Defect detection method and device for common tank type pole-mounted switchgear and computer device

Abstract

The application discloses a kind of defects detection method, device and computer equipment of common box type pole switch equipment, wherein the method comprises: target leakage current signal is collected at the grounding of target common box type pole switch equipment;Extract the target values corresponding to the respective corresponding multiple characteristic quantities of target leakage current signal, including time domain characteristic quantity and frequency domain characteristic quantity in multiple characteristic quantities;Select the target values corresponding to the respective corresponding multiple characteristic quantities of target leakage current signal in the characteristic quantity set matched with the preset characteristic quantity, as input value, the characteristic quantity set includes the characteristic quantity that the influence degree of defect detection meets preset condition;Input value is input into preset target identification model, and defect detection result is obtained.The application solves the technical problem that the internal moisture of the pole switch equipment due to the sealing performance defect is difficult to detect, which leads to the decrease of insulation performance and the increase of fault risk.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring technology, and more specifically, to a defect detection method, apparatus, and computer equipment for a common-enclosure pole-mounted switchgear. Background Technology

[0002] Pole-mounted switchgear in power distribution systems is a crucial component for ensuring power grid safety and power supply reliability. Operating outdoors, this equipment is constantly exposed to environmental factors, particularly the deterioration of its sealing performance, which leads to gradual moisture absorption of the internal insulation materials and subsequent insulation degradation. Moisture absorption in pole-mounted switches has become a significant hidden danger that urgently needs to be addressed in power distribution network maintenance. Once the internal insulation performance of the switchgear deteriorates to a certain level, it not only threatens the normal operation of the power grid but may also cause sudden equipment failures and even threaten personnel safety. Currently, the maintenance and inspection of pole-mounted switchgear mainly rely on regular planned maintenance and abnormal operational responses. Furthermore, current inspection methods often focus on surface inspection and basic functional testing of the switchgear, lacking effective monitoring methods for the moisture absorption of the internal insulation. In particular, the lack of early warning systems for insulation faults caused by moisture absorption means that potential problems are often only discovered when they have developed to a relatively serious stage.

[0003] There is currently no effective solution to the above problems. Summary of the Invention

[0004] This invention provides a defect detection method, apparatus, and computer equipment for a common-enclosure pole-mounted switchgear, to at least solve the technical problem that current pole-mounted switchgear suffers from internal moisture due to sealing performance defects, which is difficult to detect, leading to decreased insulation performance and increased failure risk.

[0005] According to one aspect of the present invention, a defect detection method for a common-enclosure pole-mounted switchgear is provided, comprising: acquiring a target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear; extracting target values ​​corresponding to multiple feature quantities corresponding to the target leakage current signal, wherein the multiple feature quantities include time-domain feature quantities and frequency-domain feature quantities, and the frequency-domain feature quantities include harmonic factors and total harmonic distortion; selecting the target value corresponding to the feature quantity that matches a preset feature quantity set from the target values ​​corresponding to the multiple feature quantities corresponding to the target leakage current signal, as an input value, wherein the feature quantity set includes feature quantities whose influence on defect detection meets preset conditions; and inputting the input value into a preset target recognition model to obtain the defect detection result of the target common-enclosure pole-mounted switchgear.

[0006] Optionally, extracting the target value corresponding to the frequency domain feature quantity corresponding to the target leakage current signal includes: converting the target leakage current signal to the frequency domain based on the Fast Fourier Transform algorithm to obtain the spectrum of the target leakage current signal; identifying the harmonic components corresponding to the target leakage current signal based on the spectrum of the target leakage current signal, and determining the target value corresponding to the harmonic factor of the target leakage current signal; determining the effective harmonic value of the target leakage current signal based on the harmonic components; and calculating the target value corresponding to the total harmonic distortion rate of the target leakage current signal based on the effective harmonic value of the target leakage current signal.

[0007] Optionally, leakage current signals of the grounding point of the sample common-enclosure pole-mounted switchgear are collected at multiple moisture-affected stages. The sample common-enclosure pole-mounted switchgear includes common-enclosure pole-mounted switchgear corresponding to various defect types. Sample values ​​corresponding to multiple feature quantities of the leakage current signals of the sample at multiple moisture-affected stages are extracted. Based on the sample values ​​corresponding to the feature quantities in the feature quantity set, a preset residual network model is trained to obtain a target recognition model.

[0008] Optionally, a target recognition model is obtained by training a preset residual network model based on the sample values ​​corresponding to each feature in the feature set, including: determining a training set based on the sample values ​​corresponding to each feature in the feature set, wherein the training set includes the sample values ​​corresponding to each feature with labeled defect types; determining a residual network model based on a preset residual network architecture, wherein the residual network architecture is a ResNet18 architecture or a ResNet35 architecture; and training the residual network model using the training set to obtain the target recognition model.

[0009] Optionally, based on the gradient boosting model and the sample values ​​corresponding to multiple feature quantities of the sample leakage current signals of multiple moisture stages, feature quantities whose influence on defects meets the preset conditions are selected from the multiple feature quantities to obtain a feature quantity set.

[0010] Optionally, when multiple time-domain features and multiple frequency-domain features exist, based on the gradient boosting model and the sample values ​​corresponding to the multiple features corresponding to the leakage current signals of samples at multiple moisture stages, features whose influence on defects meets preset conditions are selected from the multiple features to obtain a feature set, including: determining a first sample set based on the sample values ​​corresponding to the time-domain features corresponding to the leakage current signals of samples at multiple defect types and multiple moisture stages; determining a second sample set based on the sample values ​​corresponding to the frequency-domain features corresponding to the leakage current signals of samples at multiple defect types and multiple moisture stages; inputting the first and second sample sets into the gradient boosting model to obtain the importance scores corresponding to the multiple time-domain features and the multiple frequency-domain features, where the importance score characterizes the degree of influence on defect detection; and selecting features whose importance scores meet preset conditions from the multiple time-domain features and multiple frequency-domain features to determine the feature set.

[0011] Optionally, if the defect detection result indicates the presence of a defect, the pulse discharge signal of the target common-enclosure pole-mounted switchgear and the ambient humidity are detected; based on a preset severity determination formula, the degree of defect of the target common-enclosure pole-mounted switchgear is determined according to the target leakage current signal, pulse discharge signal and ambient humidity.

[0012] Alternatively, the degree determination formula is determined in the following way: ,in, As for the degree of defect, The weight corresponding to the target leakage current. The weights corresponding to ambient humidity. The weights corresponding to the amplitude of the pulse discharge signal. The level corresponding to the target leakage current. This represents the humidity level corresponding to the ambient humidity. This refers to the level corresponding to the amplitude of the pulse discharge signal.

[0013] Alternatively, the formula for determining the degree is as follows:

[0014] ,

[0015] in, As for the degree of defect, For the target leakage current, The amplitude of the pulse discharge signal. The humidity of the surrounding environment.

[0016] According to another aspect of the present invention, a defect detection device for a common-enclosure pole-mounted switchgear is also provided, comprising: a data acquisition module for acquiring a target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear; an extraction module for extracting target values ​​corresponding to multiple feature quantities corresponding to the target leakage current signal, wherein the multiple feature quantities include time-domain feature quantities and frequency-domain feature quantities, and the frequency-domain feature quantities include harmonic factors and total harmonic distortion; a selection module for selecting target values ​​corresponding to feature quantities that match a preset feature quantity set from the target values ​​corresponding to the multiple feature quantities corresponding to the target leakage current signal, as input values, wherein the feature quantity set includes feature quantities whose influence on defect detection meets preset conditions; and a detection module for inputting the input values ​​into a preset target recognition model to obtain the defect detection result of the target common-enclosure pole-mounted switchgear.

[0017] According to another aspect of the present invention, a non-volatile storage medium is also provided, the non-volatile storage medium including a stored program, wherein, when the program is running, the device where the non-volatile storage medium is located is controlled to execute any of the above-described defect detection methods for co-enclosure pole-mounted switchgear.

[0018] According to another aspect of the present invention, a computer device is also provided, the computer device including a processor for running a program, wherein the program executes any of the above-described defect detection methods for co-enclosure pole-mounted switchgear.

[0019] According to another aspect of the present invention, a computer program product is also provided, including a computer program that, when executed by a processor, implements any of the above-described defect detection methods for co-enclosure pole-mounted switchgear.

[0020] In this embodiment of the invention, a defect detection method for a common-enclosure pole-mounted switchgear is adopted. This method involves acquiring the target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear; extracting the target values ​​corresponding to multiple feature quantities corresponding to the target leakage current signal, including time-domain and frequency-domain feature quantities, with the frequency-domain feature quantities including harmonic factors and total harmonic distortion (THD); selecting the target value corresponding to the feature quantity that matches a preset feature quantity set from the target values ​​corresponding to the multiple feature quantities of the target leakage current signal, as the input value; wherein the feature quantity set includes feature quantities whose influence on defect detection meets preset conditions; and inputting the input value into a preset target recognition model to obtain the defect detection result of the target common-enclosure pole-mounted switchgear. This achieves the goal of timely detection of defects in the common-enclosure pole-mounted switchgear based on leakage current, thereby reducing the technical effect of fault risk. Furthermore, it solves the technical problem that current pole-mounted switchgear suffers from internal moisture due to sealing performance defects, which is difficult to detect, leading to decreased insulation performance and increased fault risk. Attached Figure Description

[0021] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0022] Figure 1 A hardware block diagram of a computer terminal for implementing a defect detection method for a common-enclosure pole-mounted switchgear is shown.

[0023] Figure 2 This is a flowchart illustrating a defect detection method for a common-enclosure pole-mounted switchgear according to an embodiment of the present invention.

[0024] Figure 3 This is a structural block diagram of a defect detection device for a common-enclosure pole-mounted switchgear provided according to an embodiment of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] First, some nouns or terms that appear in the description of the embodiments of this application shall be interpreted as follows:

[0028] Common-enclosure pole-mounted switchgear is an integrated and compact outdoor power distribution device widely used in medium-voltage power distribution lines. This type of switchgear houses electrical components such as circuit breakers, disconnectors, load switches, current transformers, and voltage transformers in a sealed metal enclosure (common enclosure). The enclosure is filled with insulating gas (usually sulfur hexafluoride SF6 or dry air) to protect the components and achieve insulation.

[0029] According to an embodiment of the present invention, a method embodiment for defect detection of a common-enclosure pole-mounted switchgear is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0030] The method embodiment provided in Embodiment 1 of this application can be executed on a mobile terminal, computer terminal, or similar computing device. Figure 1 A hardware block diagram of a computer terminal for implementing a defect detection method for co-enclosure pole-mounted switchgear is shown. Figure 1 As shown, the computer terminal 10 may include one or more processors (shown as 102a, 102b, ..., 102n in the figure) (the processor may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data. In addition, it may also include: a display, an input / output interface (I / O interface), a universal serial bus (USB) port (which may be included as one of the ports of a BUS bus), a network interface, a power supply, and / or a camera. Those skilled in the art will understand that... Figure 1The structure shown is for illustrative purposes only and does not limit the structure of the aforementioned electronic device. For example, computer terminal 10 may also include... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0031] It should be noted that the aforementioned one or more processors and / or other data processing circuits are generally referred to herein as "data processing circuits". These data processing circuits may be embodied, in whole or in part, in software, hardware, firmware, or any other combination thereof. Furthermore, the data processing circuits may be a single, independent processing module, or may be integrated, in whole or in part, into any other element within the computer terminal 10. As involved in the embodiments of this application, the data processing circuits serve as a processor control mechanism (e.g., selection of a variable resistor termination path connected to an interface).

[0032] The memory 104 can be used to store software programs and modules for application software, such as the program instructions / data storage device corresponding to the defect detection method for the common-enclosure pole-mounted switchgear in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory 104, thereby implementing the defect detection method for the common-enclosure pole-mounted switchgear described above. The memory 104 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory 104 may further include memory remotely located relative to the processor, and these remote memories can be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0033] The display can be, for example, a touchscreen liquid crystal display (LCD) that allows the user to interact with the user interface of the computer terminal 10.

[0034] Figure 2 This is a flowchart illustrating a defect detection method for a common-enclosure pole-mounted switchgear according to an embodiment of the present invention, as shown below. Figure 2 As shown, the method includes the following steps:

[0035] Step S202: Collect the target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear.

[0036] In this step, suitable leakage current detection instruments can be selected, typically including a wideband current sensor, a coaxial RF cable, and a high-frequency signal synchronous acquisition device. These devices should have a wideband response range, such as 100kHz to 100MHz, effectively resist electromagnetic interference, and ensure the accuracy and integrity of the data. Secure the wideband current sensor to the ground wire of the pole-mounted switchgear, ensuring correct placement and contact to guarantee accurate signal acquisition. Connect the sensor to the input of the acquisition device using a coaxial RF cable, ensuring the cable's double shielding is intact to avoid external electromagnetic interference. Set the acquisition device parameters, such as the sampling rate (e.g., 100MSa / s) and frequency response range (e.g., 100kHz to 100MHz), and connect it to the data analysis software on a computer or control terminal. Begin acquiring the leakage current signal, monitoring the leakage current changes of the pole-mounted switchgear under normal operating or experimental conditions to obtain the target leakage current signal.

[0037] By continuously monitoring leakage current, the insulation status of pole-mounted switchgear can be effectively monitored, potential faults can be dealt with in a timely manner, and the safe operation of the power system can be ensured.

[0038] Step S204: Extract the target values ​​corresponding to each of the multiple feature quantities corresponding to the target leakage current signal. Among the multiple feature quantities are time-domain feature quantities and frequency-domain feature quantities. The frequency-domain feature quantities include harmonic factors and total harmonic distortion rate.

[0039] In this step, since the leakage current signal may be affected by external electromagnetic interference, filtering or denoising processing can be performed first to improve the accuracy of feature extraction. The time-domain features can be directly extracted from the target leakage current signal, including but not limited to:

[0040] RMS (Reliable Mean Square): Calculates the effective value of the leakage current signal, reflecting the energy level of the signal.

[0041] Maximum value: Identifies the largest amplitude value in the signal waveform.

[0042] Minimum value: Identifies the minimum amplitude in a signal waveform.

[0043] Peak-to-peak value: Calculates the difference between the maximum and minimum values.

[0044] Average value: Calculate the average of the signal values.

[0045] Standard deviation: reflects the degree of fluctuation in signal strength.

[0046] Skewness and kurtosis: assess the skewness and kurtosis of the signal distribution.

[0047] However, when extracting frequency domain features, it is necessary to convert the time-domain signal to the frequency domain to analyze its frequency components. For example, performing an FFT (Fast Fourier Transform) on the acquired time-domain leakage current signal converts it to a frequency-domain signal, obtaining the amplitude and phase information of the signal at different frequencies. For each harmonic of interest (e.g., 2nd, 3rd, 4th, etc.), the amplitude ratio of that harmonic to the fundamental frequency (usually 50Hz or 60Hz) is calculated. Total Harmonic Distortion (THD) is an important indicator for measuring the degree of nonlinearity and distortion of a signal. It is determined by the square root of the ratio of the sum of the squares of the amplitudes of all harmonics to the square of the amplitude of the fundamental frequency, multiplied by 100% to express it as a percentage.

[0048] Through the above steps, the time-frequency domain characteristics of leakage current signals can be systematically extracted and analyzed, providing quantitative and reliable data support for the detection of insulation defects in pole-mounted switchgear.

[0049] Step S206: Select the target value of the feature quantity that matches the preset feature quantity set from the target values ​​of the multiple feature quantities corresponding to the target leakage current signal, and use it as the input value. The feature quantity set includes feature quantities whose influence on defect detection meets the preset conditions.

[0050] In this step, the features in the feature set can be those that significantly influence insulation defect detection, such as the time-domain statistical characteristics (RMS, maximum, average, etc.) and frequency-domain characteristics (harmonic factor, total harmonic distortion (THD), etc.) of leakage current. That is, using the values ​​of these features for defect detection is more effective. For example, a significant increase in the harmonic factor or THD may indicate a deterioration in insulation condition. The selected features and their corresponding target values ​​can be organized into a dataset, typically represented in tabular or vector form, as input to the identification algorithm.

[0051] Step S208: Input the input values ​​into the preset target recognition model to obtain the defect detection results of the target common-box pole-mounted switchgear.

[0052] In this step, the extracted and filtered input values ​​are fed into a pre-set target recognition model to obtain the defect detection results of the common-enclosure pole-mounted switchgear. This target recognition model is typically trained on a large amount of labeled leakage current signal data and is capable of identifying insulation defects at different stages. Through this process, the target recognition model based on leakage current characteristics can achieve automated detection of insulation defects in common-enclosure pole-mounted switchgear, providing timely and effective operation and maintenance guidance, thereby ensuring the safe and stable operation of the power system.

[0053] Because the time-frequency domain characteristics of leakage current are closely related to the degree of moisture absorption of the insulation of pole-mounted switchgear, analyzing these changes can accurately determine insulation defects in the equipment. Therefore, by collecting leakage current signals at the grounding point of the pole-mounted switchgear and extracting key characteristic quantities using time-frequency domain analysis techniques, including harmonic factors and total harmonic distortion (THD), the insulation status of the switchgear can be comprehensively reflected. The technology in this embodiment can effectively detect insulation defects in pole-mounted switchgear, improving the safety and reliability of power distribution network operation. Furthermore, by combining it with partial discharge detection technology, the accuracy and sensitivity of detection can be further improved, overcoming potential limitations in leakage current detection under specific conditions.

[0054] Through the above steps, the goal of timely detection of defects in common-enclosure pole-mounted switchgear based on leakage current can be achieved, thereby reducing the technical effect of fault risk. This solves the technical problem that current pole-mounted switchgear suffers from internal moisture due to sealing performance defects, which is difficult to detect, leading to decreased insulation performance and increased fault risk.

[0055] As an optional embodiment, extracting the target value corresponding to the frequency domain feature quantity of the target leakage current signal includes: converting the target leakage current signal to the frequency domain based on the Fast Fourier Transform algorithm to obtain the spectrum of the target leakage current signal; identifying the harmonic components of the target leakage current signal based on the spectrum of the target leakage current signal, and determining the target value corresponding to the harmonic factor of the target leakage current signal; determining the effective harmonic value of the target leakage current signal based on the harmonic components of the target leakage current signal; and calculating the target value corresponding to the total harmonic distortion rate of the target leakage current signal based on the effective harmonic value of the target leakage current signal.

[0056] Optionally, a wideband current sensor is used to acquire the leakage current signal at the grounding wire of the pole-mounted switchgear. The acquired signal undergoes preprocessing such as denoising and filtering to ensure signal purity and reduce the impact of external interference on feature extraction. The preprocessed time-domain leakage current signal is then converted to the frequency domain using a Fast Fourier Transform (FFT) to obtain the signal's spectrum. This step converts the signal's energy distribution into a frequency component distribution, facilitating the identification and analysis of harmonic components. The fundamental frequency (typically corresponding to the grid frequency, such as 50Hz or 60Hz) and other harmonic components (such as 2, 3, and 4 times the fundamental frequency) of the leakage current signal are identified in the spectrum. For each harmonic component, its amplitude ratio relative to the fundamental frequency is calculated, i.e., the harmonic factor. This provides information on the intensity of the harmonic component relative to the dominant frequency, which is very useful for identifying abnormal states. For each identified harmonic component (and the fundamental frequency), its effective value is calculated. The effective value is an indicator of signal energy magnitude and is crucial for the energy distribution analysis of frequency components. Based on the RMS harmonic values, the Total Harmonic Distortion (THD) is calculated. This is an indicator for assessing the degree of signal distortion, reflecting the proportion of total energy comprised of harmonic components other than the fundamental frequency. The calculated harmonic factors and THD values ​​are compared with target values ​​in a preset set of characteristic quantities to identify which characteristic quantity values ​​exceed the normal operating range, potentially indicating insulation defects or moisture in the equipment. Through these steps, frequency domain characteristic quantities can be effectively extracted from the target leakage current signal and matched with the preset set of characteristic quantities to identify and assess the insulation status of pole-mounted switchgear, providing data support and decision-making basis for the safe operation of the power system.

[0057] This optional embodiment utilizes the Fast Fourier Transform (FFT) algorithm for frequency domain feature extraction, efficiently converting time-domain signals into frequency-domain signals, facilitating the analysis of harmonic components. Harmonic factors and total harmonic distortion (THD) reflect the harmonic content and distortion degree of the signal, thereby revealing changes in the insulation state of the pole-mounted switchgear, more accurately identifying insulation defects, and providing a scientific basis for equipment maintenance and fault early warning. Other frequency domain analysis methods, such as wavelet transform or spectral analysis, can also be employed to adapt to different types of leakage current signals, improving the flexibility and applicability of detection.

[0058] As an optional embodiment, leakage current signals of sample common-enclosure pole-mounted switchgear are collected at the grounding point during multiple moisture-affected stages. The sample common-enclosure pole-mounted switchgear includes common-enclosure pole-mounted switchgear corresponding to various defect types. Sample values ​​corresponding to multiple feature quantities of the sample leakage current signals during multiple moisture-affected stages are extracted. Based on the sample values ​​corresponding to the feature quantities in the feature quantity set, a preset residual network model is trained to obtain a target recognition model.

[0059] Optionally, this optional embodiment constructs a sample dataset containing various defect types by collecting leakage current signals of pole-mounted switchgear at different moisture stages, providing a rich data foundation for feature training. The residual network model is a deep learning model capable of learning and recognizing complex signal features through multi-layer convolutions and skip connections, thereby achieving accurate identification of insulation defects in pole-mounted switchgear.

[0060] Specifically, representative pole-mounted switchgear with multiple defect types can be selected as samples to ensure coverage of different insulation defect types, such as slight dampness, severe dampness, and partial discharge. In experimental or field environments, conditions for different defect types are set to simulate or monitor leakage current signals at different stages of the equipment. For example, to simulate different levels of dampness, a constant voltage plus moisture method can be used to collect leakage current signals from the normal state of the equipment to the dampness stage, and even flashover or breakdown. Then, the time-domain characteristics of the sample leakage current signals are calculated, such as RMS, maximum, minimum, mean, and standard deviation. The sample leakage current signals are converted to the frequency domain, and frequency domain features are analyzed and extracted, such as fundamental and harmonic amplitudes, harmonic factors, and total harmonic distortion (THD). Each sample data is labeled to indicate its dampness stage or defect type for supervised learning of the model. Residual Network (ResNet) is selected as the target recognition model. ResNet solves the gradient vanishing problem of deep neural networks by introducing residual connections, making it suitable for the extraction and classification of complex signal features. As needed, ImageNet pre-trained weights can be used for model initialization to leverage features learned from large datasets for transfer learning. The sample dataset is divided into training and validation sets, typically in a ratio of 80:20 or 90:10. The cleaned feature sample values ​​are used as input to train the ResNet model. The model training process is optimized by adjusting parameters such as the learning rate, batch size, and number of iterations. Multiple rounds of training are performed using the training set data, with model parameters adjusted after each training round based on the performance on the validation set, until the model achieves satisfactory performance on the validation set. Then, performance metrics such as accuracy, recall, and F1 score are calculated on the validation set to evaluate the model's classification performance. A confusion matrix can be constructed to visually compare the model's classification results with the true labels, aiding in further performance analysis. Once the model achieves satisfactory performance, its parameters and structure are saved for future use in actual equipment maintenance.

[0061] Using the methods described above, a residual network model can be trained based on various features of the leakage current signal. This model can accurately identify the moisture stage and defect type of common-enclosure pole-mounted switchgear, providing strong technical support for intelligent operation and maintenance of power equipment. The key to model training lies in the selection of features, the construction of the dataset, the optimization of model parameters, and the evaluation of model performance, ensuring that the model can capture subtle changes in the signal and cope with different types of insulation defects. In other optional embodiments, other machine learning algorithms, such as support vector machines or neural networks, can be introduced to adapt to the detection needs of different scenarios and address the potential limitations of residual networks under certain specific conditions.

[0062] As an optional embodiment, a target recognition model is obtained by training a preset residual network model based on the sample values ​​corresponding to each feature in the feature set, including: determining a training set based on the sample values ​​corresponding to each feature in the feature set, wherein the training set includes the sample values ​​corresponding to each feature with labeled defect types; determining a residual network model based on a preset residual network architecture, wherein the residual network architecture is a ResNet18 architecture or a ResNet35 architecture; and training the residual network model using the training set to obtain the target recognition model.

[0063] Optionally, this alternative embodiment constructs an intelligent identification model for insulation defects in pole-mounted switchgear by determining a training set containing labeled defect types and selecting ResNet18 or ResNet35 as the residual network architecture. The residual network, by introducing skip connections, solves the gradient vanishing problem in deep networks, enabling it to effectively learn and identify complex features in leakage current signals.

[0064] From leakage current signals collected at different moisture stages from a sample-type column-mounted switchgear, sample values ​​corresponding to a set of feature quantities are extracted, including time-domain features (such as RMS value and maximum value) and frequency-domain features (such as harmonic factor and total harmonic distortion, THD). Each sample value is assigned a label representing the corresponding defect type or moisture stage. For example, the data can be categorized into "slightly damp," "severely damp," and "critically damp." The feature values ​​and their corresponding defect type labels are combined into input-output pairs to form a training set. The training set is ensured to cover all types of defects and have a balanced number of samples across categories to avoid model bias. Based on the data volume, computational resources, and task complexity, a residual network architecture, such as ResNet18 or ResNet35, is selected. ResNet18 is suitable for situations with moderate data volume and limited computational resources, while ResNet35 can handle more complex data but requires more computational resources. Pre-trained weights, such as those trained on the ImageNet dataset, are loaded for model initialization. This step provides the model with initial learning capabilities, especially in feature extraction. Setting training hyperparameters includes learning rate, batch size, and number of iterations. The choice of these parameters directly affects the model's training speed and final performance. The training set is fed into the residual network model, and the weights are adjusted using the backpropagation algorithm to minimize the loss function. Commonly used loss functions include cross-entropy loss and mean squared error loss, depending on the task type. During training, the model performance is periodically evaluated using a validation set to adjust hyperparameters and prevent overfitting. When the model achieves satisfactory performance on the validation set, the model's weights and structure are saved, resulting in the object recognition model.

[0065] In power equipment operation and maintenance, leakage current signals collected on-site are converted into characteristic numerical values ​​and input into a target identification model for real-time or periodic defect detection and condition assessment. The model's output can guide maintenance personnel in developing equipment maintenance strategies, such as equipment replacement and repair plans, to prevent equipment failures and ensure the safe operation of the power grid.

[0066] Through the above steps, the residual network architecture can be effectively used to identify and classify insulation defects in power equipment, providing accurate equipment condition monitoring and helping power systems achieve intelligent operation and maintenance. In other optional embodiments, the model's recognition accuracy and generalization ability can be further improved by adjusting the network architecture or optimizing the algorithm, addressing potential shortcomings of the model under certain specific conditions.

[0067] As an optional embodiment, based on the gradient boosting model and the sample values ​​corresponding to multiple feature quantities of the sample leakage current signals of multiple moisture stages, feature quantities whose influence on defects meets the preset conditions are selected from multiple feature quantities to obtain a feature quantity set.

[0068] Optionally, based on a gradient boosting model (such as XGBoost) and sample leakage current signals from multiple moisture-affected stages, feature quantities whose influence on insulation defects meets preset conditions are selected from multiple feature quantities to construct a feature quantity set.

[0069] First, a series of features potentially related to insulation defect states are pre-selected from previous signal analysis and research, including time-domain statistical characteristics of leakage current (such as RMS, maximum, and mean values) and frequency-domain characteristics (such as harmonic factors and total harmonic distortion (THD)). Leakage current signals from multiple moisture-affected stages are converted into feature values, forming a sample dataset containing all pre-selected features. Each sample is ensured to contain measurements of all pre-selected features and a corresponding moisture-affected stage label. A gradient boosting model, such as XGBoost, is selected and initialized. The XGBoost model is used to train the feature set; the model automatically evaluates the contribution of each feature to predicting the moisture-affected stage, i.e., the feature importance score. A preset feature importance threshold is used to determine which features substantially contribute to the identification of insulation defect states. Features are ranked according to the feature importance scores calculated by the XGBoost model, and those with scores below the preset threshold are removed. This step helps reduce model complexity and improve its prediction efficiency and accuracy.

[0070] Using the selected feature set, retrain XGBoost or other models and observe whether the model performance improves. Cross-validation can also be used to evaluate the stability of the new model on different datasets, confirming whether feature selection improves the model's generalization ability. Feature selection can be an iterative process, involving multiple adjustments to preset conditions and thresholds, retraining the model, and validating the results until the optimal feature set is found. The final feature set, containing features that significantly contribute to insulation defect state detection, can then be used for subsequent model training and state evaluation.

[0071] Through this series of steps, XGBoost-based feature selection can effectively identify the most critical features for detecting insulation defects in pole-mounted switches from a large pool of features, thereby constructing a more accurate and efficient condition assessment model. This not only improves the model's predictive performance but also reduces computational costs, making the model more suitable for real-time on-site detection applications. The gradient boosting model, through iterative optimization, can identify the features that have the greatest impact on defect detection, thus improving the model's recognition efficiency and accuracy. In other optional embodiments, other feature selection methods, such as principal component analysis or LASSO regression, can be employed to adapt to different types of signal data, improving the flexibility and applicability of feature selection.

[0072] As an optional embodiment, when multiple time-domain features and multiple frequency-domain features exist, based on the gradient boosting model and the sample values ​​corresponding to the multiple features corresponding to the leakage current signals of samples at multiple moisture stages, features whose influence on defects meets preset conditions are selected from the multiple features to obtain a feature set. This includes: determining a first sample set based on the sample values ​​corresponding to the time-domain features corresponding to the leakage current signals of samples at multiple defect types and multiple moisture stages; determining a second sample set based on the sample values ​​corresponding to the frequency-domain features corresponding to the leakage current signals of samples at multiple defect types and multiple moisture stages; inputting the first and second sample sets into the gradient boosting model to obtain the importance scores corresponding to the multiple time-domain features and the multiple frequency-domain features, where the importance score characterizes the degree of influence on defect detection; and selecting features whose importance scores meet preset conditions from the multiple time-domain features and multiple frequency-domain features to determine the feature set.

[0073] Optionally, based on leakage current signals from samples with various defect types and different moisture absorption stages, a set of features is determined through time-domain and frequency-domain feature analysis, combined with a gradient boosting model (such as XGBoost). Time-domain and frequency-domain feature sample sets can be constructed separately. Leakage current signals from common-enclosure pole-mounted switchgear containing various defect types at different moisture absorption stages can be collected. Time-domain features, such as RMS value, maximum value, mean, standard deviation, and waveform indicators, are extracted from the signals. The sample values ​​of the time-domain features are paired with their corresponding defect types and moisture absorption stages to construct the first sample set. The leakage current signal is converted from the time domain to the frequency domain using a Fast Fourier Transform (FFT) to obtain the signal spectrum. Harmonic components are identified from the spectrum, and frequency-domain features, such as fundamental frequency and harmonic amplitude, harmonic factor, and total harmonic distortion (THD), are extracted. The sample values ​​of the frequency-domain features are paired with their corresponding defect types and moisture absorption stages to construct the second sample set.

[0074] The first and second sample sets are input into the gradient boosting model, and the model will give an importance score for each feature, which reflects the importance of the feature to defect detection.

[0075] From multiple time-domain and frequency-domain features, those features with scores higher than a preset threshold are selected to form a preliminary feature set. The selected feature set can be used to retrain the model, and its performance can be evaluated using cross-validation. Based on the model's performance and specific metrics (such as precision and recall), the importance score threshold is adjusted until the optimal feature set is found. The above steps determine the final feature set, which contains important features that significantly impact insulation defect detection.

[0076] Specifically, different types of features are input into the XGBoost algorithm to obtain importance scores. Among the time-domain features, the effective value, mean square value, and maximum value receive higher scores. Among the frequency-domain features, the total harmonic distortion rate and the 3rd-5th harmonic factor score significantly higher than the other features.

[0077] By inputting time-domain and frequency-domain features into the gradient boosting model and scoring their importance, the most influential set of features for detecting insulation defects in pole-mounted switchgear is selected. In other alternative embodiments, the feature set can be further optimized by combining other feature selection methods, such as random forests or support vector machines, to address potential limitations of the gradient boosting model under certain conditions.

[0078] As an optional embodiment, if the defect detection result indicates the presence of a defect, the pulse discharge signal of the target common-enclosure pole-mounted switchgear and the ambient humidity are detected; based on a preset degree determination formula, the degree of defect of the target common-enclosure pole-mounted switchgear is determined according to the target leakage current signal, pulse discharge signal and ambient humidity.

[0079] Optionally, if insulation defects are detected in the common-enclosure pole-mounted switchgear, the severity of the defects can be further assessed. A wideband current sensor is used to continuously collect leakage current signals from the grounding wire of the target common-enclosure pole-mounted switchgear, and the values ​​are recorded. Simultaneously, the device is monitored for pulse discharge phenomena, recording their presence, amplitude, and frequency. The humidity of the environment in which the device is located can also be recorded, using a professional hygrometer for real-time monitoring to ensure data accuracy.

[0080] The leakage current can be assessed and compared with a baseline value during normal operation to determine if it exceeds the normal range. If a pulse discharge is detected, its amplitude and frequency are analyzed to determine the severity of the discharge. The current ambient humidity level is assessed to determine if the equipment is in a risk area that could lead to moisture damage. A formula or algorithm can be pre-defined to quantify the degree of equipment defects by integrating parameters such as leakage current, pulse discharge, and ambient humidity. For example, a weighted summation formula can be used, assigning different weights to each factor based on its impact on insulation defects. Based on historical data and expert experience, different defect severity levels, such as minor, severe, or critical, can be defined for different scoring intervals. The calculated defect severity score is compared with the pre-defined scoring intervals to determine the equipment's defect level. Appropriate measures can then be implemented for different defect levels.

[0081] For example, minor defects: increase inspection frequency and monitor equipment operating status; no immediate action is required. Serious defects: add the equipment to the priority monitoring list, increase inspection frequency, and consider scheduling power outages for maintenance or replacement. Critical defects: take emergency measures, such as immediately stopping equipment operation and conducting disassembly and inspection, to prevent the fault from causing larger power grid safety problems.

[0082] By comprehensively analyzing leakage current, pulse discharge signals, and ambient humidity, the degree of insulation defects in pole-mounted switchgear can be accurately determined, enabling the development of reasonable operation and maintenance strategies to ensure the reliability and safety of the power system. The key lies in the rational construction of the formula and the precise setting of parameter weights to ensure that the score accurately reflects the true condition of the equipment.

[0083] As an optional embodiment, the degree determination formula is determined in the following manner: ,in, As for the degree of defect, The weight corresponding to the target leakage current. The weights corresponding to ambient humidity. The weights corresponding to the amplitude of the pulse discharge signal. The level corresponding to the target leakage current. This represents the humidity level corresponding to the ambient humidity. This refers to the level corresponding to the amplitude of the pulse discharge signal.

[0084] Optionally, a weighted summation formula can be used to determine the degree of defect. The weights reflect the relative importance of each parameter to the equipment condition and defect degree. There are many ways to determine the weights. For example, a data-driven approach can be used, training a model with historical data. The model automatically learns the importance of each parameter, thus deriving its weight. Alternatively, a fault tree model can be constructed to analyze the probability of each factor causing equipment failure, thereby determining the weights. Dimensionality reduction techniques can be used to identify the main components in the data, i.e., the feature vectors of the dataset; the variance contribution rate of these components can be used as weights. Multiple methods can also be combined, such as data-driven methods and expert evaluation, to obtain a more comprehensive weight determination. Regardless of the method used, weight determination requires a large amount of data and a deep understanding of power equipment failure modes. Furthermore, as equipment condition changes and environmental conditions vary, the weights can be updated and adjusted periodically to maintain the validity and accuracy of the evaluation formula. Specifically, the target leakage current level classification can be based on statistical data to set leakage current levels. For example, leakage current can be classified into the following levels:

[0085] Slight moisture (Level I): Leakage current is above the normal range, but does not exceed a certain threshold (e.g., 0.5-1.0mA).

[0086] Severe moisture (Level II): The leakage current exceeds the range of slight moisture, but has not yet reached a critical state (e.g., 1.0-3.0mA).

[0087] Critical condition (Level III): The leakage current is very high, indicating a serious defect in the equipment (e.g., above 3.0mA).

[0088] Environmental humidity level classification involves collecting and analyzing equipment operating data under different humidity conditions, especially humidity-related equipment malfunction data, and then dividing the environmental humidity into several levels. For example:

[0089] Normal (Level I): Ambient humidity is within a safe range (e.g., <80%).

[0090] Minor risk (Level II): Humidity is slightly higher than normal, but has not yet caused significant impact (e.g., 80%-85%).

[0091] High risk (Level III): Humidity reaches a level that may seriously affect the insulation of the equipment (e.g., >85%).

[0092] The pulse discharge amplitude level can be determined based on the pulse discharge amplitude of the equipment under normal conditions and when different levels of defects occur. Then, a threshold value for the pulse discharge amplitude is set to distinguish between different levels of defects, for example:

[0093] No discharge (Level I): Under normal operating conditions, the equipment has no pulse discharge signal.

[0094] Minor discharge (Level II): The pulse discharge amplitude is small, which may indicate that there is a minor partial discharge phenomenon in the equipment (e.g., <5V).

[0095] Severe discharge (Level III): The pulse discharge amplitude is high, which may indicate that the insulating material has been significantly damaged (e.g., ≥5V).

[0096] As an optional embodiment, the degree determination formula is as follows: ,in, As for the degree of defect, For the target leakage current, The amplitude of the pulse discharge signal. The humidity of the surrounding environment.

[0097] Optionally, the severity of the defect can be classified into three levels: general, severe, and critical.

[0098] General moisture defect (Level I): The leakage current reaches 0.5-1.0mA, there is no pulse discharge signal, and the ambient humidity is <80%, indicating that the pole-mounted switch has a slight moisture defect.

[0099] Severe moisture defect (Level II): The leakage current reaches 1.0-3.0mA, the pulse discharge amplitude is <5V, and the ambient humidity is between 80%-85%, indicating that the pole-mounted switch has a moderate moisture defect.

[0100] Critical Moisture Defect (Level III): If the detected leakage current reaches 3.0mA or higher, the pulse discharge amplitude is ≥5V, and the ambient humidity is >85%, it is determined that the pole-mounted switch has a serious moisture defect and the risk of failure is high.

[0101] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, because according to the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0102] Through the above description of the embodiments, those skilled in the art can clearly understand that the defect detection method for the common-enclosure pole-mounted switchgear according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platform. Of course, it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0103] According to embodiments of the present invention, a defect detection device for a common-enclosure pole-mounted switchgear is also provided for implementing the above-described defect detection method for common-enclosure pole-mounted switchgear. Figure 3 This is a structural block diagram of a defect detection device for a common-enclosure pole-mounted switchgear according to an embodiment of the present invention, as shown below. Figure 3 As shown, the defect detection device for the common-enclosure pole-mounted switchgear includes: a data acquisition module 302, an extraction module 304, a selection module 306, and a detection module 308. The defect detection device for the common-enclosure pole-mounted switchgear is described below.

[0104] The acquisition module 302 is used to acquire the target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear.

[0105] The extraction module 304 is connected to the acquisition module 302 and is used to extract the target values ​​corresponding to multiple feature quantities of the target leakage current signal. Among the multiple feature quantities are time-domain feature quantities and frequency-domain feature quantities. The frequency-domain feature quantities include harmonic factors and total harmonic distortion rate.

[0106] The selection module 306, connected to the extraction module 304, is used to select the target value of the feature quantity that matches the preset feature quantity set from the target values ​​of multiple feature quantities corresponding to the target leakage current signal, and use it as the input value. The feature quantity set includes feature quantities whose influence on defect detection meets the preset conditions.

[0107] The detection module 308, connected to the selection module 306, is used to input the input values ​​into the preset target recognition model to obtain the defect detection results of the target common-enclosure pole-mounted switchgear.

[0108] It should be noted that the acquisition module 302, extraction module 304, selection module 306, and detection module 308 mentioned above correspond to steps S202 to S208 in the embodiments. Multiple modules and their corresponding steps implement the same instances and application scenarios, but are not limited to the content disclosed in the above embodiments. It should also be noted that the above modules, as part of the device, can run on the computer terminal 10 provided in the embodiments.

[0109] Embodiments of the present invention may provide a computer device. Optionally, in this embodiment, the computer device may be located in at least one of a plurality of network devices in a computer network. The computer device includes a memory and a processor.

[0110] The memory can be used to store software programs and modules, such as the program instructions / modules corresponding to the defect detection method and device for the shared-enclosure pole-mounted switchgear in this embodiment of the invention. The processor executes various functional applications and data processing by running the software programs and modules stored in the memory, thereby realizing the aforementioned defect detection method for the shared-enclosure pole-mounted switchgear. The memory may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, the memory may further include memory remotely located relative to the processor, and these remote memories can be connected to a computer terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0111] The processor can access the information and application program stored in the memory via the transmission device to perform the following steps: acquiring the target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear; extracting the target values ​​corresponding to multiple feature quantities corresponding to the target leakage current signal, wherein the multiple feature quantities include time-domain feature quantities and frequency-domain feature quantities, and the frequency-domain feature quantities include harmonic factors and total harmonic distortion; selecting the target value corresponding to the feature quantity that matches the preset feature quantity set from the target values ​​corresponding to the multiple feature quantities corresponding to the target leakage current signal, as the input value, wherein the feature quantity set includes feature quantities whose influence on defect detection meets preset conditions; inputting the input value into the preset target recognition model to obtain the defect detection result of the target common-enclosure pole-mounted switchgear.

[0112] Optionally, the processor may also execute program code for the following steps: extracting the target value corresponding to the frequency domain feature quantity corresponding to the target leakage current signal, including: converting the target leakage current signal to the frequency domain based on the Fast Fourier Transform algorithm to obtain the spectrum diagram corresponding to the target leakage current signal; identifying the harmonic components corresponding to the target leakage current signal based on the spectrum diagram corresponding to the target leakage current signal, and determining the target value corresponding to the harmonic factor corresponding to the target leakage current signal; determining the effective harmonic value corresponding to the target leakage current signal based on the harmonic components corresponding to the target leakage current signal; and calculating the target value corresponding to the total harmonic distortion rate corresponding to the target leakage current signal based on the effective harmonic value corresponding to the target leakage current signal.

[0113] Optionally, the processor may also execute program code for the following steps: acquiring sample leakage current signals at the grounding point of the sample common-enclosure pole-mounted switchgear during multiple moisture-affected stages, wherein the sample common-enclosure pole-mounted switchgear includes common-enclosure pole-mounted switchgear corresponding to various defect types; extracting sample values ​​corresponding to multiple feature quantities corresponding to the sample leakage current signals during multiple moisture-affected stages; and training a preset residual network model based on the sample values ​​corresponding to the feature quantities in the feature quantity set to obtain a target recognition model.

[0114] Optionally, the processor may also execute program code for the following steps: training a preset residual network model based on the sample values ​​corresponding to each feature quantity in the feature quantity set to obtain a target recognition model, including: determining a training set based on the sample values ​​corresponding to each feature quantity in the feature quantity set, wherein the training set includes sample values ​​corresponding to each feature quantity labeled with defect types; determining a residual network model based on a preset residual network architecture, wherein the residual network architecture is a ResNet18 architecture or a ResNet35 architecture; and training the residual network model using the training set to obtain a target recognition model.

[0115] Optionally, the processor may also execute program code that performs the following steps: based on the gradient boosting model and the sample values ​​corresponding to multiple feature quantities of the sample leakage current signals of multiple moisture stages, select feature quantities whose influence on defects meets preset conditions from the multiple feature quantities to obtain a feature quantity set.

[0116] Optionally, the processor may also execute program code with the following steps: In the presence of multiple time-domain features and multiple frequency-domain features, based on the gradient boosting model and the sample values ​​corresponding to the multiple features of the leakage current signals of samples at multiple moisture stages, select features whose influence on defects meets preset conditions from the multiple features to obtain a feature set, including: determining a first sample set based on the sample values ​​corresponding to the time-domain features of the leakage current signals of samples at multiple defect types and multiple moisture stages; determining a second sample set based on the sample values ​​corresponding to the frequency-domain features of the leakage current signals of samples at multiple defect types and multiple moisture stages; inputting the first and second sample sets into the gradient boosting model to obtain the importance scores corresponding to the multiple time-domain features and the multiple frequency-domain features, where the importance score characterizes the degree of influence on defect detection; and selecting features whose importance scores meet preset conditions from the multiple time-domain features and multiple frequency-domain features to determine the feature set.

[0117] Optionally, the processor may also execute program code for the following steps: if the defect detection result indicates the presence of a defect, detect the pulse discharge signal and ambient humidity of the target common-enclosure pole-mounted switchgear; based on a preset degree determination formula, determine the degree of defect of the target common-enclosure pole-mounted switchgear according to the target leakage current signal, pulse discharge signal, and ambient humidity.

[0118] Optionally, the degree determination formula in the above processor is determined in the following way: ,in, As for the degree of defect, The weight corresponding to the target leakage current. The weights corresponding to ambient humidity. The weights corresponding to the amplitude of the pulse discharge signal. The level corresponding to the target leakage current. This represents the humidity level corresponding to the ambient humidity. This refers to the level corresponding to the amplitude of the pulse discharge signal.

[0119] Optionally, the degree determination formula in the above processor is as follows: ,in, As for the degree of defect, For the target leakage current, The amplitude of the pulse discharge signal. The humidity of the surrounding environment.

[0120] This invention provides a method for defect detection of a common-enclosure pole-mounted switchgear. The method involves acquiring a target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear; extracting target values ​​corresponding to multiple feature quantities corresponding to the target leakage current signal, including time-domain and frequency-domain feature quantities, with the frequency-domain feature quantities including harmonic factors and total harmonic distortion (THD); selecting the target value corresponding to the feature quantity that matches a preset feature quantity set from the target values ​​corresponding to the multiple feature quantities of the target leakage current signal, as the input value; wherein the feature quantity set includes feature quantities whose influence on defect detection meets preset conditions; and inputting the input value into a preset target recognition model to obtain the defect detection result of the target common-enclosure pole-mounted switchgear. This achieves the goal of timely detection of defects in common-enclosure pole-mounted switchgear based on leakage current, thereby reducing the technical effect of fault risk. It also solves the technical problem that current pole-mounted switchgear suffers from internal moisture due to sealing performance defects, which is difficult to detect, leading to decreased insulation performance and increased fault risk.

[0121] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be implemented by a program instructing the hardware related to the terminal device. The program can be stored in a non-volatile storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0122] Embodiments of the present invention also provide a non-volatile storage medium. Optionally, in this embodiment, the aforementioned non-volatile storage medium can be used to store the program code executed by the defect detection method for the common-enclosure pole-mounted switchgear provided in the above embodiments.

[0123] Optionally, in this embodiment, the non-volatile storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0124] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: acquiring the target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear; extracting the target values ​​corresponding to each of the multiple feature quantities corresponding to the target leakage current signal, wherein the multiple feature quantities include time-domain feature quantities and frequency-domain feature quantities, and the frequency-domain feature quantities include harmonic factors and total harmonic distortion; selecting the target value corresponding to the feature quantity that matches the preset feature quantity set from the target values ​​corresponding to the multiple feature quantities corresponding to the target leakage current signal, as the input value, wherein the feature quantity set includes feature quantities whose influence on defect detection meets preset conditions; inputting the input value into the preset target recognition model to obtain the defect detection result of the target common-enclosure pole-mounted switchgear.

[0125] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: extracting the target value corresponding to the frequency domain feature quantity corresponding to the target leakage current signal, including: converting the target leakage current signal to the frequency domain based on the Fast Fourier Transform algorithm to obtain the spectrum diagram corresponding to the target leakage current signal; identifying the harmonic components corresponding to the target leakage current signal based on the spectrum diagram corresponding to the target leakage current signal, and determining the target value corresponding to the harmonic factor corresponding to the target leakage current signal; determining the effective harmonic value corresponding to the target leakage current signal based on the harmonic components corresponding to the target leakage current signal; and calculating the target value corresponding to the total harmonic distortion rate corresponding to the target leakage current signal based on the effective harmonic value corresponding to the target leakage current signal.

[0126] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: acquiring sample leakage current signals at the grounding point of the sample co-enclosure pole-mounted switchgear during multiple moisture-affected stages, wherein the sample co-enclosure pole-mounted switchgear includes co-enclosure pole-mounted switchgear corresponding to various defect types; extracting sample values ​​corresponding to multiple feature quantities corresponding to the sample leakage current signals during multiple moisture-affected stages; and training a preset residual network model based on the sample values ​​corresponding to the feature quantities in the feature quantity set to obtain a target recognition model.

[0127] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: training a preset residual network model based on the sample values ​​corresponding to each feature quantity in the feature quantity set to obtain a target recognition model, including: determining a training set based on the sample values ​​corresponding to each feature quantity in the feature quantity set, wherein the training set includes sample values ​​corresponding to each feature quantity with labeled defect types; determining a residual network model based on a preset residual network architecture, wherein the residual network architecture is a ResNet18 architecture or a ResNet35 architecture; and training the residual network model using the training set to obtain a target recognition model.

[0128] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: based on the gradient boosting model and the sample values ​​corresponding to multiple feature quantities of the sample leakage current signals of multiple moisture stages, the feature quantities whose influence on the defect meets the preset conditions are selected from the multiple feature quantities to obtain a feature quantity set.

[0129] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: Given multiple time-domain features and multiple frequency-domain features, based on the gradient boosting model and the sample values ​​corresponding to the multiple features of the leakage current signals of samples at multiple moisture stages, a feature set is obtained by selecting features whose influence on defects meets preset conditions from the multiple features, including: determining a first sample set based on the sample values ​​corresponding to the time-domain features of the leakage current signals of samples at multiple defect types and multiple moisture stages; determining a second sample set based on the sample values ​​corresponding to the frequency-domain features of the leakage current signals of samples at multiple defect types and multiple moisture stages; inputting the first and second sample sets into the gradient boosting model to obtain the importance scores corresponding to the multiple time-domain features and the multiple frequency-domain features, wherein the importance score characterizes the degree of influence on defect detection; and selecting features whose importance scores meet preset conditions from the multiple time-domain features and multiple frequency-domain features to determine the feature set.

[0130] Optionally, in this embodiment, the non-volatile storage medium is configured to store program code for performing the following steps: when the defect detection result indicates the presence of a defect, detecting the pulse discharge signal of the target common-enclosure pole-mounted switchgear and the ambient humidity; based on a preset degree determination formula, determining the degree of defect of the target common-enclosure pole-mounted switchgear according to the target leakage current signal, the pulse discharge signal, and the ambient humidity.

[0131] Optionally, in this embodiment, the formula for determining the degree of neutralization of the non-volatile storage medium is determined in the following manner: ,in, As for the degree of defect, The weight corresponding to the target leakage current. The weights corresponding to ambient humidity. The weights corresponding to the amplitude of the pulse discharge signal. The level corresponding to the target leakage current. This represents the humidity level corresponding to the ambient humidity. This refers to the level corresponding to the amplitude of the pulse discharge signal.

[0132] Optionally, in this embodiment, the degree determination formula in the non-volatile storage medium is as follows: ,in, As for the degree of defect, For the target leakage current, The amplitude of the pulse discharge signal. The humidity of the surrounding environment.

[0133] Embodiments of the present invention also provide a computer program product, including a computer program. Optionally, in this embodiment, when the computer program is executed by a processor, it can: acquire the target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear; extract the target values ​​corresponding to each of the multiple feature quantities corresponding to the target leakage current signal, wherein the multiple feature quantities include time-domain feature quantities and frequency-domain feature quantities, and the frequency-domain feature quantities include harmonic factors and total harmonic distortion; select the target value corresponding to the feature quantity that matches the preset feature quantity set from the target values ​​corresponding to the multiple feature quantities corresponding to the target leakage current signal, as the input value, wherein the feature quantity set includes feature quantities whose influence on defect detection meets preset conditions; input the input value into a preset target recognition model to obtain the defect detection result of the target common-enclosure pole-mounted switchgear.

[0134] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0135] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0136] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0137] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0138] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0139] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a non-volatile storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, or optical disks.

[0140] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A defect detection method for a common-enclosure pole-mounted switchgear, characterized in that, include: Collect the target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear; Extract the target values ​​corresponding to each of the multiple feature quantities corresponding to the target leakage current signal, wherein the multiple feature quantities include time-domain feature quantities and frequency-domain feature quantities, and the frequency-domain feature quantities include harmonic factor and total harmonic distortion rate; Select the target value of the feature quantity that matches the preset feature quantity set from the target value of each of the multiple feature quantities corresponding to the target leakage current signal, and use it as the input value. The feature quantity set includes feature quantities whose influence on defect detection meets the preset conditions. The input values ​​are input into a preset target recognition model to obtain the defect detection results of the target common-enclosure pole-mounted switchgear; The set of features is determined through the following steps: Leakage current signals of the grounding point of the sample common-enclosure pole-mounted switchgear were collected during multiple moisture-affected stages. The sample common-enclosure pole-mounted switchgear included common-enclosure pole-mounted switchgear corresponding to various defect types. Extract the sample values ​​corresponding to each of the multiple feature quantities corresponding to the sample leakage current signals of the multiple moisture-affected stages; Set a threshold for feature importance; Based on the gradient boosting model and the sample values ​​corresponding to multiple feature quantities of the sample leakage current signals at multiple moisture stages, the feature importance score corresponding to each of the multiple feature quantities is calculated. Based on the feature importance scores corresponding to each of the multiple feature quantities, they are sorted, and feature quantities with feature importance scores lower than the feature importance threshold are removed to obtain the filtered feature quantity set. The gradient boosting model is trained using the filtered set of features, the feature importance threshold is adjusted, and the gradient boosting model is retrained. The above steps are repeated until the set of features is obtained, wherein the gradient boosting model is an XGBoost model. Specifically, if the defect detection result indicates the presence of a defect, the pulse discharge signal of the target common-enclosure pole-mounted switchgear and the ambient humidity are detected. Based on a preset severity determination formula, the degree of defect of the target common-enclosure pole-mounted switchgear is determined according to the target leakage current signal, the pulse discharge signal, and the ambient humidity. The severity determination formula is determined in the following manner: ， in, The degree of defect, The weight corresponding to the target leakage current. The weight corresponding to the ambient humidity. The weights corresponding to the amplitude of the pulse discharge signal are... The level corresponding to the target leakage current. This refers to the level corresponding to the humidity of the surrounding environment. The level corresponding to the amplitude of the pulse discharge signal.

2. The method according to claim 1, characterized in that, Extracting the target value corresponding to the frequency domain feature quantity of the target leakage current signal, including: Based on the Fast Fourier Transform algorithm, the target leakage current signal is converted to the frequency domain to obtain the spectrum of the target leakage current signal; Based on the spectrum diagram corresponding to the target leakage current signal, the harmonic components corresponding to the target leakage current signal are identified, and the target value of the harmonic factor corresponding to the target leakage current signal is determined. Based on the harmonic components corresponding to the target leakage current signal, the effective value of the harmonics corresponding to the target leakage current signal is determined; Based on the effective harmonic value of the target leakage current signal, calculate the target value corresponding to the total harmonic distortion rate of the target leakage current signal.

3. The method according to claim 1, characterized in that, Also includes: Based on the sample values ​​corresponding to each feature quantity in the feature quantity set, the preset residual network model is trained to obtain the target recognition model.

4. The method according to claim 3, characterized in that, The step of training a preset residual network model based on sample values ​​corresponding to each feature quantity in the feature quantity set to obtain the target recognition model includes: A training set is determined based on the sample values ​​corresponding to each feature in the feature set, wherein the training set includes the sample values ​​corresponding to each feature with labeled defect types. Based on a preset residual network architecture, the residual network model is determined, wherein the residual network architecture is a ResNet18 architecture or a ResNet35 architecture. The residual network model is trained using the training set to obtain the target recognition model.

5. The method according to claim 1, characterized in that, In the presence of multiple time-domain and multiple frequency-domain features, the sample values ​​corresponding to the multiple features based on the gradient boosting model and the sample leakage current signals of multiple moisture-affected stages are used to select features whose influence on defects meets the preset conditions from the multiple features, resulting in the feature set, including: Based on the sample values ​​corresponding to the time-domain feature quantities of the sample leakage current signals of the various defect types and the various moisture-affected stages, a first sample set is determined. Based on the sample values ​​corresponding to the frequency domain feature quantities of the sample leakage current signals of the various defect types and the various moisture stages, a second sample set is determined. The first sample set and the second sample set are input into the gradient boosting model to obtain the importance scores corresponding to each of the multiple time-domain features and the multiple frequency-domain features, wherein the importance scores characterize the degree of influence on defect detection. The feature set is determined by selecting the feature quantities whose importance scores meet the preset conditions from the plurality of time-domain feature quantities and the plurality of frequency-domain feature quantities respectively.

6. The method according to claim 1, characterized in that, The formula for determining the degree is as follows: ， in, The degree of defect, The target leakage current, The amplitude of the pulse discharge signal is [value missing]. The humidity of the environment in question.

7. A defect detection device for a common-enclosure pole-mounted switchgear, characterized in that, include: The acquisition module is used to acquire the target leakage current signal at the grounding point of the target common-enclosure pole-mounted switchgear; An extraction module is used to extract the target values ​​corresponding to each of the multiple feature quantities corresponding to the target leakage current signal. The multiple feature quantities include time-domain feature quantities and frequency-domain feature quantities. The frequency-domain feature quantities include harmonic factors and total harmonic distortion rate. The selection module is used to select the target value of the feature quantity that matches the preset feature quantity set from the target values ​​of the multiple feature quantities corresponding to the target leakage current signal, and use it as the input value. The feature quantity set includes feature quantities whose influence on defect detection meets the preset conditions. The detection module is used to input the input values ​​into a preset target recognition model to obtain the defect detection results of the target common-enclosure pole-mounted switchgear; The device is further configured to determine the set of feature quantities based on the following steps: Leakage current signals of the grounding point of the sample common-enclosure pole-mounted switchgear were collected during multiple moisture-affected stages. The sample common-enclosure pole-mounted switchgear included common-enclosure pole-mounted switchgear corresponding to various defect types. Extract the sample values ​​corresponding to each of the multiple feature quantities corresponding to the sample leakage current signals of the multiple moisture-affected stages; Set a threshold for feature importance; Based on the gradient boosting model and the sample values ​​corresponding to multiple feature quantities of the sample leakage current signals at multiple moisture stages, the feature importance score corresponding to each of the multiple feature quantities is calculated. Based on the feature importance scores corresponding to each of the multiple feature quantities, they are sorted, and feature quantities with feature importance scores lower than the feature importance threshold are removed to obtain the filtered feature quantity set. The gradient boosting model is trained using the filtered set of features, the feature importance threshold is adjusted, and the gradient boosting model is retrained. The above steps are repeated until the set of features is obtained, wherein the gradient boosting model is an XGBoost model. Specifically, if the defect detection result indicates the presence of a defect, the pulse discharge signal of the target common-enclosure pole-mounted switchgear and the ambient humidity are detected. Based on a preset severity determination formula, the degree of defect of the target common-enclosure pole-mounted switchgear is determined according to the target leakage current signal, the pulse discharge signal, and the ambient humidity. The severity determination formula is determined in the following manner: ， in, The degree of defect, The weight corresponding to the target leakage current. The weight corresponding to the ambient humidity. The weights corresponding to the amplitude of the pulse discharge signal are... The level corresponding to the target leakage current. This refers to the level corresponding to the humidity of the surrounding environment. The level corresponding to the amplitude of the pulse discharge signal.

8. A non-volatile storage medium, characterized in that, The non-volatile storage medium includes a stored program, wherein, when the program is executed, it controls the device containing the non-volatile storage medium to perform the defect detection method for the common-enclosure pole-mounted switchgear according to any one of claims 1 to 6.

9. A computer device, characterized in that, include: Memory and processor The memory stores computer programs; The processor is configured to execute a computer program stored in the memory, wherein when the computer program is executed, the processor performs the defect detection method for the common-enclosure pole-mounted switchgear according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the defect detection method for the common-enclosure pole-mounted switchgear as described in any one of claims 1 to 6.

Images