Intelligent power grid operation and maintenance management platform and method

Abstract

The invention relates to the technical field of intelligent management, and particularly discloses an intelligent power grid operation and maintenance management platform and method, and the method employs an artificial intelligence technology based on deep learning to carry out the image semantic analysis of a surface state image of power grid equipment, so as to capture the surface state features of the power grid equipment. Equipment appearance abnormal information in the image is mined, so that whether the surface of the power grid equipment is corroded or not is intelligently identified. Therefore, the surface damage of the power grid equipment can be found and processed in time, and equipment operation and maintenance personnel can carry out repair and maintenance conveniently, so that the normal operation of the equipment is ensured, and the operation and maintenance efficiency and safety of the intelligent power grid are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent management technology, and more specifically, to a smart grid operation and maintenance management platform and method. Background Art

[0002] With the rapid development of society and the continuous growth of electricity demand, smart grid has become an important part of the modern power system. Smart grid not only improves the transmission efficiency of electric energy, but also realizes real-time monitoring and precise control of the grid, thus ensuring the stable supply of electric energy. However, with the continuous expansion of the scale of smart grid and the increasing number of equipment, the operation and maintenance management of equipment in smart grid still faces many challenges.

[0003] Traditional power grid operation and maintenance management mainly relies on manual inspection and regular maintenance. This manual inspection method requires operation and maintenance personnel to go to the location of the equipment for inspection, which is not only inefficient, but also easily limited by factors such as manpower, material resources and time. In addition, due to the wide variety of power grid equipment, wide distribution, and complex operating environment, traditional operation and maintenance management methods are difficult to meet the efficient, safe and reliable operation requirements of modern smart grids. Therefore, a smart grid operation and maintenance management platform and method are expected. Summary of the invention

[0004] In order to solve the above technical problems, this application is proposed. The embodiment of the present application provides a smart grid operation and maintenance management platform and method, which uses artificial intelligence technology based on deep learning to perform image semantic analysis on the surface state image of the power grid equipment to capture the surface state characteristics of the power grid equipment and mine the abnormal appearance information of the equipment in the image, so as to intelligently identify whether the surface of the power grid equipment is corroded. In this way, the surface damage of the power grid equipment can be discovered and handled in time, which is convenient for the equipment operation and maintenance personnel to carry out repairs and maintenance, thereby ensuring the normal operation of the equipment and improving the operation and maintenance efficiency and safety of the smart grid.

[0005] Accordingly, according to one aspect of the present application, a smart grid operation and maintenance management platform is provided, which includes:

[0006] The device surface image acquisition module is used to obtain the surface state image of the monitored power grid device collected by the unmanned aerial vehicle inspection aircraft, wherein the monitored power grid device is a transformer;

[0007] An image semantic feature mining module, used for performing image semantic feature extraction and global feature clustering on the surface state image to obtain a power grid equipment surface state clustering feature matrix;

[0008] A monitoring result generating module, used for determining the monitoring result based on the grid equipment surface state clustering feature matrix;

[0009] The response and operation and maintenance prompt module is used to generate operation and maintenance management prompt information in response to the monitoring result that the surface of the monitored power grid equipment is corroded.

[0010] In the above-mentioned smart grid operation and maintenance management platform, the image semantic feature mining module includes: an image data enhancement unit, which is used to pass the surface state image through the data enhancement module to obtain a set of enhanced surface state images; an equipment surface state feature extraction unit, which is used to pass each enhanced surface state image in the set of enhanced surface state images through a surface state feature extractor based on a convolutional neural network model to obtain a set of grid equipment surface state feature maps; a feature distillation unit, which is used to pass each grid equipment surface state feature map in the set of grid equipment surface state feature maps through a feature distillation network to obtain a set of distilled grid equipment surface state feature vectors; a global feature clustering unit, which is used to pass the set of distilled grid equipment surface state feature vectors through a joint clustering analysis network to obtain the grid equipment surface state clustering feature matrix.

[0011] In the above-mentioned smart grid operation and maintenance management platform, the data enhancement module is used to rotate and preprocess the surface state image, and the preprocessing includes image noise reduction and color transformation.

[0012] In the above-mentioned smart grid operation and maintenance management platform, the feature distillation unit is used to: expand the feature map of the surface state feature map of the power grid equipment to obtain a sequence of surface state feature vectors of the power grid equipment; process the sequence of surface state feature vectors of the power grid equipment with the following feature distillation formula to obtain the distilled surface state feature vector of the power grid equipment; wherein the feature distillation formula is:

[0013]

[0014] Among them, v i is the ith power grid device surface state feature vector in the sequence of power grid device surface state feature vectors, v j is the jth grid device surface state feature vector in the sequence of grid device surface state feature vectors, ‖·‖1 represents the 1-norm of the feature vector, M is the length of the sequence of grid device surface state feature vectors -1, V k is the representation of the sequence of the surface state feature vectors of the power grid equipment, D i represents the characteristic difference coefficient, exp(·) represents the natural exponential function operation, N s represents the total number of characteristic difference coefficients, P k is the surface state feature vector of the distillation power grid device.

[0015] In the above-mentioned smart grid operation and maintenance management platform, the global feature clustering unit is used to: calculate the relevant weights of any two distillation power grid equipment surface state feature vectors in the set of the distillation power grid equipment surface state feature vectors using the following adjacency association formula to form an adjacency matrix composed of multiple such relevant weights; wherein the adjacency association formula is:

[0016]

[0017] Among them, v i is the i-th distillation power grid device surface state feature vector in the set of distillation power grid device surface state feature vectors, v j is the jth distillation power grid device surface state feature vector in the set of distillation power grid device surface state feature vectors, σ is the stretching factor, ‖·‖2 2 represents the square of the vector norm, exp(·) represents the exponential operation, and W i,j is the correlation weight between the i-th distillation power grid equipment surface state feature vector and the j-th distillation power grid equipment surface state feature vector; calculates the degree matrix of the set of the distillation power grid equipment surface state feature vectors; calculates the Laplace matrix based on the adjacency matrix and the degree matrix; standardizes the Laplace matrix to obtain a standardized Laplace matrix; arranges each eigenvalue in the standardized Laplace matrix in descending order, takes the first K eigenvalues, and decomposes the standardized Laplace matrix based on the first K eigenvalues ​​to obtain K decomposed eigenvectors, wherein K is the number of the distillation power grid equipment surface state feature vectors; after standardizing the K decomposed eigenvectors to obtain K standardized decomposed eigenvectors, the K standardized decomposed eigenvectors are reconstructed into the power grid equipment surface state clustering feature matrix.

[0018] In the above-mentioned smart grid operation and maintenance management platform, the monitoring result generation module includes: an equipment surface state identification unit, which is used to cluster the feature matrix of the surface state of the power grid equipment through a classifier-based surface state monitor to obtain the monitoring result, and the monitoring result is used to indicate whether the surface of the monitored power grid equipment is corroded.

[0019] The above-mentioned smart grid operation and maintenance management platform also includes a training module for training the data enhancement module, the surface state feature extractor based on the convolutional neural network model, the feature distillation network, the joint clustering analysis network and the classifier-based surface state monitor.

[0020] In the above-mentioned smart grid operation and maintenance management platform, the training module includes: a training data acquisition unit, which is used to acquire training data, wherein the training data includes a training surface state image of the monitored power grid equipment collected by an unmanned aerial vehicle inspection aircraft, and a true value of whether the surface of the monitored power grid equipment is corroded, wherein the monitored power grid equipment is a transformer; a training image data enhancement unit, which is used to pass the training surface state image through the data enhancement module to obtain a set of training enhanced surface state images, wherein the data enhancement module is used to rotate and preprocess the training surface state image, and the preprocessing includes image denoising and color transformation; a training surface state extraction unit, which is used to pass each training enhanced surface state image in the set of training enhanced surface state images through the surface state feature extractor based on the convolutional neural network model to obtain a set of training power grid equipment surface state feature maps; a training feature distillation unit, which is used to extract the training surface state image from the training power grid equipment surface state feature map; Each training power grid equipment surface state feature map in the set of state feature maps passes through the feature distillation network to obtain a set of training distilled power grid equipment surface state feature vectors; a training feature clustering analysis unit is used to pass the set of training distilled power grid equipment surface state feature vectors through the joint clustering analysis network to obtain a training power grid equipment surface state clustering feature matrix; a classification loss estimation unit is used to pass the training power grid equipment surface state clustering feature matrix through the classifier-based surface state monitor to obtain a classification loss function value; a training unit is used to train the data enhancement module, the surface state feature extractor based on the convolutional neural network model, the feature distillation network, the joint clustering analysis network and the classifier-based surface state monitor with the classification loss function value, wherein, in each round of iteration of the training, the training power grid equipment surface state clustering feature vector obtained after the training power grid equipment surface state clustering feature matrix is ​​expanded is optimized.

[0021] According to another aspect of the present application, a smart grid operation and maintenance management method is provided, which includes:

[0022] Acquire a surface state image of a monitored power grid device collected by an unmanned aerial vehicle inspection aircraft, wherein the monitored power grid device is a transformer;

[0023] Performing image semantic feature extraction and global feature clustering on the surface state image to obtain a power grid equipment surface state clustering feature matrix;

[0024] Determining monitoring results based on the grid equipment surface state clustering feature matrix;

[0025] In response to the monitoring result that the surface of the monitored power grid equipment is corroded, operation and maintenance management prompt information is generated.

[0026] Compared with the prior art, the smart grid operation and maintenance management platform and method provided by the present application uses deep learning-based artificial intelligence technology to perform image semantic analysis on the surface state image of the power grid equipment to capture the surface state characteristics of the power grid equipment and mine the abnormal appearance information of the equipment in the image, thereby intelligently identifying whether the surface of the power grid equipment is corroded. In this way, surface damage of the power grid equipment can be discovered and handled in a timely manner, making it easier for equipment operation and maintenance personnel to perform repairs and maintenance, thereby ensuring the normal operation of the equipment and improving the operation and maintenance efficiency and safety of the smart grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0028] Figure 1 A block diagram of a smart grid operation and maintenance management platform according to an embodiment of the present application.

[0029] Figure 2 Schematic diagram of the architecture of the smart grid operation and maintenance management platform according to an embodiment of the present application.

[0030] Figure 3 It is a block diagram of an image semantic feature mining module in a smart grid operation and maintenance management platform according to an embodiment of the present application.

[0031] Figure 4 It is a block diagram of the training module in the smart grid operation and maintenance management platform according to an embodiment of the present application.

[0032] Figure 5 Flow chart of a smart grid operation and maintenance management method according to an embodiment of the present application. DETAILED DESCRIPTION

[0033] Below, the embodiments of the present application will be described in more detail in conjunction with the accompanying drawings, and the above-mentioned and other purposes, features and advantages of the present application will become more obvious. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the example embodiments described herein. At the same time, the accompanying drawings are used to provide a further understanding of the embodiments of the present application, and constitute a part of the specification, and are used together with the embodiments of the present application to explain the present application, and do not constitute a limitation on the present application. In the drawings, the same reference numerals generally represent the same parts or steps.

[0034] Figure 1FIG. 1 is a block diagram of a smart grid operation and maintenance management platform according to an embodiment of the present application. Figure 1 As shown, according to the smart grid operation and maintenance management platform 100 of the embodiment of the present application, it includes: an equipment surface image acquisition module 110, which is used to obtain the surface status image of the monitored power grid equipment collected by the unmanned aerial vehicle inspection aircraft, wherein the monitored power grid equipment is a transformer; an image semantic feature mining module 120, which is used to extract image semantic features and perform global feature clustering on the surface status image to obtain a power grid equipment surface status clustering feature matrix; a monitoring result generation module 130, which is used to determine the monitoring result based on the power grid equipment surface status clustering feature matrix; a response and operation and maintenance prompt module 140, which is used to generate operation and maintenance management prompt information in response to the monitoring result that the surface of the monitored power grid equipment is corroded.

[0035] As mentioned in the background technology above, traditional power grid operation and maintenance management mainly relies on manual inspection and regular maintenance. This manual inspection method requires operation and maintenance personnel to go to the location of the equipment for inspection, which is not only inefficient, but also easily limited by factors such as manpower, material resources and time. In addition, as the scale of the power grid expands and the number of devices increases, traditional operation and maintenance management methods are gradually unable to meet the efficient, safe and reliable operation requirements of modern smart grids.

[0036] In response to the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to perform image semantic analysis on the surface state image of power grid equipment to capture the surface state characteristics of power grid equipment and mine the abnormal appearance information of the equipment in the image, so as to intelligently identify whether the surface of the power grid equipment is corroded. In this way, surface damage of power grid equipment can be discovered and handled in a timely manner, which is convenient for equipment operation and maintenance personnel to carry out repairs and maintenance, thereby ensuring the normal operation of the equipment and improving the operation and maintenance efficiency and safety of the smart grid.

[0037] In the above-mentioned smart grid operation and maintenance management platform 100, the device surface image acquisition module 110 is used to obtain the surface state image of the monitored power grid equipment collected by the unmanned aerial vehicle inspection aircraft, wherein the monitored power grid equipment is a transformer. It should be understood that compared with the traditional manual inspection method, the unmanned aerial vehicle inspection aircraft can efficiently cover a large range of transformers, quickly obtain the surface state image of the transformer, capture the detailed equipment surface state information, and realize fast and comprehensive equipment surface state monitoring, thereby improving the efficiency and timeliness of equipment operation and maintenance. In addition, the use of unmanned aerial vehicle inspections can reduce the safety risks to personnel, especially in high-risk or difficult-to-access areas, such as high-voltage transmission lines. By collecting equipment images by drones, people can avoid direct contact with power grid equipment and improve safety.

[0038] In the above-mentioned smart grid operation and maintenance management platform 100, the image semantic feature mining module 120 is used to extract image semantic features and perform global feature clustering on the surface state image to obtain a grid equipment surface state clustering feature matrix. Specifically, Figure 3 FIG. 1 is a block diagram of an image semantic feature mining module in a smart grid operation and maintenance management platform according to an embodiment of the present application. Figure 3 As shown, the image semantic feature mining module 120 includes: an image data enhancement unit 121, which is used to pass the surface state image through a data enhancement module to obtain a set of enhanced surface state images; an equipment surface state feature extraction unit 122, which is used to pass each enhanced surface state image in the set of enhanced surface state images through a surface state feature extractor based on a convolutional neural network model to obtain a set of power grid equipment surface state feature maps; a feature distillation unit 123, which is used to pass each power grid equipment surface state feature map in the set of power grid equipment surface state feature maps through a feature distillation network to obtain a set of distilled power grid equipment surface state feature vectors; a global feature clustering unit 124, which is used to pass the set of distilled power grid equipment surface state feature vectors through a joint clustering analysis network to obtain the power grid equipment surface state clustering feature matrix.

[0039] Specifically, the image data enhancement unit 121 is used to pass the surface state image through the data enhancement module to obtain a set of enhanced surface state images. In a specific example of the present application, the data enhancement module is used to rotate and preprocess the surface state image, and the preprocessing includes image noise reduction and color transformation. It should be understood that, considering that in practical applications, due to differences in acquisition conditions, device status and environmental factors, the acquired device surface state image may have problems such as noise, uneven lighting, color distortion, etc., resulting in low image quality, thereby affecting the accuracy of subsequent device appearance detection. Therefore, in order to improve the image quality, the surface state image is further rotated, image noise is reduced and color transformation is performed. Specifically, by the rotation operation, the diversity of the image can be increased and the generalization ability of the model can be improved. By removing the noise interference in the surface state image through the image noise reduction technology, the image quality can be improved and the image clarity can be improved. By performing color correction, color enhancement and other operations on the surface state image through the color transformation technology, the contrast of the image can be improved, the image color can be made more realistic, and the visualization effect of the image can be improved. In this way, by performing these data enhancement operations to expand the data set and generate a collection of enhanced surface state images, the diversity of the data is increased, the generalization ability of the model is improved, and the risk of overfitting is reduced.

[0040] Specifically, the device surface state feature extraction unit 122 is used to pass each enhanced surface state image in the set of enhanced surface state images through a surface state feature extractor based on a convolutional neural network model to obtain a set of surface state feature maps of power grid equipment. It should be understood that as a deep learning model, convolutional neural network (CNN) performs well in computer vision tasks such as image recognition and target detection. In the technical solution of the present application, a CNN model is used to construct a surface state feature extractor, which is used to extract image feature information from each enhanced surface state image in the set of enhanced surface state images. Specifically, the surface state feature extractor based on the convolutional neural network model uses network structures such as convolutional layers and pooling layers to operate on the enhanced surface state images, and extracts features from each enhanced surface state image, which can effectively extract high-level abstract feature representations from the original image data, gradually capture features such as edges, textures, and colors in the image, and mine detailed surface state information of the transformer equipment, providing strong support for subsequent abnormal surface state identification of the equipment.

[0041] Specifically, the feature distillation unit 123 is used to pass each power grid equipment surface state feature map in the set of power grid equipment surface state feature maps through a feature distillation network to obtain a set of distilled power grid equipment surface state feature vectors. It should be understood that considering that the surface state feature map of the power grid equipment may contain irrelevant background and redundant feature information, it may affect the subsequent abnormality recognition accuracy. Therefore, in order to further improve the effectiveness and computational efficiency of the features, in the technical solution of the present application, a feature distillation network is introduced to process each power grid equipment surface state feature map in the set of power grid equipment surface state feature maps. Specifically, feature distillation is a feature fusion and dimensionality reduction technology, which calculates the differences between the local features of each position in the power grid equipment surface state feature map, screens out a more compact and meaningful feature representation, suppresses irrelevant background and redundant information, and converts high-dimensional feature maps into a more compact feature vector representation, so that the features are more distinguishable and expressive, and provide a more accurate and efficient feature representation for subsequent equipment appearance abnormality recognition.

[0042] In a specific example of the present application, the feature distillation unit 123 is used to: perform feature map expansion on the surface state feature map of the power grid equipment to obtain a sequence of surface state feature vectors of the power grid equipment; process the sequence of surface state feature vectors of the power grid equipment using the following feature distillation formula to obtain the distilled surface state feature vector of the power grid equipment; wherein the feature distillation formula is:

[0043]

[0044] Among them, vi is the ith power grid device surface state feature vector in the sequence of power grid device surface state feature vectors, v j is the jth grid device surface state feature vector in the sequence of grid device surface state feature vectors, ‖·‖1 represents the 1-norm of the feature vector, M is the length of the sequence of grid device surface state feature vectors -1, V k is the representation of the sequence of the surface state feature vectors of the power grid equipment, D i represents the characteristic difference coefficient, exp(·) represents the natural exponential function operation, N s represents the total number of characteristic difference coefficients, P k is the surface state feature vector of the distillation power grid device.

[0045] Specifically, the global feature clustering unit 124 is used to pass the set of surface state feature vectors of the distillation power grid equipment through a joint clustering analysis network to obtain the surface state clustering feature matrix of the power grid equipment. It should be understood that through the above-mentioned rotation, image denoising and color transformation processing, multiple enhanced surface state images are generated, and the surface state of the equipment has certain differences and similarities in different images. In the technical solution of the present application, in order to learn the invariant features of the transformer equipment under these image transformation operations to improve the accuracy of the model for equipment appearance recognition, a joint clustering analysis network is further introduced to analyze the set of surface state feature vectors of the distillation power grid equipment. Specifically, the joint clustering analysis network learns the potential correlation between the surface state feature vectors of each distillation power grid equipment, and performs cluster analysis on the set of surface state feature vectors of the distillation power grid equipment to extract the shared features between the surface state feature vectors of each distillation power grid equipment, so as to better understand the abnormal surface state of the transformer equipment and improve the accuracy and stability of equipment appearance abnormality recognition.

[0046] In a specific example of the present application, the global feature clustering unit 124 is used to calculate the relevant weights of any two distillation power grid equipment surface state feature vectors in the set of the distillation power grid equipment surface state feature vectors using the following adjacency association formula to form an adjacency matrix composed of multiple such relevant weights; wherein the adjacency association formula is:

[0047]

[0048] Among them, v i is the i-th distillation power grid device surface state feature vector in the set of distillation power grid device surface state feature vectors, v j is the jth distillation power grid device surface state feature vector in the set of distillation power grid device surface state feature vectors, σ is the stretching factor, ‖·‖22 represents the square of the vector norm, exp(·) represents the exponential operation, and W i,j is the correlation weight between the i-th distillation power grid equipment surface state feature vector and the j-th distillation power grid equipment surface state feature vector; calculates the degree matrix of the set of the distillation power grid equipment surface state feature vectors; calculates the Laplace matrix based on the adjacency matrix and the degree matrix; standardizes the Laplace matrix to obtain a standardized Laplace matrix; arranges each eigenvalue in the standardized Laplace matrix in descending order, takes the first K eigenvalues, and decomposes the standardized Laplace matrix based on the first K eigenvalues ​​to obtain K decomposed eigenvectors, wherein K is the number of the distillation power grid equipment surface state feature vectors; after standardizing the K decomposed eigenvectors to obtain K standardized decomposed eigenvectors, the K standardized decomposed eigenvectors are reconstructed into the power grid equipment surface state clustering feature matrix.

[0049] In the above-mentioned smart grid operation and maintenance management platform 100, the monitoring result generation module 130 is used to determine the monitoring result based on the grid equipment surface state clustering feature matrix. Specifically, the monitoring result generation module includes: an equipment surface state recognition unit, which is used to pass the grid equipment surface state clustering feature matrix through a surface state monitor based on a classifier to obtain the monitoring result, and the monitoring result is used to indicate whether the surface of the monitored grid equipment is corroded. It should be understood that corrosion is one of the main causes of damage and failure of grid equipment. In the technical solution of the present application, by using a large number of grid equipment surface state images of known states for training, the classifier can learn the classification boundary between the normal surface of the transformer equipment and the corroded surface. Then, the trained classifier is used as a surface state monitor to perform feature learning on the grid equipment surface state clustering feature matrix. The classifier can classify and map the grid equipment surface state clustering feature matrix according to the classification boundary learned during the training process, intelligently judge whether there is a corrosion problem on the surface of the grid equipment, and output the corresponding monitoring result.

[0050] In the technical solution described above, each distilled power grid equipment surface state feature vector in the set of distilled power grid equipment surface state feature vectors respectively expresses the image semantic features after feature distillation of each enhanced surface state image. However, considering the change in the semantic spatial distribution of the image source after the surface state image passes through the data enhancement module, each distilled power grid equipment surface state feature vector will also have feature distribution misalignment, so that after the set of distilled power grid equipment surface state feature vectors passes through the joint clustering analysis network, the power grid equipment surface state clustering feature matrix obtained will have inconsistencies between the local spatial distributions of the feature matrix spatial distribution due to the clustering differences between the local feature distributions of the set of distilled power grid equipment surface state feature vectors. That is, the power grid equipment surface state clustering feature matrix has a lack of class regression interpretability, thereby affecting the accuracy of the classification results. Based on this, the present application considers optimizing the power grid equipment surface state clustering feature matrix.

[0051] In a preferred embodiment of the present application, the monitoring result generation module includes: optimizing the power grid equipment surface state clustering feature matrix to obtain an optimized power grid equipment surface state clustering feature vector; passing the optimized power grid equipment surface state clustering feature vector through a classifier-based surface state monitor to obtain the monitoring result.

[0052] Specifically, the grid equipment surface state clustering feature matrix is ​​optimized to obtain an optimized grid equipment surface state clustering feature vector, comprising the steps of:

[0053] The autocorrelation matrix of the grid equipment surface state clustering feature vectors obtained after the grid equipment surface state clustering feature matrix is ​​expanded is calculated, that is:

[0054]

[0055] Wherein, V represents the clustering feature vector of the surface state of the power grid equipment, T represents the transpose of the vector, Represents matrix multiplication, M c represents the autocorrelation matrix;

[0056] The grid equipment surface state clustering feature vector is multiplied by the self-correlation matrix in a query manner to obtain an intermediate feature vector, namely:

[0057]

[0058] Among them, V c represents the clustering feature vector of the surface state of the power grid equipment, M c represents the autocorrelation matrix, represents matrix multiplication, V ′represents the intermediate eigenvector;

[0059] The inner product and L2 distance of the i-th row vector and the j-th row vector of the autocorrelation matrix are calculated respectively to obtain a first weight matrix M1 and a second weight matrix M2, which are expressed as:

[0060]

[0061] in, and represents the i-th row vector and the j-th row vector of the autocorrelation matrix, T represents the transpose of the vector, Represents matrix multiplication, M c represents the autocorrelation matrix, M1 represents the first weight matrix, m 1-i,j represents the eigenvalue of the (i, j)th position of the first weight matrix, M2 represents the second weight matrix, and m 2-i,j represents the eigenvalue of the (i, j)th position of the second weight matrix, Indicates the L2 distance of the calculated vector;

[0062] After matrix multiplication of the first weight matrix M1 and the second weight matrix M2, further matrix multiplication with the intermediate eigenvector is performed to obtain an optimized grid equipment surface state clustering eigenvector, that is:

[0063]

[0064] Where M1 represents the first weight matrix, M2 represents the second weight matrix, V ′ represents the intermediate eigenvector, represents matrix multiplication, V ″ Represents the optimized grid equipment surface state clustering feature vector.

[0065] That is, for the bit-by-bit scattering association response space of the eigenvalues ​​of the grid equipment surface state clustering feature matrix, a deep embedding structure in a high-order association space is obtained through the joint embedding space of a vector inner product embedding space and a vector independent embedding space, and the decomposable association response sub-dimensions of the bit-by-bit scattering association response space of the eigenvalues ​​of the grid equipment surface state clustering feature matrix are represented as a set of spatial sub-dimensions to further perform dependent spatial mapping of the decomposable association response sub-dimensions of the deep embedding structure of the grid equipment surface state clustering feature matrix, so as to improve the interpretability of the solution space of the regression solution of the grid equipment surface state clustering feature matrix based on the association distribution through the complex dimensional representation of the association response, thereby improving the accuracy of the classification result obtained by inputting the grid equipment surface state clustering feature matrix into the classifier.

[0066] In the above-mentioned smart grid operation and maintenance management platform 100, the response and operation and maintenance prompt module 140 is used to generate operation and maintenance management prompt information in response to the monitoring result that the surface of the monitored power grid equipment is corroded. That is, when it is monitored that there is a corrosion problem on the surface of the power grid equipment, the operation and maintenance management prompt information is automatically generated to remind the operation and maintenance management personnel to deal with it in time. Specifically, the operation and maintenance management prompt information may include the specific location of the corrosion, the degree of corrosion, the recommended treatment measures and other contents, to help the operation and maintenance management personnel quickly and accurately locate the problem and take corresponding treatment measures, so as to avoid further deterioration of equipment damage and ensure the normal operation and safety and stability of the power grid equipment. In addition, according to actual needs, the operation and maintenance management prompt information can also be pushed to relevant operation and maintenance management personnel through channels such as SMS, email, APP, etc., to achieve timely transmission and processing of information.

[0067] It should be understood that before using the above-mentioned neural network model, it is necessary to train the data enhancement module, the surface state feature extractor based on the convolutional neural network model, the feature distillation network, the joint clustering analysis network and the surface state monitor based on the classifier. That is to say, in the smart grid operation and maintenance management platform of the present application, it also includes a training module for training the data enhancement module, the surface state feature extractor based on the convolutional neural network model, the feature distillation network, the joint clustering analysis network and the surface state monitor based on the classifier.

[0068] Figure 4 FIG. 1 is a block diagram of a training module in a smart grid operation and maintenance management platform according to an embodiment of the present application. Figure 4As shown, the training module 200 includes: a training data acquisition unit 210, which is used to acquire training data, wherein the training data includes a training surface state image of the monitored power grid equipment collected by an unmanned aerial vehicle inspection aircraft, and a true value of whether the surface of the monitored power grid equipment is corroded, wherein the monitored power grid equipment is a transformer; a training image data enhancement unit 220, which is used to pass the training surface state image through the data enhancement module to obtain a set of training enhanced surface state images, wherein the data enhancement module is used to rotate and preprocess the training surface state image, and the preprocessing includes image denoising and color transformation; a training surface state extraction unit 230, which is used to pass each training enhanced surface state image in the set of training enhanced surface state images through the surface state feature extractor based on the convolutional neural network model to obtain a set of training power grid equipment surface state feature maps; a training feature distillation unit 240, which is used to extract the surface state feature map of the training power grid equipment. Each training power grid equipment surface state feature graph in the set of graphs passes through the feature distillation network to obtain a set of training distilled power grid equipment surface state feature vectors; a training feature clustering analysis unit 250 is used to pass the set of training distilled power grid equipment surface state feature vectors through the joint clustering analysis network to obtain a training power grid equipment surface state clustering feature matrix; a classification loss estimation unit 260 is used to pass the training power grid equipment surface state clustering feature matrix through the classifier-based surface state monitor to obtain a classification loss function value; a training unit 270 is used to train the data enhancement module, the surface state feature extractor based on the convolutional neural network model, the feature distillation network, the joint clustering analysis network and the classifier-based surface state monitor with the classification loss function value, wherein, in each round of iteration of the training, the training power grid equipment surface state clustering feature vector obtained after expanding the training power grid equipment surface state clustering feature matrix is ​​optimized.

[0069] In a preferred embodiment of the present application, each time the training power grid device surface state clustering feature vector after the training power grid device surface state clustering feature matrix is ​​expanded is classified and iterated by the classifier, the training power grid device surface state clustering feature vector is optimized by the following optimization formula, wherein the optimization formula is:

[0070]

[0071]

[0072] Where V is the clustering feature vector of the surface state of the training power grid equipment, v i and v jare the eigenvalues ​​of the i-th and j-th positions of the training grid equipment surface state clustering feature vector V, M μ (i, j) is the mean value between every two eigenvalues ​​of the training grid equipment surface state clustering eigenvector, M μ is the mean matrix of the clustering feature vectors of the surface state of the training power grid equipment, M σ (i, j) is the average difference between every two eigenvalues ​​of the training grid equipment surface state clustering feature vector, M μ is the average difference matrix of the surface state clustering feature vectors of the training power grid equipment, and V′ is the optimized surface state clustering feature vectors of the training power grid equipment.

[0073] That is, the local statistical information distribution of the training power grid equipment surface state clustering feature vector V is introduced as an external information source to perform feature vector retrieval enhancement, so as to avoid the distribution illusion of the training power grid equipment surface state clustering feature vector V caused by local overflow information distribution based on the structuring of local statistical dense information, thereby obtaining the information credible response reasoning of the training power grid equipment surface state clustering feature vector V based on the retention of the local distribution group dimension, so as to obtain the credible distribution response of the training power grid equipment surface state clustering feature vector V in the probability density space based on the discretized local feature distribution, thereby improving the convergence effect of the probability density space, so as to improve the training speed and the accuracy of the training results.

[0074] In summary, the smart grid operation and maintenance management platform according to the embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to perform image semantic analysis on the surface state image of the power grid equipment to capture the surface state characteristics of the power grid equipment and mine the abnormal appearance information of the equipment in the image, so as to intelligently identify whether the surface of the power grid equipment is corroded. In this way, surface damage of the power grid equipment can be discovered and handled in a timely manner, which is convenient for equipment operation and maintenance personnel to carry out repairs and maintenance, thereby ensuring the normal operation of the equipment and improving the operation and maintenance efficiency and safety of the smart grid.

[0075] Figure 5 FIG. 1 is a flow chart of a smart grid operation and maintenance management method according to an embodiment of the present application. Figure 5 As shown, the method for smart grid operation and maintenance management according to an embodiment of the present application includes the steps of: S1, acquiring a surface state image of a monitored power grid device collected by an unmanned aerial vehicle inspection aircraft, wherein the monitored power grid device is a transformer; S2, performing image semantic feature extraction and global feature clustering on the surface state image to obtain a power grid device surface state clustering feature matrix; S3, determining a monitoring result based on the power grid device surface state clustering feature matrix; S4, generating operation and maintenance management prompt information in response to the monitoring result that the surface of the monitored power grid device is corroded.

[0076] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned smart grid operation and maintenance management method have been referred to above. Figures 1 to 4 The smart grid operation and maintenance management platform has been introduced in detail, and therefore, its repeated description will be omitted.

[0077] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0078] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed platform and method can be implemented in other ways. For example, the platform embodiment described above is only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0079] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0080] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0081] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the platform claim can also be implemented by one unit through software or hardware.

[0082] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A smart grid operation and maintenance management platform, characterized in that: include: The device surface image acquisition module is used to obtain the surface state image of the monitored power grid device collected by the unmanned aerial vehicle inspection aircraft, wherein the monitored power grid device is a transformer; An image semantic feature mining module, used for performing image semantic feature extraction and global feature clustering on the surface state image to obtain a power grid equipment surface state clustering feature matrix; A monitoring result generating module, used for determining the monitoring result based on the grid equipment surface state clustering feature matrix; The response and operation and maintenance prompt module is used to generate operation and maintenance management prompt information in response to the monitoring result that the surface of the monitored power grid equipment is corroded.

2. The smart grid operation and maintenance management platform according to claim 1, characterized in that: The image semantic feature mining module comprises: An image data enhancement unit, used for passing the surface state image through a data enhancement module to obtain a set of enhanced surface state images; An equipment surface state feature extraction unit, used for respectively passing each enhanced surface state image in the set of enhanced surface state images through a surface state feature extractor based on a convolutional neural network model to obtain a set of power grid equipment surface state feature maps; A feature distillation unit, used for passing each power grid device surface state feature map in the set of power grid device surface state feature maps through a feature distillation network to obtain a set of distilled power grid device surface state feature vectors; The global feature clustering unit is used to obtain the grid equipment surface state clustering feature matrix by passing the set of distilled grid equipment surface state feature vectors through a joint clustering analysis network.

3. The smart grid operation and maintenance management platform according to claim 2, characterized in that: The data enhancement module is used to rotate and preprocess the surface state image, and the preprocessing includes image denoising and color conversion.

4. The smart grid operation and maintenance management platform according to claim 3, characterized in that: The characteristic distillation unit is used for: Expanding the surface state characteristic graph of the power grid equipment to obtain a sequence of surface state characteristic vectors of the power grid equipment; The sequence of the surface state feature vectors of the power grid equipment is processed by the following feature distillation formula to obtain the distilled surface state feature vector of the power grid equipment; wherein the feature distillation formula is: Among them, v i is the ith power grid device surface state feature vector in the sequence of power grid device surface state feature vectors, v j is the jth grid device surface state feature vector in the sequence of grid device surface state feature vectors, ‖·‖1 represents the 1-norm of the feature vector, M is the length of the sequence of grid device surface state feature vectors -1, V k is the representation of the sequence of the surface state feature vectors of the power grid equipment, D i represents the characteristic difference coefficient, exp(·) represents the natural exponential function operation, N s represents the total number of characteristic difference coefficients, P k is the surface state feature vector of the distillation power grid device.

5. The smart grid operation and maintenance management platform according to claim 4, characterized in that: The global feature clustering unit is used to: The following adjacency association formula is used to calculate the correlation weights of any two distillation power grid equipment surface state feature vectors in the set of the distillation power grid equipment surface state feature vectors to form an adjacency matrix with multiple of the correlation weights; wherein the adjacency association formula is: Among them, v i is the i-th distillation power grid device surface state feature vector in the set of distillation power grid device surface state feature vectors, v j is the jth distillation power grid device surface state feature vector in the set of distillation power grid device surface state feature vectors, σ is the stretching factor, ‖·‖2 2 represents the square of the vector norm, exp(·) represents the exponential operation, and W i,j is the correlation weight between the i-th distillation power grid device surface state feature vector and the j-th distillation power grid device surface state feature vector; Calculating a degree matrix of a set of surface state feature vectors of the distillation power grid device; Calculate a Laplacian matrix based on the adjacency matrix and the degree matrix; Normalizing the Laplace matrix to obtain a normalized Laplace matrix; After arranging the eigenvalues ​​in the standardized Laplace matrix in descending order, taking the first K eigenvalues, and decomposing the standardized Laplace matrix based on the first K eigenvalues ​​to obtain K decomposition eigenvectors, wherein K is the number of the surface state eigenvectors of the distillation power grid equipment; After the K decomposition feature vectors are standardized to obtain K standardized decomposition feature vectors, the K standardized decomposition feature vectors are reconstructed into the power grid equipment surface state clustering feature matrix.

6. The smart grid operation and maintenance management platform according to claim 5, characterized in that: The monitoring result generating module comprises: The equipment surface state identification unit is used to obtain the monitoring result by passing the grid equipment surface state clustering feature matrix through a surface state monitor based on a classifier, and the monitoring result is used to indicate whether the surface of the monitored grid equipment is corroded.

7. The smart grid operation and maintenance management platform according to claim 6, characterized in that: It also includes a training module for training the data enhancement module, the surface state feature extractor based on the convolutional neural network model, the feature distillation network, the joint clustering analysis network and the classifier-based surface state monitor.

8. The smart grid operation and maintenance management platform according to claim 7, characterized in that: The training module comprises: A training data acquisition unit, used to acquire training data, wherein the training data includes a training surface state image of a monitored power grid device collected by an unmanned aerial vehicle inspection aircraft, and a true value of whether the surface of the monitored power grid device is corroded, wherein the monitored power grid device is a transformer; a training image data enhancement unit, configured to pass the training surface state image through the data enhancement module to obtain a set of training enhanced surface state images, wherein the data enhancement module is configured to rotate and preprocess the training surface state image, wherein the preprocessing includes image denoising and color transformation; A training surface state extraction unit, used for passing each training enhanced surface state image in the set of training enhanced surface state images through the surface state feature extractor based on the convolutional neural network model to obtain a set of training power grid equipment surface state feature maps; A training feature distillation unit, used for passing each training power grid device surface state feature map in the set of training power grid device surface state feature maps through the feature distillation network to obtain a set of training distilled power grid device surface state feature vectors; A training feature clustering analysis unit, used for passing the set of training distilled power grid equipment surface state feature vectors through the joint clustering analysis network to obtain a training power grid equipment surface state clustering feature matrix; A classification loss estimation unit, used for passing the training grid equipment surface state clustering feature matrix through the classifier-based surface state monitor to obtain a classification loss function value; A training unit is used to train the data enhancement module, the surface state feature extractor based on the convolutional neural network model, the feature distillation network, the joint clustering analysis network and the classifier-based surface state monitor with the classification loss function value, wherein in each round of iteration of the training, the training power grid equipment surface state clustering feature vector obtained by expanding the training power grid equipment surface state clustering feature matrix is ​​optimized.

9. A smart grid operation and maintenance management method, characterized in that: include: Acquire a surface state image of a monitored power grid device collected by an unmanned aerial vehicle inspection aircraft, wherein the monitored power grid device is a transformer; Performing image semantic feature extraction and global feature clustering on the surface state image to obtain a power grid equipment surface state clustering feature matrix; Determining monitoring results based on the grid equipment surface state clustering feature matrix; In response to the monitoring result that the surface of the monitored power grid equipment is corroded, operation and maintenance management prompt information is generated.

Images