GIS partial discharge pattern recognition method, device, product, medium and equipment

Abstract

The application discloses a GIS partial discharge mode recognition method, device, product, medium and equipment, relates to the partial discharge classification recognition technical field, and the method comprises the steps of acquiring GIS partial discharge data; normalizing the partial discharge data and converting the partial discharge data into a gray-scale image; inputting the gray-scale image into a partial discharge mode recognition model to output a GIS partial discharge mode recognition result; and the partial discharge mode recognition model is obtained by introducing sequentially connected SKNet and coordinate attention mechanisms before the residual connection of a first residual module and a second residual module of a ResNet model framework. The GIS partial discharge mode recognition method, device, product, medium and equipment can effectively improve the accuracy of GIS partial discharge mode recognition.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of partial discharge classification and identification, in particular to a GIS partial discharge pattern identification method, device, product, medium and equipment. BACKGROUND

[0002] Gas Insulated Switchgear (GIS) is a key device widely used in modern power systems, which has a complex structure and a very high working field strength. Once an insulation defect occurs inside the GIS, it is easy to cause equipment failure, leading to large-scale power outages and causing huge economic losses. Insulation failure has always had an important impact on the reliability of GIS operation. Partial discharge, as an early form of insulation failure inside GIS, is a potential threat to the normal operation of GIS. If partial discharge faults can be detected and eliminated as early as possible during equipment operation, it can reduce equipment losses and improve power supply reliability. However, due to the pure sealing structure of GIS, it is difficult to carry out various preventive tests like traditional open-type equipment to find potential defects in time. Masoud Karimi et al. in Iran introduced deep belief network into the field of partial discharge identification, found that the classifier based on deep belief network can learn important distinguishable features from data, and effectively distinguish three types of corona discharge, surface discharge and internal discharge, but this method does not study the fine classification of discharge faults, nor does it classify the discharge intensity (high, medium and low), which limits its judgment of the severity of discharge in practical application. SUMMARY

[0003] The purpose of the present application is to provide a GIS partial discharge pattern identification method, device, product, medium and equipment, which can effectively improve the accuracy of GIS partial discharge pattern identification.

[0004] To achieve the above purpose, the present application provides the following solutions:

[0005] In a first aspect, the present application provides a GIS partial discharge pattern identification method, comprising:

[0006] obtaining partial discharge data of GIS;

[0007] normalizing the partial discharge data and converting it into a grayscale image;

[0008] inputting the grayscale image into a partial discharge pattern identification model to output a partial discharge pattern identification result of GIS; the partial discharge pattern identification model is obtained by introducing SKNet and coordinate attention mechanism in sequence in series before the residual connection of the first residual module and the second residual module of the ResNet model framework.

[0009] Optionally, the acquiring GIS partial discharge data specifically comprises:

[0010] The GIS partial discharge data is acquired by using a UHF sensor to collect phase-resolved partial discharge data of the GIS.

[0011] Optionally, the normalizing and converting the partial discharge data into a gray image specifically comprises:

[0012] The partial discharge data is organized into a two-dimensional data matrix and normalized to obtain a normalized data matrix.

[0013] The data in the normalized data matrix is multiplied by 255 and converted into an integer format, and then converted into a gray image through an image processing library.

[0014] Optionally, the ResNet model comprises a convolutional layer, a max-pooling layer, a first residual module, a second residual module, an average-pooling layer, and a fully connected layer connected in sequence.

[0015] Optionally, the first residual module has a channel number of 64, and the second residual module has a channel number of 128.

[0016] Optionally, the GIS partial discharge pattern recognition result comprises a normal signal without partial discharge, an interference signal, a solid suspension partial discharge high-intensity signal, a solid suspension partial discharge medium-intensity signal, a solid suspension partial discharge low-intensity signal, a sharp-point corona partial discharge high-intensity signal, a sharp-point corona partial discharge medium-intensity signal, a sharp-point corona partial discharge low-intensity signal, a metal particle partial discharge high-intensity signal, a metal particle partial discharge medium-intensity signal, a metal particle partial discharge low-intensity signal, a suspension potential partial discharge high-intensity signal, a suspension potential partial discharge medium-intensity signal, and a suspension potential partial discharge low-intensity signal.

[0017] In a second aspect, the application provides a GIS partial discharge pattern recognition device, comprising:

[0018] A data acquisition module for acquiring GIS partial discharge data;

[0019] A gray image acquisition module for normalizing and converting the partial discharge data into a gray image;

[0020] A partial discharge pattern recognition result output module for inputting the gray image into a partial discharge pattern recognition model and outputting a GIS partial discharge pattern recognition result; the partial discharge pattern recognition model is obtained by introducing a SKNet and a coordinate attention mechanism in sequence before the residual connection of a first residual module and a second residual module of a ResNet model framework.

[0021] In a third aspect, the present application provides a computer program product comprising a computer program which, when executed by a processor, implements the GIS partial discharge pattern recognition method.

[0022] In a fourth aspect, the present application provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, implements the GIS partial discharge pattern recognition method.

[0023] In a fifth aspect, the present application provides a computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor executing the computer program to implement the GIS partial discharge pattern recognition method.

[0024] According to the specific embodiments provided by the present application, the present application has the following technical effects:

[0025] The present application provides a GIS partial discharge pattern recognition method, device, product, medium and equipment, mainly relates to a GIS partial discharge pattern recognition method based on deep learning algorithm, by introducing the SKNet and the coordinate attention mechanism CoordAttention which are sequentially connected before the residual connection of the first residual module and the second residual module of the ResNet model framework, construct the partial discharge pattern recognition model of the present application, use the partial discharge pattern recognition model of the present application to output the partial discharge pattern recognition result of GIS, which can effectively improve the accuracy of GIS partial discharge pattern recognition. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0027] Figure 1 The GIS partial discharge pattern recognition method provided by the present application is shown in the flowchart;

[0028] Figure 2 The ResNet model framework structure provided by the present application is shown in the schematic diagram;

[0029] Figure 3 The SKNet model framework structure provided by the present application is shown in the schematic diagram;

[0030] Figure 4 The implementation process of the coordinate attention mechanism provided by the present application is shown in the schematic diagram;

[0031] Figure 5A classification effect schematic diagram of the ResNet18 model provided in the present application is shown in FIG. 1, wherein, Figure 5 (a) is a fault type classification diagram of the ResNet18 model; Figure 5 (b) is a fault type and intensity classification diagram of the ResNet18 model;

[0032] Figure 6 A classification effect schematic diagram of the ResNet model provided in the present application is shown in FIG. 2, wherein, Figure 6 (a) is a fault type classification diagram of the ResNet model; Figure 6 (b) is a fault type and intensity classification diagram of the ResNet model;

[0033] Figure 7 A classification effect schematic diagram of the partial discharge pattern recognition model provided in the present application is shown in FIG. 3, wherein, Figure 7 (a) is a fault type classification diagram of the partial discharge pattern recognition model; Figure 7 (b) is a fault type and intensity classification diagram of the partial discharge pattern recognition model. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0035] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0036] To solve the technical defects that the discharge fault cannot be classified in detail and the discharge intensity (high, medium and low) cannot be classified in the background art, the application provides a GIS partial discharge pattern recognition method based on deep learning, wherein a partial discharge pattern recognition model is constructed, and the partial discharge data is subjected to deep and feature extraction by using the partial discharge pattern recognition model to classify the partial discharge fault types in the GIS, including 6 data types of normal signals without partial discharge, interference signals, solid suspension partial discharge signals, sharp-point corona partial discharge signals, metal particle partial discharge signals and suspended potential partial discharge signals. On this basis, in order to classify the discharge intensity, each fault type is further classified into high-intensity discharge, medium-intensity discharge and low-intensity discharge, and is classified into 14 patterns. The partial discharge pattern recognition model of the application can realize effective classification of two modes of 6 classification and 14 classification, and the partial discharge pattern recognition model directly outputs one of 6 classifications or one of 14 classifications.

[0037] As shown in Figure 1 , the GIS partial discharge pattern recognition method disclosed by the application comprises:

[0038] Step 1: Obtain the partial discharge data of GIS.

[0039] The application uses an ultrahigh frequency sensor to collect phase-resolved partial discharge (PRPD) data of GIS, obtains the partial discharge data of GIS, also known as PRPD data. The PRPD data is a conventional data form collected for partial discharge equipment, which can effectively reflect the number and intensity of partial discharge under different phase conditions.

[0040] Step 2: Normalize the partial discharge data and convert it into a gray-scale image.

[0041] After the PRPD data is collected by the GIS partial discharge pattern recognition method of the application, the PRPD data is first organized into a two-dimensional data matrix, and in order to effectively convert the data into a gray-scale image, the two-dimensional data matrix is normalized to obtain a normalized data matrix. This operation scales the amplitude of the signal to the range of [0, 1], which facilitates the generation of a gray-scale image subsequently, wherein the normalization formula is:

[0042] (1)

[0043] wherein, is the normalized data matrix; is the PRPD data in the two-dimensional data matrix; and These are the minimum and maximum values ​​of the PRPD data in the two-dimensional data matrix, respectively.

[0044] Normalization enhances the contrast of grayscale images, making signal features more prominent. The normalized data matrix can be converted into a grayscale image using an image processing library. In this step, the data in the normalized data matrix is ​​multiplied by 255 and converted to an integer format (such as uint8) to fit the pixel value range (0 to 255) of the grayscale image, ensuring that the final grayscale image clearly displays the spatial distribution and intensity of partial discharge activity.

[0045] Step 3: Input the grayscale image into the partial discharge pattern recognition model and output the partial discharge pattern recognition result of GIS; the partial discharge pattern recognition model is obtained by introducing a sequentially cascaded SKNet and coordinate attention mechanism before the residual connection of the first residual module and the second residual module in the ResNet model framework.

[0046] After obtaining the grayscale image, the grayscale image is fed into the ResNet model framework, which is as follows: Figure 2 As shown, the structure specifically includes a convolutional layer, a max pooling layer, a first residual module with 64 channels, a second residual module with 128 channels, an average pooling layer, and a fully connected layer, all connected in sequence. After average pooling by the average pooling layer, the fully connected layer converts the data to the classification number dimension for output prediction of the GIS partial discharge pattern recognition result. Specifically, the grayscale image passes through the convolutional layer and the max pooling layer before entering the convolutional module (first residual module and second residual module). To improve the deep learning capability for PRPD data, this application introduces a Selective Kernel Network (SKNet) and a CoordAttention mechanism in the two residual modules. SKNet and CoordAttention are concatenated and placed before the residual connection between the two residual modules.

[0047] For SKNet, its specific structure and feature extraction process are as follows: Figure 3 As shown, the input feature variables in SKNet These are deep features extracted from the first residual module of the ResNet model, first based on feature variables. Perform two forms of grouped convolution and Their convolutional kernel sizes are 3×3 and 5×5, respectively, after... Feature variables are obtained after convolution transformation ,go through Feature variables are obtained after convolution transformation In the partial discharge pattern recognition task, the PRPD data may involve complex spatial and temporal features, such as the intensity of the discharge, the waveform characteristics, and the disturbance of the noise, and the scales of these features are different, so the multi-scale convolution can help to extract features under different receptive fields and effectively capture local and global information. This multi-branch convolution operation makes each branch use different convolution kernel scales for feature extraction, ensuring that the partial discharge pattern recognition model for pattern recognition can process complex and hierarchical signals.

[0048] After that and are element-wise summed to obtain the feature vector :

[0049] (2)

[0050] After the global average pooling operation , the spatial scale of the feature vector is compressed to obtain the channel feature vector :

[0051] (3)

[0052] wherein, represents the global average pooling operation; represents the height of the feature vector ; represents the width of the feature vector ; i , j ) represents the pixel position in the feature vector .

[0053] The partial discharge signal may exhibit different information at different scales, and these information are complementary. By stacking the features extracted by different convolution kernels and performing certain form of feature transformation and fusion, the partial discharge pattern recognition model not only relies on the information in a single receptive field. Partial discharge data often contain both subtle high-frequency features and global pattern information, and the above operations can handle such diverse data features and improve recognition accuracy.

[0054] After that, the feature is obtained by dimensionality reduction operation through the fully connected layer, and then the weights of each feature scale of the feature are obtained by dimensionality reduction and softmax operation , and in the Select stage, they are weighted and summed with and to obtain the final learned feature​ Through the Select operation, SKNet can dynamically select the optimal feature extraction path during training. This adaptive feature selection mechanism enables SKNet to automatically identify the most prominent feature patterns in the discharge signal and amplify these features, thereby improving its ability to distinguish between different discharge types.

[0055] The implementation process of CoordAttention is as follows: Figure 4 As shown, the input to CoordAttention is the final learned features obtained from SKNet. First, the input features Perform average pooling operations on the X and Y axes to obtain features. In width and height The feature maps in two directions are used to obtain accurate location information encoding of the PRPD data. The specific formula is as follows:

[0056] (4)

[0057] in, and They represent Compression features along height and width; express Height; express The width; h , p )and( q , w )express The pixel position in the data, where by Manipulate the compression width direction feature, through Operational compression height direction features.

[0058] The feature maps from two directions, which obtain the global receptive field at the height and width scales, are concatenated together. The number of channels is compressed using a convolutional module to reduce computational cost. The final feature map is obtained after batch normalization and a non-linear activation function. Then, through convolution and the sigmoid function, the feature map is transformed into attention weights along the width and height directions:

[0059] (5)

[0060] in, and Attention weights are assigned to the height (X-axis) and width (Y-axis), respectively. Represents the sigmoid function; and represents the convolution operation along the height and width.

[0061] Finally, the Through the multiplication weighting calculation, the final feature vector with attention weight is obtained:

[0062] (6)

[0063] wherein, represents the learning feature of channel attention, represents the position of the feature vector and the attention weight.

[0064] The coordinate attention mechanism of the application encodes the features in the horizontal and vertical directions independently and then fuses them, so that the partial discharge pattern recognition model has higher flexibility and accuracy in the spatial dimension. The partial discharge pattern often has specific spatial distribution characteristics, and the use of this mechanism can better capture these spatial characteristics, thereby improving the accuracy of classification.

[0065] In summary, the application effectively combines SKNet and the coordinate attention mechanism, and through global fusion of features in different directions, the partial discharge pattern recognition model can better fuse the features of each discharge pattern, improve the feature extraction capability of the partial discharge pattern recognition model for partial discharge data, and enhance the classification effect. In addition, the application uses a simplified ResNet model, which reduces the parameter amount and computational overhead while maintaining high feature extraction capability, and is suitable for deployment in resource-limited environments.

[0066] As a specific embodiment, the application also compares the partial discharge pattern recognition model with the ResNet18 model and the simplified ResNet in various aspects, and the obtained data is shown in Table 1 and Table 2.

[0067] Table 1 Comparison of computational complexity and parameter amount of different models

[0068]

[0069] Table 2 Fault classification accuracy of different models

[0070]

[0071] It can be found from Table 1 that the simplified ResNet model has a significant reduction in model parameters and computational complexity compared to the ResNet18 model, and the lightweight of the model greatly promotes the practical application in power equipment. However, as can be seen from Table 2, although the simplified ResNet significantly reduces the model size, the classification ability of the model for different fault types is also significantly reduced, which restricts the feature learning ability of the model facing complex partial discharge data. The partial discharge pattern recognition model based on the simplified ResNet adds a key attention module, so that the computational complexity and model parameters only increase slightly, but the feature learning ability of the partial discharge pattern recognition model is significantly improved.

[0072] The partial discharge pattern recognition model directly outputs one of the six classifications (or 14 classifications) of fault codes 0-5 (or 0-13). The statistical effect diagram of all predicted classifications of each model in the test case is as shown in Figures 5 to 7 The corresponding relationship between the class label and the pattern type of the partial discharge is as follows:

[0073] (1) Six kinds of partial discharge data: 0-normal signal without generating partial discharge, 1-interference signal, 2-solid suspension partial discharge signal, 3-sharp corona partial discharge signal, 4-metal particle partial discharge signal, and 5-suspended potential partial discharge signal.

[0074] (2) 14 kinds of partial discharge data: 0-normal signal without generating partial discharge, 1-interference signal, 2-solid suspension partial discharge high-intensity signal, 3-solid suspension partial discharge medium-intensity signal, 4-solid suspension partial discharge low-intensity signal, 5-sharp corona partial discharge high-intensity signal, 6-sharp corona partial discharge medium-intensity signal, 7-sharp corona partial discharge low-intensity signal, 8-metal particle partial discharge high-intensity signal, 9-metal particle partial discharge medium-intensity signal, 10-metal particle partial discharge low-intensity signal, 11-suspended potential partial discharge high-intensity signal, 12-suspended potential partial discharge medium-intensity signal, and 13-suspended potential partial discharge low-intensity signal.

[0075] By observing the pattern recognition effect of the ResNet18 model for 6 kinds of partial discharge data, it can be found that there is only a pattern misjudgment between classes 2 and 3, and the characteristic learning and pattern recognition of various partial discharge faults can be basically met. For the pattern recognition of 14 kinds of partial discharge data, the accuracy of more categories is less than 70%, and the classification effect is low. For the simplified ResNet model, the feature extraction capability of the model is significantly reduced, and the pattern recognition effect for 6 kinds of partial discharge data and 14 kinds of partial discharge data is difficult to meet the fault classification demand of actual application. The partial discharge pattern recognition model provided in the application not only improves the key feature learning capability of the model for partial discharge data on the basis of ensuring the lightweight of the model, but also realizes the effective division of various partial discharge data, which proves that the model can correctly identify different fault types and meet the actual application demand of the partial discharge scene.

[0076] Compared with the existing partial discharge pattern recognition technology, the GIS partial discharge pattern recognition method has the following advantages:

[0077] 1) The SKNet is integrated into the residual module of the simplified ResNet model in the application, the flexibility of feature extraction is enhanced through the selective convolution kernel of different receptive fields, and the partial discharge pattern recognition model has more advantages in feature extraction of partial discharge through multi-scale information fusion.

[0078] (2) The GIS partial discharge pattern recognition method can simultaneously pay attention to the global information of space and channel by introducing CoordAttention, and can strengthen feature expression and improve the extraction of complex spatio-temporal features in partial discharge data by combining the position-sensitive coordinate attention mechanism.

[0079] (3) The GIS partial discharge pattern recognition method uses grouped convolution, selective convolution and dimension reduction design in SKNet and CoordAttention, and uses a simplified ResNet model, which effectively reduces the computational complexity and parameter amount while ensuring the performance of the partial discharge pattern recognition model, and is conducive to the lightweight deployment of the model.

[0080] Based on the GIS partial discharge pattern recognition method, the application further provides a GIS partial discharge pattern recognition device, which comprises:

[0081] A data acquisition module for acquiring partial discharge data of GIS.

[0082] A grayscale image acquisition module for normalizing and converting the partial discharge data into a grayscale image.

[0083] The local discharge mode recognition result output module is configured to input the gray image into a local discharge mode recognition model to output a local discharge mode recognition result of the GIS.

[0084] In some embodiments, the present application also provides a computer program product comprising a computer program which, when executed by a processor, implements the GIS local discharge mode recognition method.

[0085] In some embodiments, the present application also provides a computer readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the GIS local discharge mode recognition method.

[0086] In some embodiments, the present application also provides a computer device comprising a processor, a memory, an input / output interface (I / O), a communication interface, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the GIS local discharge mode recognition method. Specifically, the running device of the GIS local discharge mode recognition method provided by the present application is a computer, the processor version is i5-12600KF, and the graphics card version is RTX-3060Ti.

[0087] The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store transactions to be processed. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through network connection. The computer program is executed by the processor to implement the GIS local discharge mode recognition method.

[0088] It should be noted that the object information (including but not limited to object device information, object personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the object or fully authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards.

[0089] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0090] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0091] The principles and implementation modes of the present application are described by using specific examples in this paper, and the above-mentioned embodiments are only used to help understand the method and its core idea; at the same time, for those skilled in the art, according to the idea of the present application, the specific implementation mode and application range will be changed. In conclusion, the content of the present application should not be understood as a limitation.

Claims

1. A method for partial discharge pattern recognition in GIS, characterized in that, include: Obtain partial discharge data from GIS; The partial discharge data is normalized and converted into a grayscale image; A grayscale image is input into a partial discharge pattern recognition model, and the partial discharge pattern recognition result of GIS is output. The partial discharge pattern recognition model is obtained by introducing a sequentially cascaded SKNet and coordinate attention mechanism inside the first residual module of the ResNet model framework. Specifically, the SKNet and the coordinate attention mechanism are embedded after the convolution operation in the first residual module and before the second residual module. Integrating SKNet into the residual module of the ResNet model enhances feature extraction through selective convolutional kernels with different receptive fields; By introducing a coordinate attention mechanism, the partial discharge pattern recognition model simultaneously focuses on global information in both space and channel, and combines a position-sensitive coordinate attention mechanism to enhance feature representation.

2. The GIS partial discharge pattern recognition method according to claim 1, characterized in that, The acquisition of partial discharge data from GIS specifically includes: A UHF sensor was used to collect phase-resolved partial discharge data of GIS to obtain partial discharge data of GIS.

3. The GIS partial discharge pattern recognition method according to claim 1, characterized in that, The process of normalizing the partial discharge data and converting it into a grayscale image specifically includes: The partial discharge data is organized into a two-dimensional data matrix and then normalized to obtain the normalized data matrix. The data in the normalized data matrix is ​​multiplied by 255 and converted to integer format, and then converted to grayscale image using an image processing library.

4. The GIS partial discharge pattern recognition method according to claim 1, characterized in that, The ResNet model comprises a convolutional layer, a max pooling layer, a first residual module, a second residual module, an average pooling layer, and a fully connected layer connected in sequence.

5. The GIS partial discharge pattern recognition method according to claim 4, characterized in that, The first residual module has 64 channels; the second residual module has 128 channels.

6. The GIS partial discharge pattern recognition method according to claim 1, characterized in that, The partial discharge pattern recognition results of the GIS include: normal signals without partial discharge, interference signals, high-intensity signals of solid suspended partial discharge, medium-intensity signals of solid suspended partial discharge, low-intensity signals of solid suspended partial discharge, high-intensity signals of tip corona partial discharge, medium-intensity signals of tip corona partial discharge, low-intensity signals of tip corona partial discharge, high-intensity signals of metal particle partial discharge, medium-intensity signals of metal particle partial discharge, low-intensity signals of metal particle partial discharge, high-intensity signals of suspended potential partial discharge, medium-intensity signals of suspended potential partial discharge, and low-intensity signals of suspended potential partial discharge.

7. A GIS partial discharge pattern recognition device, characterized in that, include: The data acquisition module is used to acquire partial discharge data from the GIS. The grayscale image acquisition module is used to normalize the partial discharge data and convert it into a grayscale image; The partial discharge pattern recognition result output module is used to input grayscale images into the partial discharge pattern recognition model and output the partial discharge pattern recognition results of GIS. The partial discharge pattern recognition model is specifically obtained by introducing sequentially cascaded SKNet and coordinate attention mechanism inside the first residual module of the ResNet model framework. The SKNet and the coordinate attention mechanism are embedded after the convolution operation in the first residual module and before the second residual module. Integrating SKNet into the residual module of the ResNet model enhances feature extraction through selective convolutional kernels with different receptive fields; By introducing a coordinate attention mechanism, the partial discharge pattern recognition model simultaneously focuses on global information in both space and channel, and combines a position-sensitive coordinate attention mechanism to enhance feature representation.

8. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the GIS partial discharge pattern recognition method according to any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the GIS partial discharge pattern recognition method according to any one of claims 1-6.

10. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the GIS partial discharge pattern recognition method according to any one of claims 1-6.

Images