length of the signal samples in the original dataset is standardized, specifically including: If the signal length is greater than a preset length, data points within the preset length range are extracted as valid signal samples. If the signal length is less than the preset length, zero-padding is performed at the end of the signal sample until the signal length corresponding to the zero-padding signal sample is equal to the preset length.

[0008] In one implementation of this application, the original dataset is enhanced using data augmentation techniques, specifically including: The signal sample is multiplied by a random scaling factor through the time-domain perturbation to simulate signal energy fluctuations, and the signal sample is randomly cyclically shifted on the time axis to simulate acquisition-triggered jitter.

[0009] In one implementation of this application, the original dataset is augmented using data augmentation techniques to obtain an expanded sample dataset, specifically including: By superimposing noise based on signal-to-noise ratio, a specified signal-to-noise ratio value is selected within a preset signal-to-noise ratio range; Calculate the noise power that needs to be superimposed on the signal samples based on the specified signal-to-noise ratio value; Based on the noise power, Gaussian white noise is generated and superimposed on the signal sample to obtain the sample augmented dataset.

[0010] In one implementation of this application, calculating the noise power to be superimposed on the signal samples based on the specified signal-to-noise ratio value specifically includes: The noise power to be superimposed on the signal sample is calculated using the following formula, based on the specified signal-to-noise ratio value:

[0011]

[0012] in, Indicates signal power. The noise power is represented by SNR, the specified signal-to-noise ratio is represented by L, and L represents the total number of data points contained in the signal sample. This represents the amplitude of the signal sample at the t-th data point.

[0013] In one implementation of this application, generating Gaussian white noise based on the noise power specifically includes: Based on the noise power, Gaussian white noise with a variance equal to the noise power and a mean of 0 is generated.

[0014] This application provides a partial discharge signal identification device based on multi-dimensional features, the device comprising: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to: Collect partial discharge signals and random interference pulses generated during the operation of the substation, and construct an original dataset based on the partial discharge signals and random interference pulses; The original dataset is augmented using data augmentation techniques to obtain an augmented sample dataset. From the sample dataset, extract the multi-dimensional features corresponding to each signal sample; wherein, the multi-dimensional features include time-domain features, frequency-domain features, and time-frequency-domain features; Based on the aforementioned multi-dimensional features, a signal classification model is trained using support vector machine technology to classify partial discharge signals and random interference pulses.

[0015] This application provides a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured as follows: Collect partial discharge signals and random interference pulses generated during the operation of the substation, and construct an original dataset based on the partial discharge signals and random interference pulses; The original dataset is augmented using data augmentation techniques to obtain an augmented sample dataset. From the sample dataset, extract the multi-dimensional features corresponding to each signal sample; wherein, the multi-dimensional features include time-domain features, frequency-domain features, and time-frequency-domain features; Based on the aforementioned multi-dimensional features, a signal classification model is trained using support vector machine technology to classify partial discharge signals and random interference pulses.

[0016] The partial discharge signal identification method based on multi-dimensional features proposed in this application can bring the following beneficial effects: By employing data augmentation techniques such as signal length standardization, temporal perturbation, and controllable noise superposition, the scale of training samples is effectively expanded based on the original dataset, thereby significantly improving the generalization ability of the signal classification model during training. Secondly, by extracting time-domain, frequency-domain, and time-frequency-domain features, the fundamental differences between partial discharge signals and random interference pulses in waveform complexity, energy distribution, and statistical characteristics are comprehensively reflected. This overcomes the limited recognition capability of single-dimensional features and effectively improves the classification accuracy of partial discharge signals and interference pulses. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart illustrating a partial discharge signal identification method based on multi-dimensional features according to an embodiment of this application. Figure 2 A schematic diagram of a confusion matrix of test set classification results provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a partial discharge signal identification device based on multi-dimensional features, provided in an embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0019] The technical solutions provided by the various embodiments of this application are described in detail below with reference to the accompanying drawings.

[0020] like Figure 1 As shown in the embodiment of this application, a partial discharge signal identification method based on multi-dimensional features is provided, including: S101: Collect partial discharge signals and random interference pulses generated during the operation of the substation, and construct the original dataset based on the partial discharge signals and random interference pulses.

[0021] Partial discharge signals refer to the electrical signals generated when partial discharge occurs in the insulation of substation equipment. Random interference pulses, on the other hand, are pulse signals generated by factors such as electromagnetic interference at the substation site, switching equipment operation, and discharge from lines or nearby equipment. During on-site measurements at the substation, data is collected at different time periods and under different operating conditions using signal acquisition equipment. The collected partial discharge signals and random interference pulses are initially screened and preprocessed to remove obvious erroneous data and outliers, thus constructing the original dataset.

[0022] S102: The original dataset is augmented using data augmentation techniques to obtain an augmented sample dataset.

[0023] Since the amount of data in the original dataset may be insufficient to support effective model learning and training, and its data features may not be rich enough, data augmentation techniques are needed to expand and optimize the original dataset to improve the model's generalization ability and robustness. Data augmentation techniques refer to the techniques of generating new and reasonable training samples by transforming existing data. To make the generated new data more reliable, the original dataset needs to be augmented by simulating the changes and interference processes of signals in the actual physical environment. This includes three steps: signal length normalization, time-domain perturbation, and noise superposition based on signal-to-noise ratio. These steps need to be performed in sequence.

[0024] In one embodiment, signal length standardization refers to processing signal samples of different lengths in the original dataset to achieve a uniform length standard. This is because, during actual data acquisition, the signal lengths of partial discharge signals and random interference pulses may differ, while signal classification models typically require input signals of a fixed length. Signal length standardization ensures that the signal length input to the model is consistent, avoiding adverse effects on model training and recognition results due to differences in signal length. Therefore, this embodiment uses signal length standardization technology to standardize the signal lengths of signal samples in the original dataset, ensuring that the processed signal sample lengths match the signal input length required by the signal classification model.

[0025] Specifically, if the signal length exceeds the preset length, it indicates that the signal sample contains too many data points. To ensure that the signal length input to the model meets the requirements, data points within the preset length range need to be truncated as valid signal samples. For an input signal x(t), its length is denoted as len(x), and the specified length of the model input signal is set to N. When the signal length exceeds the preset length N, only data points within the preset length range are retained, i.e. If the preset length is 1000 data points, and the length of a certain signal sample is 1200 data points, then the first 1000 data points are taken as the valid data of the signal sample.

[0026] If the signal length is less than the preset length, it indicates that the signal sample contains too few data points. To meet the input signal length requirements of the signal classification model, zeros need to be padded at the end of the signal sample until the length of the padded signal sample equals the preset length. For example, if the preset length is 1000 data points, and a signal sample has 800 data points, then 200 zeros are padded at the end of the signal sample to make its length reach 1000 data points. The specific process is represented by the following formula:

[0027] In one embodiment, after standardizing the signal length, the signal samples need to be further perturbed in the time domain. Time-domain perturbation simulates energy fluctuations and acquisition trigger jitter that may occur in the signal in a real physical environment. Through time-domain perturbation, the characteristics of the signal samples can be further enriched, improving the model's adaptability to different signal characteristics. Specifically, the signal sample after signal length standardization is multiplied by a random scaling factor, which is a value randomly selected within a certain range, such as between 0.8 and 1.5. In a real physical environment, the energy of the signal is not constant, and the acquisition device may also experience some jitter when triggering signal acquisition. By multiplying the signal sample by the random scaling factor, the fluctuations in signal energy can be simulated.

[0028] Next, the signal sample is randomly cyclically shifted along the time axis. Random cyclic shifting refers to moving the signal sample left or right along the time axis by a certain time step, and when the signal moves beyond its boundary, the excess portion is cyclically moved to the other end of the signal. For example, for a signal sample x(t), it is randomly cyclically shifted along the time axis by a random integer s (-20 ≤ s ≤ 20) to simulate jitter triggered by data acquisition. Assuming a signal sample with 1000 data points, if a shift step of 20 is selected, shifting the signal sample 20 data points to the right moves the original last 20 data points to the beginning of the signal, and the remaining data points move sequentially to the right. This random cyclic shifting operation can simulate jitter triggered by data acquisition.

[0029] After completing the time-domain perturbation processing, noise superposition processing based on the signal-to-noise ratio (SNR) is also required. In actual substation environments, partial discharge signals are subject to various noise interferences. SNR-based noise superposition simulates this noise interference. First, a specified SNR value is selected within a preset SNR range. This preset SNR range can be set according to the actual noise conditions of the substation environment; for example, it can be set to -5dB to 20dB. A value is randomly selected from this range as the specified SNR value. Then, based on the specified SNR value, the noise power to be superimposed on the signal sample is calculated. Specifically, the noise power can be calculated using the following formula:

[0030]

[0031] in, Indicates signal power. The noise power is represented by SNR, the specified signal-to-noise ratio is represented by L, and L represents the total number of data points contained in the signal sample. This represents the amplitude of the signal sample at the t-th data point.

[0032] Finally, Gaussian white noise is generated based on the calculated noise power. Gaussian white noise is a type of random noise with a Gaussian distribution, a mean of 0, and a variance equal to the noise power. A random number generator is used to generate a sequence of random numbers conforming to a Gaussian distribution. These random numbers have a mean of 0 and a variance equal to the calculated noise power; this random number sequence is the generated Gaussian white noise. The generated Gaussian white noise is then superimposed onto the signal samples, completing the signal-to-noise ratio-based noise superposition processing and obtaining the augmented sample dataset. This data augmentation technique effectively expands the original dataset, improves the model's generalization ability and robustness, and enables the model to better cope with the complexities of real-world substation environments. S103: Extract multi-dimensional features corresponding to each signal sample from the sample dataset; where multi-dimensional features include time-domain features, frequency-domain features, and time-frequency-domain features.

[0033] After augmenting the original dataset using data augmentation techniques, a sufficient number of sample datasets were obtained, denoted as follows: , , … The length of the signal sample is N. Features are extracted from each sample signal in three dimensions: time domain, frequency domain, and time-frequency domain. Based on the extracted multi-dimensional features, a signal recognition model is constructed to improve the recognition accuracy of partial discharge signals and random interference pulses.

[0034] In terms of time-domain features, the standard deviation of the signal can be extracted ( ), pulse factor ( ), kurtosis ( ) and skewness ( Features include: standard deviation, impulse factor, kurtosis, and skewness. Standard deviation reflects the degree of fluctuation in the signal over time; impulse factor reflects the impulse characteristics of the signal, highlighting the intensity of impulses; kurtosis describes the peaks in the signal; and skewness measures the asymmetry of the signal distribution, with positive values ​​indicating right skewness and negative values ​​indicating left skewness. The calculation processes for these features are as follows:

[0035]

[0036]

[0037]

[0038] In the above formula, This represents each signal sample in the sample dataset. The mean of a signal sample can be represented by the formula... Calculated.

[0039] Regarding frequency domain features, the desired feature to be extracted is the frequency centroid. First, the signal sample is transformed from the time domain to the frequency domain using a Fast Fourier Transform to obtain the signal's spectrum. Then, the main frequency components of the spectrum are extracted. This can be expressed as the following formula:

[0040] in, For the frequency centroid, For signal samples Discrete Fourier Transform, For frequency.

[0041] Regarding time-frequency domain features, the feature to be extracted is time-frequency entropy, which reflects the complexity and uncertainty of the signal in the time-frequency domain. First, the signal sample is transformed from the time domain to the time-frequency domain using wavelet transform to obtain the signal's time-frequency distribution. Wavelet transform has good time-frequency localization properties, enabling simultaneous analysis of the signal in both the time and frequency domains. After obtaining the signal's time-frequency distribution, the time-frequency entropy is calculated based on this distribution. The specific calculation process is as follows:

[0042]

[0043] in, Let f represent the probability distribution function, where t is time and f is frequency. For signal samples The short-time Fourier transform.

[0044] S104: Based on multi-dimensional features, a signal classification model is obtained by training with support vector machine technology, so as to classify partial discharge signals and random interference pulses through the signal classification model.

[0045] After extracting the multi-dimensional features corresponding to each signal sample through the above process, all feature data are randomly divided into training, validation, and test sets according to a specific data ratio for model training and testing. For example, the training, validation, and test sets can be constructed in a 70%:15%:15% ratio. A signal classification model is constructed using traditional support vector machine (SVM) technology. The SVM classifier is trained using the feature data and its corresponding labels, ultimately resulting in a signal classification model capable of distinguishing between partial discharge signals and random interference pulses.

[0046] Using the above method, classification tests were conducted, collecting 100 different types of partial discharge signals and 100 random interference pulses measured in the substation. Data augmentation techniques were used to expand the original dataset, resulting in time-domain waveform sequences of 1000 partial discharge signals and 1000 random interference pulses. Finally, a sample dataset of 2000 samples was constructed, with each sample presented as a time-series signal and its corresponding label. The label indicates the signal type: label "1" represents a partial discharge signal, while label "0" represents a random interference pulse.

[0047] Multi-dimensional features are extracted from each sample signal. A signal classification model is trained using these features, and the classification results are validated using a test set. The confusion matrix of the classification results is shown below. Figure 2As shown in the diagram, according to the confusion matrix, 148 partial discharge signals and 149 interference pulses were accurately identified, 1 partial discharge signal was incorrectly identified as an interference pulse, and 2 interference pulses were incorrectly identified as partial discharge signals. Analysis of the classification results indicates that the overall classification accuracy of the signal classification model reached 99.0%.

[0048] The above are embodiments of the methods proposed in this application. Based on the same idea, some embodiments of this application also provide devices and non-volatile computer storage media corresponding to the above methods.

[0049] Figure 3 This is a schematic diagram of a partial discharge signal identification device based on multi-dimensional features, provided as an embodiment of this application. Figure 3 As shown, it includes: At least one processor; and, At least one processor-communication-connected memory; wherein, The memory stores instructions that can be executed by at least one processor, and the instructions, when executed by at least one processor, enable at least one processor to: Collect partial discharge signals and random interference pulses generated during the operation of the substation, and construct an original dataset based on the partial discharge signals and random interference pulses; The original dataset is augmented using data augmentation techniques to obtain an augmented sample dataset. From the sample dataset, extract the multi-dimensional features corresponding to each signal sample; wherein, the multi-dimensional features include time-domain features, frequency-domain features, and time-frequency-domain features; Based on the aforementioned multi-dimensional features, a signal classification model is trained using support vector machine technology to classify partial discharge signals and random interference pulses.

[0050] This application provides a non-volatile computer storage medium storing computer-executable instructions, which are configured as follows: Collect partial discharge signals and random interference pulses generated during the operation of the substation, and construct an original dataset based on the partial discharge signals and random interference pulses; The original dataset is augmented using data augmentation techniques to obtain an augmented sample dataset. From the sample dataset, extract the multi-dimensional features corresponding to each signal sample; wherein, the multi-dimensional features include time-domain features, frequency-domain features, and time-frequency-domain features; Based on the aforementioned multi-dimensional features, a signal classification model is trained using support vector machine technology to classify partial discharge signals and random interference pulses.

[0051] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device and medium embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the description of the method embodiments.

[0052] The devices and media provided in this application are one-to-one with the methods. Therefore, the devices and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0053] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0054] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0055] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0056] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0057] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0058] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0059] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0060] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0061] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A partial discharge signal recognition method based on multi-dimensional features, characterized in that, The method comprises: Collecting partial discharge signals and random interference pulses generated during the operation of a substation, and constructing an original data set according to the partial discharge signals and the random interference pulses; Through a data enhancement technique, the original data set is expanded to obtain a sample data set after sample expansion; From the sample data set, multi-dimensional features corresponding to each signal sample are extracted; wherein the multi-dimensional features include time domain features, frequency domain features and time-frequency domain features; Based on the multi-dimensional features, a signal classification model is trained through a support vector machine technique, so as to classify the partial discharge signals and the random interference pulses through the signal classification model.

2. The partial discharge signal recognition method based on multi-dimensional features according to claim 1, characterized in that, The data enhancement technique includes signal length standardization, time domain disturbance and noise superposition based on signal-to-noise ratio.

3. The partial discharge signal recognition method based on multi-dimensional features according to claim 2, characterized in that, Through the data enhancement technique, the original data set is enhanced, specifically including: Through the signal length standardization, the signal length of the signal sample in the original data set is standardized to make the signal length of the processed signal sample consistent with the signal input length required by the signal classification model.

4. The partial discharge signal recognition method based on multi-dimensional features according to claim 3, characterized in that, The signal length of the signal sample in the original data set is standardized, specifically including: In the case that the signal length is greater than a preset length, data points within the preset length range are intercepted as effective signal samples; In the case that the signal length is less than the preset length, a zero padding operation is performed at the end of the signal sample until the signal length of the zero-padded signal sample is equal to the preset length.

5. The partial discharge signal recognition method based on multi-dimensional features according to claim 2, characterized in that, Through the data enhancement technique, the original data set is enhanced, specifically including: Through the time domain disturbance, the signal sample is multiplied by a random scaling factor to simulate signal energy fluctuation, and the signal sample is randomly circularly translated on the time axis to simulate acquisition trigger jitter.

6. The partial discharge signal recognition method based on multi-dimensional features according to claim 2, characterized in that, Through the data enhancement technique, the original data set is expanded to obtain a sample data set after sample expansion, specifically including: Through the noise superposition based on signal-to-noise ratio, a specified signal-to-noise ratio value is selected within a preset signal-to-noise ratio range; According to the specified signal-to-noise ratio value, the noise power to be superimposed on the signal sample is calculated; According to the noise power, Gaussian white noise is generated, and the Gaussian white noise is superimposed into the signal sample to obtain a sample data set after sample expansion.

7. The partial discharge signal recognition method based on multi-dimensional features according to claim 6, characterized in that, According to the specified signal-to-noise ratio value, the noise power to be superimposed on the signal sample is calculated, specifically including: According to the specified signal-to-noise ratio value, the noise power to be superimposed on the signal sample is calculated by the following formula: wherein represents the signal power, represents the noise power, SNR represents a specified signal-to-noise ratio value, and L represents all data points contained in the signal sample, represents the amplitude of the signal sample at the tth data point.

8. The partial discharge signal recognition method based on multi-dimensional features according to claim 6, characterized in that, According to the noise power, Gaussian white noise is generated, specifically including: According to the noise power, Gaussian white noise with a variance of the noise power and a mean of 0 is generated.

9. A partial discharge signal recognition device based on multi-dimensional features, characterized by, The device comprises: At least one processor; and a memory connected in communication with the at least one processor; Wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform a partial discharge signal recognition method based on multi-dimensional features according to any one of claims 1-8.

10. A non-transitory computer storage medium storing computer-executable instructions that, when executed, cause a computer to perform: The computer executable instructions are configured to: The method for recognizing partial discharge signals based on multi-dimensional features according to any one of claims 1-8.

Images