Transmission line fault identification system and method based on artificial intelligence

Through deep learning technology, the feature extraction and analysis of the voltage and current data of the transmission line is solved, and the problem of low manual inspection efficiency is achieved, efficient and accurate fault recognition is achieved, and the uncertainty of manual identification is reduced.

CN117368639BActive Publication Date: 2024-11-15吴迪
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
CN202311351797.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-10-18
Publication Date
2024-11-15
Estimated Expiration
2043-10-18

AI Technical Summary

Technical Problem

The prior art relies on manual inspection in the identification of high-voltage transmission line faults, which are low in efficiency and poor in reliability, making it difficult to effectively identify faults of complex network transmission lines, especially in the case of complex terrain in many regions.

Method used

The transmission line fault identification system based on deep learning is adopted to extract and correlate feature analysis of the amplitude data of three-phase voltage, three-phase current and zero-sequence current in the time period before and after the fault, and use the convolutional neural network and The Gaussian density graph performs feature fusion and optimization to generate fault classification results.

Benefits of technology

It improves the accuracy and efficiency of transmission line fault classification, reduces the uncertainty of manual identification, and can identify fault types more quickly and accurately.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117368639B_ABST
    Figure CN117368639B_ABST
Patent Text Reader

Abstract

The present application relates to the field of intelligent detection technology, and specifically discloses a power transmission line fault identification system and method based on artificial intelligence, which uses deep learning technology to extract time series variation features and analyze correlation features of amplitude data of three-phase voltage, three-phase current, and zero-sequence current at multiple predetermined time points before and after the fault, and thereby determines the type of power transmission line fault. In this way, the accuracy and efficiency of power transmission line fault classification can be effectively improved, and the uncertainty of manual identification can be reduced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of intelligent detection technology, and more specifically, to a power transmission line fault identification system and method based on artificial intelligence. Background Art

[0002] As an important part of the power system, high-voltage transmission lines are bridges connecting various power stations and end users, and shoulder the important mission of power transmission. Once a line failure occurs, it will inevitably seriously affect industrial and agricultural production and the lives of the people. As the scale of the power grid continues to expand, the safe and stable operation of the power system has an important impact on people's quality of life. However, due to various reasons such as equipment failure, weather conditions, external interference, and operational errors, short circuits, overloads, and other faults may occur in the transmission and distribution network. In order to ensure the stable operation of the transmission and distribution network, regular equipment inspection and maintenance work is required.

[0003] my country's power transmission lines are distributed in a network-like structure with many branches. In most areas, the terrain is complex and the line failure rate is very high. The manual inspection method is too dependent on the professional qualities and personal abilities of the staff, has low reliability and poor work efficiency, and increases the difficulty and time of fault identification.

[0004] Therefore, a power transmission line fault identification system and method based on artificial intelligence is expected. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of the present application provides a power transmission line fault identification system and method based on artificial intelligence, which uses deep learning technology to extract time series change features and analyze correlation features of the amplitude data of three-phase voltage, three-phase current, and zero-sequence current at multiple predetermined time points in the time period before and after the fault, and use this to determine the fault type of the power transmission line. In this way, the accuracy and efficiency of power transmission line fault classification can be effectively improved, and the uncertainty of manual identification can be reduced.

[0006] Accordingly, according to one aspect of the present application, there is provided a power transmission line fault identification system based on artificial intelligence, comprising:

[0007] A data acquisition module is used to obtain monitoring parameters at multiple predetermined time points in the period before and after the transmission line fault, wherein the monitoring parameters include amplitude data of three-phase voltage, three-phase current and zero-sequence current;

[0008] A monitoring parameter feature extraction module, used for arranging the monitoring parameters of the plurality of predetermined time points into a plurality of monitoring parameter input vectors according to the time dimension, and then respectively passing the plurality of monitoring parameter input vectors through a first convolutional neural network model including a first convolutional layer and a second convolutional layer to obtain a plurality of multi-scale monitoring parameter feature vectors;

[0009] A Gaussian fusion module, used for fusing the plurality of multi-scale monitoring parameter feature vectors based on a Gaussian density map to obtain a monitoring parameter collaborative feature matrix;

[0010] A fusion feature extraction module, used for passing the monitoring parameter collaborative feature matrix through a second convolutional neural network model using mutually transposed convolution kernels in adjacent layers to obtain a classification feature matrix;

[0011] An optimization module, used for performing geometric ordering on the classification feature matrix based on parameterized features to obtain an optimized classification feature matrix;

[0012] The identification result generation module is used to pass the optimized classification feature matrix through a multi-label classifier to obtain a classification result, and the classification result is used to represent the probability value of the transmission line having various types of faults.

[0013] In the above-mentioned artificial intelligence-based transmission line fault identification system, the monitoring parameter feature extraction module includes: a first scale feature extraction unit, which is used to use each layer of the first convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing of each feature matrix along the channel dimension and nonlinear activation processing on the input data in the forward transfer of the layer to output a first scale monitoring parameter feature vector from the last layer of the first convolutional layer; a second scale feature extraction unit, which is used to use each layer of the second convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing of each feature matrix along the channel dimension and nonlinear activation processing on the input data in the forward transfer of the layer to output a second scale monitoring parameter feature vector from the last layer of the second convolutional layer; a multi-scale feature fusion unit, which is used to fuse the first scale monitoring parameter feature vector and the second scale monitoring parameter feature vector to obtain the multi-scale monitoring parameter feature vector.

[0014] In the above-mentioned power transmission line fault identification system based on artificial intelligence, the Gaussian fusion module includes: a Gaussian density map construction unit, which is used to construct the Gaussian density map of the multiple multi-scale monitoring parameter feature vectors using the following Gaussian formula; wherein the Gaussian formula is: Wherein, μ is the position mean vector between the plurality of multi-scale monitoring parameter feature vectors, and σ is the variance between the eigenvalues ​​at each position in the plurality of multi-scale monitoring parameter feature vectors, represents a Gaussian density probability function, x represents a variable of a Gaussian density map; a Gaussian discretization unit is used to perform Gaussian discretization processing on the Gaussian distribution of each position in the Gaussian density map to obtain the monitoring parameter collaborative feature matrix.

[0015] In the above-mentioned artificial intelligence-based transmission line fault identification system, the fusion feature extraction module is used to: use the adjacent layers of the second convolutional neural network model using mutually transposed convolution kernels to perform the following on the input data in the forward transfer of the layer: perform convolution processing, pooling processing and nonlinear processing based on the first two-dimensional convolution kernel on the input data to obtain a first activation feature map; perform convolution processing, pooling processing and nonlinear activation processing based on the second two-dimensional convolution kernel on the first activation feature map to obtain a second activation feature map, wherein the first two-dimensional convolution kernel and the second two-dimensional convolution kernel are transposed to each other; wherein the input of the first layer of the second convolutional neural network model using mutually transposed convolution kernels in the adjacent layers is the monitoring parameter collaborative feature matrix, and the output of the last layer of the second convolutional neural network model using mutually transposed convolution kernels in the adjacent layers is the classification feature matrix.

[0016] In the above-mentioned artificial intelligence-based transmission line fault identification system, the optimization module includes: a parameterized coding vector construction unit, which is used to construct a parameterized coding vector for each position in the classification feature matrix to obtain a plurality of pixel position parameterized coding vectors, wherein the parameterized coding vector includes coordinates, gradient values ​​along the X-axis direction, gradient values ​​along the Y-axis direction, and eigenvalues; a parameterized feature extraction unit, which is used to pass the plurality of pixel position parameterized coding vectors through a parameterized feature extractor based on a fully connected layer to obtain a plurality of pixel position parameterized coding feature vectors; a cosine similarity calculation unit, which is used to calculate the cosine similarity between any two pixel position parameterized coding feature vectors in the plurality of pixel position parameterized coding feature vectors to obtain an ordered geometric topology matrix; a geometric topology feature extraction unit, which is used to pass the ordered geometric topology matrix through a geometric topology feature extractor based on a convolutional layer to obtain an ordered geometric topology feature matrix; and a fusion unit, which is used to fuse the classification feature matrix and the ordered geometric topology feature matrix to obtain the optimized classification feature matrix.

[0017] In the above-mentioned artificial intelligence-based transmission line fault identification system, the identification result generation module includes: an expansion unit, which is used to expand the optimized classification feature matrix into a classification feature vector according to a row vector or a column vector; a fully connected unit, which is used to use the fully connected layer of the classifier to fully connect the classification feature vector to obtain a fully connected encoded feature vector; a probabilization unit, which is used to input the fully connected encoded feature vector into the Softmax classification function of the classifier to obtain the probability value of the optimized classification feature matrix belonging to each classification label, and the classification label includes all possible fault types of the transmission line.

[0018] According to another aspect of the present application, a method for identifying power transmission line faults based on artificial intelligence is provided, which comprises:

[0019] Acquire monitoring parameters at multiple predetermined time points in a period before and after a transmission line fault, wherein the monitoring parameters include amplitude data of three-phase voltage, three-phase current, and zero-sequence current;

[0020] Arranging the monitoring parameters of the plurality of predetermined time points into a plurality of monitoring parameter input vectors according to the time dimension respectively, and then passing the plurality of monitoring parameter input vectors through a first convolutional neural network model including a first convolutional layer and a second convolutional layer respectively to obtain a plurality of multi-scale monitoring parameter feature vectors;

[0021] Based on the Gaussian density map, the plurality of multi-scale monitoring parameter feature vectors are fused to obtain a monitoring parameter collaborative feature matrix;

[0022] The monitoring parameter collaborative feature matrix is ​​passed through a second convolutional neural network model using mutually transposed convolution kernels in adjacent layers to obtain a classification feature matrix;

[0023] Performing geometric ordering on the classification feature matrix based on parameterized features to obtain an optimized classification feature matrix;

[0024] The optimized classification feature matrix is ​​passed through a multi-label classifier to obtain a classification result, and the classification result is used to represent the probability value of the transmission line having various types of faults.

[0025] In the above-mentioned artificial intelligence-based transmission line fault identification method, the monitoring parameters of the multiple predetermined time points are arranged as multiple monitoring parameter input vectors according to the time dimension and then respectively passed through a first convolutional neural network model including a first convolutional layer and a second convolutional layer to obtain multiple multi-scale monitoring parameter feature vectors, including: using each layer of the first convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing and nonlinear activation processing of each feature matrix along the channel dimension on the input data in the forward transfer of the layer to output a first-scale monitoring parameter feature vector from the last layer of the first convolutional layer; using each layer of the second convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing and nonlinear activation processing of each feature matrix along the channel dimension on the input data in the forward transfer of the layer to output a second-scale monitoring parameter feature vector from the last layer of the second convolutional layer; fusing the first-scale monitoring parameter feature vector and the second-scale monitoring parameter feature vector to obtain the multi-scale monitoring parameter feature vector.

[0026] In the above-mentioned power transmission line fault identification method based on artificial intelligence, the multiple multi-scale monitoring parameter feature vectors are fused based on the Gaussian density map to obtain the monitoring parameter collaborative feature matrix, including: constructing the Gaussian density map of the multiple multi-scale monitoring parameter feature vectors using the following Gaussian formula; wherein the Gaussian formula is: Wherein, μ is the position mean vector between the plurality of multi-scale monitoring parameter feature vectors, and σ is the variance between the eigenvalues ​​at each position in the plurality of multi-scale monitoring parameter feature vectors, represents a Gaussian density probability function, x represents a variable of a Gaussian density map; and Gaussian discretization is performed on the Gaussian distribution of each position in the Gaussian density map to obtain the monitoring parameter collaborative feature matrix.

[0027] In the above-mentioned artificial intelligence-based transmission line fault identification method, the monitoring parameter collaborative feature matrix is ​​passed through the second convolutional neural network model using mutually transposed convolution kernels in adjacent layers to obtain a classification feature matrix, including: using the adjacent layers of the second convolutional neural network model using mutually transposed convolution kernels to perform the following on the input data in the forward transfer of the layer: performing convolution processing, pooling processing and nonlinear processing based on a first two-dimensional convolution kernel on the input data to obtain a first activation feature map; performing convolution processing, pooling processing and nonlinear activation processing based on a second two-dimensional convolution kernel on the first activation feature map to obtain a second activation feature map, wherein the first two-dimensional convolution kernel and the second two-dimensional convolution kernel are transposed of each other; wherein the input of the first layer of the second convolutional neural network model using mutually transposed convolution kernels in the adjacent layers is the monitoring parameter collaborative feature matrix, and the output of the last layer of the second convolutional neural network model using mutually transposed convolution kernels in the adjacent layers is the classification feature matrix.

[0028] Compared with the prior art, the power transmission line fault identification system and method based on artificial intelligence provided by the present application utilizes deep learning technology to extract time series variation features and analyze correlation features of the amplitude data of three-phase voltage, three-phase current, and zero-sequence current at multiple predetermined time points before and after the fault, and thereby determine the type of power transmission line fault. In this way, the accuracy and efficiency of power transmission line fault classification can be effectively improved, and the uncertainty of manual identification can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] 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.

[0030] Figure 1 4 is a block diagram of an artificial intelligence-based power transmission line fault identification system according to an embodiment of the present application.

[0031] Figure 2Schematic diagram of the architecture of an artificial intelligence-based transmission line fault identification system according to an embodiment of the present application.

[0032] Figure 3 It is a block diagram of a monitoring parameter feature extraction module in an artificial intelligence-based transmission line fault identification system according to an embodiment of the present application.

[0033] Figure 4 It is a block diagram of a Gaussian fusion module in an artificial intelligence-based transmission line fault identification system according to an embodiment of the present application.

[0034] Figure 5 The present invention is a block diagram of an identification result generation module in an artificial intelligence-based power transmission line fault identification system according to an embodiment of the present application.

[0035] Figure 6 The present invention is a flowchart of a method for identifying power transmission line faults based on artificial intelligence according to an embodiment of the present application. DETAILED DESCRIPTION

[0036] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. 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 exemplary embodiments described here.

[0037] Exemplary Systems

[0038] Figure 1 FIG. 1 is a block diagram of a power transmission line fault identification system based on artificial intelligence according to an embodiment of the present application. Figure 1As shown, according to the embodiment of the present application, the power transmission line fault identification system 100 based on artificial intelligence includes: a data acquisition module 110, which is used to obtain monitoring parameters at multiple predetermined time points in the time period before and after the transmission line fault, and the monitoring parameters include amplitude data of three-phase voltage, three-phase current, and zero-sequence current; a monitoring parameter feature extraction module 120, which is used to arrange the monitoring parameters of the multiple predetermined time points into multiple monitoring parameter input vectors according to the time dimension, and then pass them through a first convolutional neural network model including a first convolutional layer and a second convolutional layer to obtain multiple multi-scale monitoring parameter feature vectors; a Gaussian fusion module 130, which is used to obtain a plurality of multi-scale monitoring parameter feature vectors based on the time dimension; The Gaussian density map fuses the multiple multi-scale monitoring parameter feature vectors to obtain a monitoring parameter collaborative feature matrix; a fusion feature extraction module 140 is used to pass the monitoring parameter collaborative feature matrix through adjacent layers using a second convolutional neural network model with mutually transposed convolution kernels to obtain a classification feature matrix; an optimization module 150 is used to geometrically order the classification feature matrix based on parameterized features to obtain an optimized classification feature matrix; an identification result generation module 160 is used to pass the optimized classification feature matrix through a multi-label classifier to obtain a classification result, and the classification result is used to represent the probability value of the transmission line having various types of faults.

[0039] Figure 2 FIG. 1 is a schematic diagram of the architecture of a power transmission line fault identification system based on artificial intelligence according to an embodiment of the present application. Figure 2 As shown, first, the monitoring parameters of multiple predetermined time points in the time period before and after the transmission line fault are obtained, and the monitoring parameters include the amplitude data of three-phase voltage, three-phase current, and zero-sequence current. Then, the monitoring parameters of the multiple predetermined time points are arranged as multiple monitoring parameter input vectors according to the time dimension, and then passed through the first convolutional neural network model including the first convolutional layer and the second convolutional layer to obtain multiple multi-scale monitoring parameter feature vectors. Next, the multiple multi-scale monitoring parameter feature vectors are fused based on the Gaussian density map to obtain a monitoring parameter collaborative feature matrix. Then, the monitoring parameter collaborative feature matrix is ​​passed through the second convolutional neural network model with mutually transposed convolution kernels in adjacent layers to obtain a classification feature matrix. Secondly, the classification feature matrix is ​​geometrically ordered based on parameterized features to obtain an optimized classification feature matrix. Finally, the optimized classification feature matrix is ​​passed through a multi-label classifier to obtain a classification result, and the classification result is used to represent the probability value of the transmission line having various types of faults.

[0040] In the above-mentioned power transmission line fault identification system 100 based on artificial intelligence, the data acquisition module 110 is used to obtain monitoring parameters at multiple predetermined time points in the time period before and after the power transmission line fault, and the monitoring parameters include amplitude data of three-phase voltage, three-phase current, and zero-sequence current. As mentioned in the above-mentioned background technology, my country's power transmission lines are distributed in a network-like structure with many branches, and the terrain in most areas is complex, and the line failure rate is very high. The manual inspection method is too dependent on the professional quality and personal ability of the staff, with low reliability and poor work efficiency, which increases the difficulty and time of fault identification. Therefore, an intelligent and efficient power transmission line fault identification solution is expected.

[0041] Accordingly, considering that when a fault occurs, the voltage, current and other parameters of the transmission line usually change significantly, and different fault types have corresponding parameter change characteristics, it is possible to construct a transmission line fault identification scheme by performing feature analysis on the voltage, current and other parameters before and after the transmission line fault. Based on this, in the technical scheme of the present application, deep learning technology is used to extract time series change features and analyze the correlation features of the amplitude data of the three-phase voltage, three-phase current and zero-sequence current at multiple predetermined time points in the time period before and after the fault, and use this to determine the fault type of the transmission line. In this way, the accuracy and efficiency of transmission line fault classification can be effectively improved, and the uncertainty of manual identification can be reduced. Specifically, in the technical scheme of the present application, first, the monitoring parameters of multiple predetermined time points in the time period before and after the transmission line fault are obtained by sensors, and the monitoring parameters include the amplitude data of the three-phase voltage, three-phase current and zero-sequence current.

[0042] In the above-mentioned power transmission line fault identification system 100 based on artificial intelligence, the monitoring parameter feature extraction module 120 is used to arrange the monitoring parameters of the multiple predetermined time points into multiple monitoring parameter input vectors according to the time dimension, and then pass through the first convolutional neural network model including the first convolutional layer and the second convolutional layer to obtain multiple multi-scale monitoring parameter feature vectors. It should be understood that the transmission line fault usually causes the monitoring parameters to change in time. By arranging the monitoring parameters according to the time dimension, the temporal correlation can be retained, so that the model can learn the dynamic change pattern before and after the fault occurs. At the same time, the features on different time scales are of great significance for fault classification and identification. Here, by using the convolutional neural network model including the first convolutional layer and the second convolutional layer, features of multiple scales are extracted at different levels. For example, the first convolutional layer can extract local features, while the second convolutional layer can extract higher-level abstract features. In this way, fault features on different time scales can be captured and the accuracy of fault classification can be improved. In addition, the convolutional neural network has the characteristics of parameter sharing and weight sharing, which can effectively reduce the number of parameters of the model and improve the training efficiency and generalization ability of the model. By sharing weights and parameters at multiple time points, the feature extraction process can be shared at different time points, improving the model's ability to extract fault features.

[0043] Figure 3 FIG. 1 is a block diagram of a monitoring parameter feature extraction module in a power transmission line fault identification system based on artificial intelligence according to an embodiment of the present application. Figure 3 As shown, the monitoring parameter feature extraction module 120 includes: a first scale feature extraction unit 121, which is used to use each layer of the first convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing of each feature matrix along the channel dimension and non-linear activation processing on the input data in the forward transfer of the layer so as to output a first scale monitoring parameter feature vector from the last layer of the first convolutional layer; a second scale feature extraction unit 122, which is used to use each layer of the second convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing of each feature matrix along the channel dimension and non-linear activation processing on the input data in the forward transfer of the layer so as to output a second scale monitoring parameter feature vector from the last layer of the second convolutional layer; a multi-scale feature fusion unit 123, which is used to fuse the first scale monitoring parameter feature vector and the second scale monitoring parameter feature vector to obtain the multi-scale monitoring parameter feature vector.

[0044] In the above-mentioned artificial intelligence-based transmission line fault identification system 100, the Gaussian fusion module 130 is used to fuse the multiple multi-scale monitoring parameter feature vectors based on the Gaussian density map to obtain the monitoring parameter collaborative feature matrix. In order to integrate the features of different types of monitoring parameters to improve the accuracy of fault classification and identification, the multiple multi-scale monitoring parameter feature vectors are further fused using the Gaussian density map. It should be understood that the Gaussian density map is a commonly used probability density estimation method, which is often used to describe the distribution of data. By fusing the multiple multi-scale monitoring parameter feature vectors based on the Gaussian density map, the weights of the multiple multi-scale monitoring parameter feature vectors can be associated with their distribution in the feature space, so as to better capture the spatial distribution information of the fault features and improve the accuracy of fault classification and identification. In addition, the feature information of different parameters is integrated into a unified feature representation, which reduces the redundancy of feature dimensions and extracts more robust and representative features, which is beneficial to subsequent classifier training and fault identification tasks.

[0045] Figure 4 FIG. 1 is a block diagram of a Gaussian fusion module in a power transmission line fault identification system based on artificial intelligence according to an embodiment of the present application. Figure 4 As shown, the Gaussian fusion module 130 includes: a Gaussian density map construction unit 131, which is used to construct the Gaussian density map of the multiple multi-scale monitoring parameter feature vectors using the following Gaussian formula; wherein the Gaussian formula is: Wherein, μ is the position mean vector between the plurality of multi-scale monitoring parameter feature vectors, and σ is the variance between the eigenvalues ​​at each position in the plurality of multi-scale monitoring parameter feature vectors, represents a Gaussian density probability function, and x represents a variable of the Gaussian density map; a Gaussian discretization unit 132 is used to perform Gaussian discretization processing on the Gaussian distribution of each position in the Gaussian density map to obtain the monitoring parameter collaborative feature matrix.

[0046] In the above-mentioned artificial intelligence-based transmission line fault identification system 100, the fusion feature extraction module 140 is used to pass the monitoring parameter collaborative feature matrix through the adjacent layers using the second convolutional neural network model with mutually transposed convolution kernels to obtain a classification feature matrix. It should be understood that the adjacent layers of the second convolutional neural network model use mutually transposed convolution kernels, which is equivalent to performing a reverse convolution operation on the monitoring parameter collaborative feature matrix, and can extract higher-level feature representations. In this way, encoding the monitoring parameter collaborative feature matrix through a deconvolution operation is equivalent to reconstructing the features, which can enhance the expressiveness and discrimination of the features, so that the classifier can better distinguish different types of faults. And by converting the features from a high-dimensional space to a low-dimensional feature representation through a deconvolution operation, this dimensionality reduction operation can reduce the redundancy and noise of the features and improve the training efficiency and generalization ability of the classifier.

[0047] Correspondingly, in a specific example, the fusion feature extraction module 140 is used to: use the adjacent layers of the second convolutional neural network model using mutually transposed convolution kernels to perform the following on the input data in the forward pass of the layer: perform convolution processing, pooling processing and nonlinear processing based on a first two-dimensional convolution kernel on the input data to obtain a first activation feature map; perform convolution processing, pooling processing and nonlinear activation processing based on a second two-dimensional convolution kernel on the first activation feature map to obtain a second activation feature map, wherein the first two-dimensional convolution kernel and the second two-dimensional convolution kernel are transposed of each other; wherein the input of the first layer of the second convolutional neural network model in which the adjacent layers use mutually transposed convolution kernels is the monitoring parameter collaborative feature matrix, and the output of the last layer of the second convolutional neural network model in which the adjacent layers use mutually transposed convolution kernels is the classification feature matrix.

[0048] In the above-mentioned power transmission line fault identification system 100 based on artificial intelligence, the optimization module 150 is used to geometrically order the classification feature matrix based on parameterized features to obtain an optimized classification feature matrix. In the technical solution of the present application, because the classification feature matrix is ​​obtained by a specific encoding method, there are specific data structures and internal laws inside the data structure of the classification feature matrix, that is, the parameterized features of the classification feature matrix are in an ordered structure according to certain rules. Therefore, if the geometric ordering of the parameterized features of the classification feature matrix can be utilized, the feature expression of the classification feature matrix can be optimized.

[0049] Specifically, in the technical solution of the present application, a parameterized coding vector of each position in the classification feature matrix is ​​constructed to obtain a plurality of pixel position parameterized coding vectors, wherein the parameterized coding vector includes coordinates, gradient values ​​along the X-axis direction, gradient values ​​along the Y-axis direction, and eigenvalues. Then, the plurality of pixel position parameterized coding vectors are passed through a parameterized feature extractor based on a fully connected layer to obtain a plurality of pixel position parameterized coding feature vectors, that is, the implicit information of the parameter features of each position in the classification feature matrix is ​​captured by using fully connected coding. Furthermore, the cosine similarity between any two pixel position parameterized coding feature vectors in the plurality of pixel position parameterized coding feature vectors is calculated to obtain an ordered geometric topology matrix, that is, the geometric order feature value of the parameterized feature is represented by the cosine similarity between the pixel position parameterized coding feature vectors. Furthermore, the ordered geometric topology matrix is ​​passed through a geometric topology feature extractor based on a convolutional layer to obtain an ordered geometric topology feature matrix, that is, the ordered geometric topology features implicit in the ordered geometric topology matrix are captured by convolutional coding. Then, the classification feature matrix and the ordered geometric topological feature matrix are fused to obtain the optimized classification feature matrix. For example, the ordered geometric topological feature matrix and the classification feature matrix are matrix multiplied to map the classification feature matrix into the feature space of the ordered geometric topological feature matrix to obtain the optimized classification feature matrix.

[0050] In this way, the classification feature matrix is ​​geometrically ordered based on parameterized features to arrange the classification feature matrix into an orderly structure according to certain rules to improve the expression ability and classification effect of the classification feature matrix. At the same time, the geometric ordering based on parameterized features can also make the classification feature matrix more compact and effective, which is conducive to improving the accuracy and efficiency of the classification task.

[0051] Accordingly, in a specific example, the optimization module 150 includes: a parameterized coding vector construction unit, which is used to construct a parameterized coding vector for each position in the classification feature matrix to obtain multiple pixel position parameterized coding vectors, wherein the parameterized coding vector includes coordinates, gradient values ​​along the X-axis direction, gradient values ​​along the Y-axis direction, and eigenvalues; a parameterized feature extraction unit, which is used to pass the multiple pixel position parameterized coding vectors through a parameterized feature extractor based on a fully connected layer to obtain multiple pixel position parameterized coding feature vectors; a cosine similarity calculation unit, which is used to calculate the cosine similarity between any two pixel position parameterized coding feature vectors in the multiple pixel position parameterized coding feature vectors to obtain an ordered geometric topological matrix; a geometric topological feature extraction unit, which is used to pass the ordered geometric topological matrix through a geometric topological feature extractor based on a convolutional layer to obtain an ordered geometric topological feature matrix; and a fusion unit, which is used to fuse the classification feature matrix and the ordered geometric topological feature matrix to obtain the optimized classification feature matrix.

[0052] In the above-mentioned power transmission line fault identification system 100 based on artificial intelligence, the identification result generation module 160 is used to pass the optimized classification feature matrix through a multi-label classifier to obtain a classification result, and the classification result is used to indicate the probability value of the power transmission line having various types of faults. It should be understood that the multi-label classifier can map the optimized classification feature matrix to each fault category, compare and match the state of the power transmission line with the known fault categories to determine the possible fault types, and output the probability value of each fault category, indicating the possibility that the power transmission line has this type of fault. In this way, the classification results can help the operation and maintenance personnel quickly understand the fault conditions of the line and take corresponding maintenance and treatment measures in a timely manner.

[0053] Figure 5 FIG. 1 is a block diagram of an identification result generation module in a power transmission line fault identification system based on artificial intelligence according to an embodiment of the present application. Figure 5 As shown, the recognition result generation module 160 includes: an expansion unit 161, which is used to expand the optimized classification feature matrix into a classification feature vector according to a row vector or a column vector; a full connection unit 162, which is used to use the full connection layer of the classifier to fully connect the classification feature vector to obtain a fully connected encoded feature vector; a probabilization unit 163, which is used to input the fully connected encoded feature vector into the Softmax classification function of the classifier to obtain the probability value of the optimized classification feature matrix belonging to each classification label, and the classification label includes all possible fault types of the transmission line.

[0054] In summary, the power transmission line fault identification system based on artificial intelligence according to the embodiment of the present application is explained, which uses deep learning technology to extract time series variation features and analyze correlation features of the amplitude data of three-phase voltage, three-phase current, and zero-sequence current at multiple predetermined time points in the time period before and after the fault, and thereby determines the fault type of the power transmission line. In this way, the accuracy and efficiency of power transmission line fault classification can be effectively improved, and the uncertainty of manual identification can be reduced.

[0055] Exemplary Methods

[0056] Figure 6 FIG. 1 is a flow chart of a method for identifying power transmission line faults based on artificial intelligence according to an embodiment of the present application. Figure 6 As shown, according to the embodiment of the present application, the power transmission line fault identification method based on artificial intelligence includes the following steps: S110, obtaining monitoring parameters at multiple predetermined time points in the time period before and after the transmission line fault, wherein the monitoring parameters include amplitude data of three-phase voltage, three-phase current, and zero-sequence current; S120, arranging the monitoring parameters at the multiple predetermined time points into multiple monitoring parameter input vectors according to the time dimension, and then passing through a first convolutional neural network model including a first convolutional layer and a second convolutional layer to obtain multiple multi-scale monitoring parameter feature vectors; S130, fusing the multiple multi-scale monitoring parameter feature vectors based on a Gaussian density map to obtain a monitoring parameter collaborative feature matrix; S140, passing the monitoring parameter collaborative feature matrix through a second convolutional neural network model using mutually transposed convolution kernels in adjacent layers to obtain a classification feature matrix; S150, geometrically ordering the classification feature matrix based on parameterized features to obtain an optimized classification feature matrix; S160, passing the optimized classification feature matrix through a multi-label classifier to obtain a classification result, and the classification result is used to represent the probability value of the transmission line having various types of faults.

[0057] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned power transmission line fault identification method based on artificial intelligence have been referred to above. Figures 1 to 5 The description of the artificial intelligence-based transmission line fault identification system has been introduced in detail, and therefore, its repeated description will be omitted.

[0058] The basic principles of the present application are described above in conjunction with specific embodiments. However, it should be noted that the advantages, strengths, effects, etc. mentioned in the present application are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. are required by each embodiment of the present application. In addition, the specific details disclosed above are only for the purpose of illustration and ease of understanding, not for limitation, and the above details do not limit the present application to being implemented by adopting the above specific details.

[0059] The block diagrams of the devices, apparatuses, equipment, and systems involved in this application are only illustrative examples and are not intended to require or imply that they must be connected, arranged, and configured in the manner shown in the block diagram. As will be appreciated by those skilled in the art, these devices, apparatuses, equipment, and systems can be connected, arranged, and configured in any manner. Words such as "including", "comprising", "having", etc. are open words, referring to "including but not limited to", and can be used interchangeably with them. The words "or" and "and" used here refer to the words "and / or" and can be used interchangeably with them, unless the context clearly indicates otherwise. The words "such as" used here refer to the phrase "such as but not limited to", and can be used interchangeably with them.

[0060] It should also be noted that in the apparatus, device and method of the present application, each component or each step can be decomposed and / or recombined. Such decomposition and / or recombination should be regarded as equivalent solutions of the present application.

[0061] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of the present application. Therefore, the present application is not intended to be limited to the aspects shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

[0062] The above description has been given for the purpose of illustration and description. In addition, this description is not intended to limit the embodiments of the present application to the forms disclosed herein. Although multiple example aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, changes, additions and sub-combinations thereof.

Claims

1. A power transmission line fault identification system based on artificial intelligence, characterized in that: include: A data acquisition module is used to obtain monitoring parameters at multiple predetermined time points in the period before and after the transmission line fault, wherein the monitoring parameters include amplitude data of three-phase voltage, three-phase current and zero-sequence current; A monitoring parameter feature extraction module, used for arranging the monitoring parameters of the plurality of predetermined time points into a plurality of monitoring parameter input vectors according to the time dimension, and then respectively passing the plurality of monitoring parameter input vectors through a first convolutional neural network model including a first convolutional layer and a second convolutional layer to obtain a plurality of multi-scale monitoring parameter feature vectors; A Gaussian fusion module, used for fusing the plurality of multi-scale monitoring parameter feature vectors based on a Gaussian density map to obtain a monitoring parameter collaborative feature matrix; A fusion feature extraction module, used for passing the monitoring parameter collaborative feature matrix through a second convolutional neural network model using mutually transposed convolution kernels in adjacent layers to obtain a classification feature matrix; An optimization module, used for performing geometric ordering on the classification feature matrix based on parameterized features to obtain an optimized classification feature matrix; An identification result generation module, used for passing the optimized classification feature matrix through a multi-label classifier to obtain a classification result, wherein the classification result is used to indicate the probability value of the transmission line having various types of faults; Wherein, the optimization module includes: A parameterized coding vector construction unit, used to construct a parameterized coding vector for each position in the classification feature matrix to obtain a plurality of pixel position parameterized coding vectors, wherein the parameterized coding vector includes a coordinate, a gradient value along the X-axis direction, a gradient value along the Y-axis direction, and a eigenvalue; A parameterized feature extraction unit, configured to pass the plurality of pixel position parameterized coding vectors through a parameterized feature extractor based on a fully connected layer to obtain a plurality of pixel position parameterized coding feature vectors; A cosine similarity calculation unit, used for calculating the cosine similarity between any two pixel position parameterized encoding feature vectors among the plurality of pixel position parameterized encoding feature vectors to obtain an ordered geometric topology matrix; A geometric topological feature extraction unit, used for passing the ordered geometric topological matrix through a geometric topological feature extractor based on a convolution layer to obtain an ordered geometric topological feature matrix; A fusion unit is used to fuse the classification feature matrix and the ordered geometric topology feature matrix to obtain the optimized classification feature matrix.

2. The power transmission line fault identification system based on artificial intelligence according to claim 1 is characterized in that: The monitoring parameter feature extraction module comprises: A first scale feature extraction unit, configured to use each layer of the first convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing of each feature matrix along the channel dimension, and nonlinear activation processing on the input data in the forward pass of the layer, so as to output a first scale monitoring parameter feature vector from the last layer of the first convolutional layer; A second scale feature extraction unit, configured to use each layer of the second convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing of each feature matrix along the channel dimension, and nonlinear activation processing on the input data in the forward pass of the layer, so as to output a second scale monitoring parameter feature vector from the last layer of the second convolutional layer; A multi-scale feature fusion unit is used to fuse the first-scale monitoring parameter feature vector and the second-scale monitoring parameter feature vector to obtain the multi-scale monitoring parameter feature vector.

3. The power transmission line fault identification system based on artificial intelligence according to claim 2 is characterized in that: The Gaussian fusion module comprises: A Gaussian density map construction unit is used to construct a Gaussian density map of the plurality of multi-scale monitoring parameter feature vectors using the following Gaussian formula; wherein the Gaussian formula is: Wherein, μ is the position mean vector between the plurality of multi-scale monitoring parameter feature vectors, and σ is the variance between the eigenvalues ​​at each position in the plurality of multi-scale monitoring parameter feature vectors, represents the Gaussian density probability function, and x represents the variable of the Gaussian density map; The Gaussian discretization unit is used to perform Gaussian discretization processing on the Gaussian distribution of each position in the Gaussian density map to obtain the monitoring parameter collaborative feature matrix.

4. The power transmission line fault identification system based on artificial intelligence according to claim 3 is characterized in that: The fusion feature extraction module is used to: use the adjacent layers of the second convolutional neural network model using mutually transposed convolution kernels to perform the following operations on the input data in the forward transfer of the layer: Performing convolution processing, pooling processing and nonlinear processing based on a first two-dimensional convolution kernel on the input data to obtain a first activation feature map; Performing convolution processing, pooling processing, and nonlinear activation processing based on a second two-dimensional convolution kernel on the first activation feature map to obtain a second activation feature map, wherein the first two-dimensional convolution kernel and the second two-dimensional convolution kernel are transposed of each other; Among them, the input of the first layer of the second convolutional neural network model in which the adjacent layers use mutually transposed convolution kernels is the monitoring parameter collaborative feature matrix, and the output of the last layer of the second convolutional neural network model in which the adjacent layers use mutually transposed convolution kernels is the classification feature matrix.

5. The power transmission line fault identification system based on artificial intelligence according to claim 4 is characterized in that: The recognition result generating module comprises: An expansion unit, used for expanding the optimized classification feature matrix into a classification feature vector according to row vectors or column vectors; A fully connected unit, used for performing fully connected encoding on the classification feature vector using the fully connected layer of the classifier to obtain a fully connected encoded feature vector; The probabilization unit is used to input the fully connected encoded feature vector into the Softmax classification function of the classifier to obtain the probability value of the optimized classification feature matrix belonging to each classification label, and the classification label includes all possible fault types of the transmission line.

6. A method for identifying power transmission line faults based on artificial intelligence, characterized in that: include: Acquire monitoring parameters at multiple predetermined time points in a period before and after a transmission line fault, wherein the monitoring parameters include amplitude data of three-phase voltage, three-phase current, and zero-sequence current; Arranging the monitoring parameters of the plurality of predetermined time points into a plurality of monitoring parameter input vectors according to the time dimension respectively, and then passing the plurality of monitoring parameter input vectors through a first convolutional neural network model including a first convolutional layer and a second convolutional layer respectively to obtain a plurality of multi-scale monitoring parameter feature vectors; Based on the Gaussian density map, the plurality of multi-scale monitoring parameter feature vectors are fused to obtain a monitoring parameter collaborative feature matrix; The monitoring parameter collaborative feature matrix is ​​passed through a second convolutional neural network model using mutually transposed convolution kernels in adjacent layers to obtain a classification feature matrix; Performing geometric ordering on the classification feature matrix based on parameterized features to obtain an optimized classification feature matrix; Passing the optimized classification feature matrix through a multi-label classifier to obtain a classification result, wherein the classification result is used to represent the probability value of the transmission line having various types of faults; The step of geometrically ordering the classification feature matrix based on parameterized features to obtain an optimized classification feature matrix includes: Constructing a parameterized coding vector for each position in the classification feature matrix to obtain a plurality of pixel position parameterized coding vectors, wherein the parameterized coding vector includes a coordinate, a gradient value along an X-axis direction, a gradient value along a Y-axis direction, and a eigenvalue; Passing the plurality of pixel position parameterized encoding vectors through a parameterized feature extractor based on a fully connected layer to obtain a plurality of pixel position parameterized encoding feature vectors; Calculating the cosine similarity between any two pixel position parameterized encoding feature vectors among the plurality of pixel position parameterized encoding feature vectors to obtain an ordered geometric topology matrix; Passing the ordered geometric topology matrix through a geometric topology feature extractor based on a convolutional layer to obtain an ordered geometric topology feature matrix; The classification feature matrix and the ordered geometric topology feature matrix are fused to obtain the optimized classification feature matrix.

7. The method for identifying power transmission line faults based on artificial intelligence according to claim 6, characterized in that: The monitoring parameters of the plurality of predetermined time points are arranged into a plurality of monitoring parameter input vectors according to the time dimension and then respectively passed through a first convolutional neural network model including a first convolutional layer and a second convolutional layer to obtain a plurality of multi-scale monitoring parameter feature vectors, including: Using each layer of the first convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing of each feature matrix along the channel dimension, and nonlinear activation processing on the input data in the forward pass of the layer to output a first scale monitoring parameter feature vector from the last layer of the first convolutional layer; Using each layer of the second convolutional layer of the first convolutional neural network model to perform one-dimensional convolution processing, global pooling processing of each feature matrix along the channel dimension and nonlinear activation processing on the input data in the forward pass of the layer to output a second-scale monitoring parameter feature vector from the last layer of the second convolutional layer; The first-scale monitoring parameter feature vector and the second-scale monitoring parameter feature vector are fused to obtain the multi-scale monitoring parameter feature vector.

8. The method for identifying power transmission line faults based on artificial intelligence according to claim 7, characterized in that: The plurality of multi-scale monitoring parameter feature vectors are fused based on the Gaussian density map to obtain a monitoring parameter collaborative feature matrix, including: The Gaussian density map of the plurality of multi-scale monitoring parameter feature vectors is constructed using the following Gaussian formula; wherein the Gaussian formula is: Wherein, μ is the position mean vector between the plurality of multi-scale monitoring parameter feature vectors, and σ is the variance between the eigenvalues ​​at each position in the plurality of multi-scale monitoring parameter feature vectors, represents the Gaussian density probability function, and x represents the variable of the Gaussian density map; Gaussian discretization is performed on the Gaussian distribution of each position in the Gaussian density map to obtain the monitoring parameter collaborative feature matrix.

9. The method for identifying power transmission line faults based on artificial intelligence according to claim 8, characterized in that: The monitoring parameter collaborative feature matrix is ​​passed through a second convolutional neural network model using mutually transposed convolution kernels in adjacent layers to obtain a classification feature matrix, including: Each layer of the second convolutional neural network model using the adjacent layers using mutually transposed convolution kernels performs the following operations on the input data in the forward pass of the layer: Performing convolution processing, pooling processing and nonlinear processing based on a first two-dimensional convolution kernel on the input data to obtain a first activation feature map; Performing convolution processing, pooling processing, and nonlinear activation processing based on a second two-dimensional convolution kernel on the first activation feature map to obtain a second activation feature map, wherein the first two-dimensional convolution kernel and the second two-dimensional convolution kernel are transposed of each other; Among them, the input of the first layer of the second convolutional neural network model in which the adjacent layers use mutually transposed convolution kernels is the monitoring parameter collaborative feature matrix, and the output of the last layer of the second convolutional neural network model in which the adjacent layers use mutually transposed convolution kernels is the classification feature matrix.

Citation Information

Patent Citations

  • Method and system for positioning fault of power distribution network

    CN108120900A

  • Power grid fault intelligent discrimination method based on multi-dimensional panoramic data

    CN111781467A

  • Reactor state diagnosis method and system

    CN112182490A

  • New energy power transmission line fault identification method based on convolutional network

    CN115469184A

  • Power grid line fault diagnosis method, equipment and medium

    CN116679161A