Power transmission line fault positioning method and system based on ground potential end transient state information perception

Abstract

The invention discloses a power transmission line fault positioning method and system based on ground potential end transient information perception, and relates to the technical field of line fault positioning, and the method comprises the steps: obtaining a voltage traveling wave signal and a current abrupt change rate signal detected by a trunk line and branch node sensor array in a power transmission line; processing the voltage traveling wave signal and the current abrupt change rate signal to obtain traveling wave components and polarity-amplitude calibration results of different modes; according to traveling wave components and polarity-amplitude calibration results of different modes, in combination with topological parameters of the power transmission line and spatial distribution of a ground potential end sensor, adjusting edge weights or connection relationships in the graph structure to obtain a composite feature matrix and an edge weight matrix of each node; and inputting the composite feature matrix and the edge weight matrix of each node into a graph neural network positioning model, and outputting a power transmission line fault positioning result. According to the invention, through ground potential end information fusion and the dynamic graph neural network, accurate fault positioning under a complex power transmission topology is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of transmission line fault monitoring, and in particular to a transmission line fault locating method and system based on ground potential terminal transient information perception. Background Art

[0002] As the scale of power grids expands and their structure becomes more complex, the efficiency and accuracy of transmission line fault location become the key to ensuring the safe and stable operation of power systems.

[0003] Existing methods for locating transmission line faults primarily include traveling wave methods, impedance methods, and intelligent ranging methods. The traveling wave method relies on dual-end clock synchronization and is susceptible to wave head identification errors in short lines or complex topologies. The impedance method is limited by transition resistance and parameter errors, resulting in insufficient accuracy. While intelligent ranging methods (such as neural networks) can handle nonlinear characteristics, they lack physical model constraints and have limited generalization capabilities. However, traditional methods utilize only single sensor data and fail to fully exploit spatiotemporal correlations, resulting in low fault location accuracy under complex operating conditions. Summary of the Invention

[0004] To address the problem of low fault location accuracy under complex working conditions caused by the existing technology using single sensor data and not fully exploiting spatiotemporal correlation features, this invention provides a transmission line fault location method and system based on ground potential terminal transient information perception. By integrating ground potential terminal information with a dynamic graph neural network, this method achieves accurate fault location under complex transmission topologies. The specific technical solution is as follows:

[0005] A method for locating faults in a power transmission line based on ground potential terminal transient information perception, comprising:

[0006] S1. Acquire fault transient information detected by a sensor array at the trunk and branch nodes of the transmission line deployed at the ground potential end, including voltage traveling wave signals and current mutation rate signals of sensors at each node;

[0007] S2. Perform variational modal decomposition on the voltage traveling wave signal to obtain traveling wave components of different modes, and perform polarity judgment and amplitude measurement on the current mutation rate signal to obtain a polarity-amplitude calibration result of the current mutation rate signal;

[0008] S3. Based on the traveling wave components and polarity-amplitude calibration results of different modes, combined with the topological parameters of the transmission line and the spatial distribution of the ground potential end sensors, the edge weights or connection relationships in the graph structure are adjusted to obtain the composite feature matrix and edge weight matrix of each node;

[0009] S4. Input the composite feature matrix and edge weight matrix of each node into the graph neural network positioning model, and output the transmission line fault location results, including:

[0010] The composite feature matrix of each node and the edge weight matrix are embedded by a graph embedding layer to output a node embedding vector fused with local topological information;

[0011] The node embedding vector is dynamically modeled in space-time by a space-time correlation layer, and the weights of different time points and different nodes are adaptively allocated by using an attention mechanism to output a global feature matrix enhanced by space-time correlation;

[0012] The global feature matrix is mapped to a fault location space by a fault mapping layer, and a binary branch residual network is used to learn the mapping functions of fault section identification and fault distance estimation, and finally a multi-dimensional vector result including fault probability distribution and fault point coordinates is output.

[0013] Preferably, the acquired fault transient information detected by the trunk and branch node sensor array deployed in the ground potential terminal transmission line includes:

[0014] The voltage traveling wave signal and the current mutation rate signal at the time of fault occurrence are synchronously collected by the sensor array, and the collected voltage traveling wave signal is time-synchronized and calibrated.

[0015] Preferably, the voltage traveling wave signal is decomposed by variational mode to obtain traveling wave components of different modes, including:

[0016] The voltage traveling wave signal is filtered;

[0017] Based on the mode number and the penalty factor set for variational mode decomposition, the center frequency of each mode component is solved by iterative optimization for the filtered voltage traveling wave signal;

[0018] According to the center frequency of each mode component, a preset number of main mode components are extracted as traveling wave propagation features, and the energy proportion and time-frequency distribution of each component are calculated.

[0019] Preferably, after the voltage traveling wave signal is decomposed by variational mode to obtain traveling wave components of different modes, it further includes:

[0020] The wavelet entropy value of the traveling wave component is calculated, and if the wavelet entropy value is greater than a preset threshold, the corresponding traveling wave component is removed.

[0021] Preferably, the polarity judgment and amplitude measurement of the current mutation rate signal to obtain the polarity-amplitude calibration result of the current mutation rate signal include:

[0022] The sliding window difference method is used to calculate the current mutation rate;

[0023] The current mutation rate is compared with a preset threshold to determine the mutation polarity;

[0024] The mutation polarity is normalized to generate a scalarized polarity-amplitude feature vector.

[0025] Preferably, the edge weights or connection relationships in the graph structure are adjusted based on the traveling wave components and polarity-amplitude calibration results of different modes, combined with the topological parameters of the transmission line and the spatial distribution of the ground potential end sensors, to obtain the composite feature matrix and edge weight matrix of each node, including:

[0026] The initial graph structure is constructed based on the transmission line topology parameters, where nodes represent sensor locations and edges represent electrical connection relationships between adjacent nodes.

[0027] Based on the energy proportion and propagation delay of the traveling wave component, the weight of the edge is dynamically adjusted. The weight calculation formula is:

[0028]

[0029] Where, E i 、E j are the traveling wave energies of adjacent nodes, τ ij is the traveling wave propagation delay, ∈ is the smoothing coefficient;

[0030] The polarity-amplitude characteristic vector of the current mutation rate is spliced ​​with the traveling wave energy characteristic to generate a composite characteristic matrix of the node.

[0031] Preferably, the global feature matrix is ​​mapped to the fault location space through the fault mapping layer, and the mapping functions of fault section identification and fault distance estimation are learned respectively through the two-branch residual network, and the final output includes a multidimensional vector result including the fault probability distribution and the fault point location coordinates.

[0032] A two-branch residual network is used to handle different tasks. The fully connected layer and softmax of the first branch output the probability distribution of the fault section, and the regression layer of the second branch outputs the estimated distance between the fault point and the nearest sensor.

[0033] The outputs of the two-branch residual network are weightedly fused to generate the final multi-dimensional fault location vector.

[0034] A transmission line fault location system based on ground potential terminal transient information perception, applying the aforementioned transmission line fault location method based on ground potential terminal transient information perception, comprising:

[0035] A signal acquisition unit is used to acquire fault transient information detected by a sensor array at the trunk and branch nodes of the transmission line deployed at the ground potential end, including voltage traveling wave signals and current mutation rate signals of sensors at each node;

[0036] The signal processing unit is used to perform variational modal decomposition on the voltage traveling wave signal to obtain traveling wave components of different modes, and to perform polarity judgment and amplitude measurement on the current mutation rate signal to obtain the polarity-amplitude calibration result of the current mutation rate signal;

[0037] The spatiotemporal graph structure processing unit is used to adjust the edge weights or connection relationships in the graph structure based on the traveling wave components and polarity-amplitude calibration results of different modes, combined with the topological parameters of the transmission line and the spatial distribution of the ground potential end sensors, to obtain the composite feature matrix and edge weight matrix of each node;

[0038] The fault location output unit is used to input the composite feature matrix and edge weight matrix of each node into the graph neural network location model and output the transmission line fault location results, including:

[0039] The graph embedding layer embeds the composite feature matrix and edge weight matrix of each node into the graph structure, and outputs the node embedding vector that integrates the local topological information;

[0040] The spatiotemporal dynamic modeling of node embedding vectors is performed through the spatiotemporal correlation layer, and the attention mechanism is used to adaptively assign weights at different times and nodes to output a global feature matrix with enhanced spatiotemporal correlation.

[0041] The global feature matrix is ​​mapped to the fault location space through the fault mapping layer. The mapping functions of fault section identification, fault distance estimation and fault type discrimination are learned respectively through a two-branch residual network. The final output is a multidimensional vector including fault probability distribution, fault point location coordinates and fault type confidence.

[0042] A computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the aforementioned transmission line fault location method based on ground potential terminal transient information perception.

[0043] A processor is used to run a program, wherein when the program is run, the above-mentioned transmission line fault location method based on ground potential terminal transient information perception is executed.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] The transmission line fault location method based on ground potential end transient information perception of the present invention monitors the fault transient information of the transmission line, including voltage traveling wave signals and current mutation rate signals, by deploying a sensor array at the ground potential end; processing the voltage traveling wave signal through variational mode decomposition to obtain traveling wave components of different modes, and simultaneously performing polarity judgment and amplitude measurement on the current mutation rate signal; combining the topological parameters of the transmission line and the spatial distribution of the ground potential end sensors, adjusting the edge weights or connection relationships in the graph structure, and constructing a composite feature matrix and edge weight matrix of each node, taking into account the actual situation of the transmission line and the sensor layout, so that the fault location model is more in line with the actual operating environment and improves the positioning accuracy; utilizing the graph neural network positioning model, through the processing of the graph embedding layer, the spatiotemporal correlation layer and the fault mapping layer, deep fusion of fault information is achieved, and a multi-dimensional vector including the fault probability distribution and the fault point location coordinates can be output, providing accurate fault location information. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly describes the drawings required for the specific embodiments or the description of the prior art. Similar elements or parts are generally identified by similar reference numerals throughout the drawings. Elements or parts in the drawings are not necessarily drawn to scale.

[0047] Figure 1 This is a flow chart of the transmission line fault location method based on ground potential terminal transient information perception of the present invention.

[0048] Figure 2 This is a schematic diagram of the power transmission line fault location system based on ground potential terminal transient information perception of the present invention. DETAILED DESCRIPTION

[0049] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0050] It should be understood that when used in this specification, the terms "include" and "comprising" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0051] It should also be understood that the terms used in the present specification are only for the purpose of describing particular embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms unless the context clearly indicates otherwise.

[0052] It should be further understood that the term “and / or” used in the description of the present invention refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0053] See the following examples Figure 1 and Figure 2 .

[0054] The present application provides a method for locating a transmission line fault based on ground potential transient information perception, including:

[0055] S1. Acquire fault transient information detected by a sensor array at the trunk and branch nodes of the transmission line deployed at the ground potential end, including voltage traveling wave signals and current mutation rate signals of sensors at each node;

[0056] By deploying redundant ground potential sensors within a set range on both sides of the impedance mutation points (such as T-junctions, tension towers, etc.) of the transmission line trunk, the voltage and current changes in the area are monitored in real time through the deployed sensors. The redundant settings improve data accuracy and avoid information loss due to the failure of a single sensor.

[0057] For branch nodes, bipolar sensors are installed to sense changes in the direction of the current. When a fault occurs, the direction of the fault propagation can be determined based on the direction of the current mutation.

[0058] Sensors at each node monitor voltage traveling wave signals and current mutation rate signals in real time. When any sensor detects that the voltage or current mutation rate exceeds a preset value, it immediately triggers synchronized recording for all sensors in the network, ensuring that all sensors simultaneously begin recording transient fault information, including the temporal changes in the voltage traveling wave and current mutation rate signals at each node.

[0059] S2. Perform variational modal decomposition on the voltage traveling wave signal to obtain traveling wave components of different modes, and perform polarity judgment and amplitude measurement on the current mutation rate signal to obtain a polarity-amplitude calibration result of the current mutation rate signal;

[0060] The captured voltage traveling wave signal is fed into a variational modal decomposition algorithm, which decomposes it into multiple modal traveling wave components based on the signal's frequency and time scale characteristics. Each modal component represents the signal's variation within a specific frequency range and time interval, reflecting the propagation and reflection of the voltage traveling wave on the transmission line at the moment of the fault.

[0061] The current mutation rate signal is processed to first determine its polarity—that is, to determine whether the current mutation is changing in a positive or negative direction—by analyzing the numerical sign of the signal. Simultaneously, the amplitude of the current mutation rate signal is accurately measured. This amplitude reflects the severity of the current change at the moment of the fault and is closely related to factors such as the severity of the fault and the distance to the fault point. By combining the polarity and amplitude information, a polarity-amplitude calibration result for the current mutation rate signal is obtained.

[0062] S3. Based on the traveling wave components and polarity-amplitude calibration results of different modes, combined with the topological parameters of the transmission line and the spatial distribution of the ground potential end sensors, the edge weights or connection relationships in the graph structure are adjusted to obtain the composite feature matrix and edge weight matrix of each node;

[0063] The initially constructed graph structure is adjusted based on the actual topological parameters of the transmission line and the spatial distribution of the ground potential end sensors on the transmission line (including information such as the specific deployment coordinates of the sensors at the trunk and branch nodes). Based on the traveling wave components and polarity-amplitude calibration results of different modes, it is evaluated whether the weights or connection relationships of each edge in the graph structure need to be changed. For example, if the propagation characteristics of the traveling wave component of a certain mode between two adjacent nodes change significantly, or the polarity-amplitude calibration results of the current mutation rate indicate that the fault may propagate along a specific path, the weights or connection status of the corresponding edges are adjusted accordingly to more accurately reflect the actual operating status of the transmission line after the fault occurs.

[0064] After adjustment, each node's voltage traveling wave signal characteristics, current mutation rate signal characteristics (polarity, amplitude), and related information adjusted based on topology and spatial distribution are integrated to form a composite feature matrix corresponding to each node. At the same time, based on the adjusted edge weights or connection relationships, an edge weight matrix is ​​constructed to describe the electrical connection tightness and signal transmission characteristics between nodes in the event of a fault, providing the graph structure data input to the graph neural network positioning model.

[0065] S4. Input the composite feature matrix and edge weight matrix of each node into the graph neural network positioning model, and output the transmission line fault location results, including:

[0066] The graph embedding layer embeds the composite feature matrix and edge weight matrix of each node into the graph structure, and outputs the node embedding vector that integrates the local topological information;

[0067] The spatiotemporal dynamic modeling of node embedding vectors is performed through the spatiotemporal correlation layer, and the attention mechanism is used to adaptively assign weights at different times and nodes to output a global feature matrix with enhanced spatiotemporal correlation.

[0068] The definition of the spatiotemporal correlation layer is as follows:

[0069] Input of the spatiotemporal correlation layer: node embedding vector (from the graph embedding layer).

[0070] Output of the spatiotemporal correlation layer: global feature matrix Where L is the number of spatiotemporal correlation layers and is the output feature dimension.

[0071] Spatiotemporal dynamic modeling: implemented through multi-layer graph neural networks (such as convolutional networks) and attention mechanisms:

[0072] H (l+1) =Attention(GCN(H (l) ,W))

[0073] Among them, W is the edge weight matrix, GCN is the graph convolution operation, and Attention is the attention mechanism used to dynamically adjust the weights of node features.

[0074] Final output: After L layers of processing, the global feature matrix H is obtained ( L) ,This matrix fuses the temporal and spatial correlation information and is used as the input of the fault mapping layer.

[0075] The fault localization space is a multidimensional vector space used to represent the fault localization results and is defined as:

[0076]

[0077] Among them, Y prob ∈R is the probability distribution of the fault section, satisfying Y dist ∈[0,1] is the normalized distance between the fault point and the nearest sensor, and the actual distance is calculated as d=Y dist × L (L is the sensor coverage length).

[0078] The global feature matrix is ​​mapped to the fault location space through the fault mapping layer, and the mapping functions of fault section identification and fault distance estimation are learned respectively through a two-branch residual network, and the final output is a multidimensional vector result including the fault probability distribution and the fault point location coordinates.

[0079] The fault mapping layer is defined as follows:

[0080] Input of the fault mapping layer: the global feature matrix is ​​H∈RN×D , where N is the number of nodes and D is the feature dimension;

[0081] Output of the fault mapping layer: fault location result vector Y = [Y prob ,Y dist ]∈R K+1 , where Y prob ∈RK is the probability distribution of the fault section (K is the preset number of fault sections), Y dist ∈[0,1] is the normalized distance between the fault point and the nearest sensor.

[0082] Fault section identification branch: implemented through a fully connected layer (output dimension is K) and a Softmax activation function:

[0083] Y prob =Softmax(W prob H flat +b prob )

[0084] Among them, W prob ∈R K×(N×D) is the weight matrix, b prob ∈R K is the bias vector, H flat ∈R N×D is the flattened global feature matrix.

[0085] Fault distance estimation branch: implemented through a fully connected layer (output dimension is 1) and a Sigmoid activation function:

[0086] Y dist =Sigmoid(w dist H flat +b dist )

[0087] Among them, w dist ∈R N×D is the weight vector, b dist ∈R is the bias term.

[0088] The final output is a multi-dimensional vector result containing the fault probability distribution and the coordinates of the fault point, which can provide intuitive and accurate fault location information for the operation and maintenance personnel of the transmission line, so that they can quickly go to the fault point for inspection and processing.

[0089] The present invention's transmission line fault location method based on ground-potential transient information sensing uses a ground-potential sensor array to synchronously collect voltage traveling wave signals and current mutation rate signals. Variational mode decomposition deeply analyzes the voltage traveling wave, while polarity-amplitude calibration refines the current characteristics. Combining these two methods and integrating line topology and sensor distribution information effectively overcomes the issues of single-signal susceptibility to interference and insufficient feature extraction under complex operating conditions. By adjusting the edge weights and connectivity of the graph structure based on multi-source features, the model dynamically adapts to changes in transmission line topology and environmental differences, reflecting the propagation patterns of fault traveling waves. This effectively reduces location errors caused by complex topology in lines with intertwined trunk lines and complex terrain. A graph neural network positioning model is implemented through a layered approach of graph embedding, spatiotemporal correlation, and fault mapping. Graph embedding integrates local topology, while spatiotemporal correlation utilizes an attention mechanism to focus on key information. Fault mapping, coupled with a two-branch residual network, identifies fault segments and distances. The resulting output includes a multidimensional result containing probability distribution and location coordinates, enabling accurate fault diagnosis. This provides a rich basis for operational decision-making and improves the efficiency and intelligence of power grid operations.

[0090] Specifically, the acquisition of fault transient information detected by a sensor array deployed at a trunk and branch node in a ground potential end transmission line includes:

[0091] The voltage traveling wave signal and the current mutation rate signal when a fault occurs are synchronously collected by the sensor array, and time synchronization calibration is performed on the collected voltage traveling wave signal.

[0092] The voltage traveling wave signal and current mutation rate signal when a fault occurs are synchronously collected through the sensor array, and the voltage traveling wave signal is time-synchronized and calibrated to ensure the time consistency of the collected data. This sets a precise time reference for fault location and avoids error accumulation caused by time differences.

[0093] Specifically, performing variational modal decomposition on the voltage traveling wave signal to obtain traveling wave components of different modes includes:

[0094] Filtering the voltage traveling wave signal;

[0095] For the filtered voltage traveling wave signal, the center frequency of each modal component is solved through iterative optimization based on the set mode number and penalty factor of variational modal decomposition;

[0096] According to the center frequency of each modal component, a preset number of main modal components are extracted as the traveling wave propagation characteristics, and the energy proportion and time-frequency distribution of each component are calculated.

[0097] The voltage traveling wave signal is filtered to remove noise and interference in the signal and improve the quality of the signal. Based on the set modal number and penalty factor of the variational mode decomposition, the center frequency of each modal component is solved by iterative optimization, and the center frequency of each modal component is adaptively determined according to the characteristics of the signal, so that the voltage traveling wave signal is decomposed into modal components of different frequency ranges. The main modal component is the most representative and influential part of the signal, which carries the main information when the fault occurs. By extracting the main modal component and calculating its energy proportion and time-frequency distribution, the distribution of the fault signal at different frequencies and times can be understood.

[0098] Specifically, after the voltage traveling wave signal is decomposed by variational mode decomposition to obtain traveling wave components of different modes, the method further includes:

[0099] The wavelet entropy value of the traveling wave component is calculated, and if the wavelet entropy value is greater than a preset threshold, the corresponding traveling wave component is removed.

[0100] Wavelet entropy is an index for measuring signal complexity and uncertainty. The signal is decomposed at different frequencies and time scales by wavelet transform, and then the entropy value is calculated according to the wavelet coefficients. For each traveling wave component of the voltage traveling wave signal, its corresponding wavelet entropy value can be calculated. When the wavelet entropy value of a certain traveling wave component is greater than a preset threshold, it is removed from the subsequent analysis. By removing the traveling wave component with a wavelet entropy value greater than the preset threshold, noise and interference in the signal can be removed, and only those truly fault-related characteristic components are retained.

[0101] Specifically, the polarity judgment and amplitude measurement of the current mutation rate signal to obtain the polarity-amplitude calibration result of the current mutation rate signal include:

[0102] The sliding window difference method is used to calculate the current mutation rate;

[0103] The current mutation rate is compared with a preset threshold to determine the mutation polarity;

[0104] The mutation polarity is normalized to generate a scalarized polarity-amplitude feature vector.

[0105] In specific implementation, the calculated current mutation rate is compared with a preset threshold. When the current mutation rate is greater than the positive threshold, it is determined that the current has a positive mutation; when it is less than the negative threshold, it is determined that the current has a negative mutation; and when it is within the threshold range, it is considered that the current has no obvious mutation.

[0106] By combining the sliding window differencing method with threshold judgment, it can identify sudden changes in current at the moment of a fault. Even in complex electromagnetic environments or conditions with large current fluctuations, it can reliably detect subtle fault current changes, thereby improving fault detection sensitivity. The polarity-amplitude eigenvector transforms complex current signals into concise numerical representations, reducing data dimensionality and computational complexity. Polarity information can also assist in determining fault direction, while amplitude information reflects fault severity, providing dual-dimensional key clues for fault location.

[0107] Specifically, the edge weights or connection relationships in the graph structure are adjusted based on the traveling wave components and polarity-amplitude calibration results of different modes, combined with the topological parameters of the transmission line and the spatial distribution of the ground potential end sensors, to obtain the composite feature matrix and edge weight matrix of each node, including:

[0108] The initial graph structure is constructed based on the transmission line topology parameters, where nodes represent sensor locations and edges represent electrical connection relationships between adjacent nodes.

[0109] Based on the energy proportion and propagation delay of the traveling wave component, the weight of the edge is dynamically adjusted. The weight calculation formula is:

[0110]

[0111] Where, E i 、E j are the traveling wave energies of adjacent nodes, τ ij is the traveling wave propagation delay, ∈ is the smoothing coefficient;

[0112] The polarity-amplitude characteristic vector of the current mutation rate is spliced ​​with the traveling wave energy characteristic to generate a composite characteristic matrix of the node.

[0113] The topological parameters of a transmission line include detailed information such as the line layout, node connectivity, and the electrical characteristics of each line segment. When constructing the initial graph structure, each ground potential sensor location is defined as a node in the graph. A node represents a point in an actual transmission line where transient fault information can be obtained. Edges represent the electrical connections between adjacent nodes, reflecting the current flow path and signal propagation channels within the transmission line. Because traveling wave components propagate, different nodes receive varying amounts of traveling wave energy. By calculating the proportion of traveling wave energy received by each node to the total energy, we can understand the energy distribution of the traveling wave at different locations. It takes a certain amount of time for a traveling wave to propagate from one node to another; this time interval is called the propagation delay. This propagation delay is related to factors such as the length of the transmission line, the wave velocity, and the electrical parameters of the line. By measuring or calculating the propagation delay, we can determine the propagation speed and path of the traveling wave between different nodes.

[0114] Edge weights comprehensively reflect the difference in traveling wave energy between adjacent nodes and the impact of propagation delay. The greater the energy difference and the shorter the propagation delay, the greater the edge weight, and vice versa.

[0115] By constructing an initial graph structure, the topology of the transmission line is converted into a graph. Graph theory methods are then used to analyze and process fault information, visually reflecting the electrical connections between nodes in the transmission line. Dynamically adjusting edge weights can be based on the energy contribution and propagation delay of the traveling wave components. Edges with higher weights indicate paths with lower energy loss and faster propagation speed when traveling waves propagate along these paths, representing the primary fault propagation paths. The resulting composite feature matrix combines the polarity-amplitude characteristics of the current mutation rate with the traveling wave energy characteristics, utilizing two different types of fault signature information to enhance fault signature representation.

[0116] Specifically, the global feature matrix is ​​mapped to the fault location space through the fault mapping layer, and the mapping functions for fault section identification and fault distance estimation are learned respectively through the two-branch residual network. The final output includes a multidimensional vector of fault probability distribution and fault point location coordinates. The results include:

[0117] Different tasks are handled separately through a two-branch residual network, including:

[0118] Output the probability distribution of the fault section through the fully connected layer and softmax of the first branch;

[0119] Input the global feature matrix (dimension is N×D, N is the number of nodes, D is the feature dimension), flatten the features and input them into the fully connected layer (output dimension = K, which is the preset number of fault sections). The activation function is Softmax, and the probability distribution of each fault section is output as P = [p1, p2, ..., p K ].

[0120] The regression layer of the second branch outputs the estimated distance between the fault point and the nearest sensor;

[0121] Input the same global feature matrix, fully connected layer (output dimension = 1), activation function is Sigmoid, and output is the normalized distance d∈[0,1] between the fault point and the nearest sensor (the actual distance is calculated by d×L, L is the length of the sensor coverage range).

[0122] The outputs of the two-branch residual network are weightedly fused to generate the final multi-dimensional fault location vector.

[0123] The weight coefficients α and β are set according to the importance of the task, and the final output vector is

[0124] The first branch flattens the global features and then outputs the probability distribution of the fault segment through the fully connected layer and Softmax, which can locate the fault area from a holistic perspective. The second branch outputs the distance between the fault point and the nearest sensor as a normalized distance, which can be converted into the actual distance based on the sensor coverage range. The combination of the two enables precise positioning from the area to the specific location, avoiding errors caused by single judgment.

[0125] The present application also provides a transmission line fault location system based on ground potential terminal transient information perception, which applies the aforementioned transmission line fault location method based on ground potential terminal transient information perception, including:

[0126] A signal acquisition unit is used to acquire fault transient information detected by a sensor array at the trunk and branch nodes of the transmission line deployed at the ground potential end, including voltage traveling wave signals and current mutation rate signals of sensors at each node;

[0127] The signal processing unit is used to perform variational modal decomposition on the voltage traveling wave signal to obtain traveling wave components of different modes, and to perform polarity judgment and amplitude measurement on the current mutation rate signal to obtain the polarity-amplitude calibration result of the current mutation rate signal;

[0128] The spatiotemporal graph structure processing unit is used to adjust the edge weights or connection relationships in the graph structure based on the traveling wave components and polarity-amplitude calibration results of different modes, combined with the topological parameters of the transmission line and the spatial distribution of the ground potential end sensors, to obtain the composite feature matrix and edge weight matrix of each node;

[0129] The fault location output unit is used to input the composite feature matrix and edge weight matrix of each node into the graph neural network location model and output the transmission line fault location results, including:

[0130] The graph embedding layer embeds the composite feature matrix and edge weight matrix of each node into the graph structure, and outputs the node embedding vector that integrates the local topological information;

[0131] The spatiotemporal dynamic modeling of node embedding vectors is performed through the spatiotemporal correlation layer, and the attention mechanism is used to adaptively assign weights at different times and nodes to output a global feature matrix with enhanced spatiotemporal correlation.

[0132] The global feature matrix is ​​mapped to the fault location space through the fault mapping layer. The mapping functions of fault section identification, fault distance estimation and fault type discrimination are learned respectively through a two-branch residual network. The final output is a multidimensional vector including fault probability distribution, fault point location coordinates and fault type confidence.

[0133] The functional explanation of each unit in this embodiment is the same as that in the embodiment of the transmission line fault location method based on ground potential terminal transient information perception, and the technical effect is the same, so it will not be repeated here.

[0134] An embodiment of the present application further provides a computer-readable storage medium, which includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the aforementioned transmission line fault location method based on ground potential terminal transient information perception.

[0135] The technical effect of this embodiment is the same as the technical effect of the transmission line fault location method based on ground potential terminal transient information perception in the embodiment, and will not be repeated here.

[0136] The present invention can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations, such as personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments that include any of the above.

[0137] An embodiment of the present application further provides a processor, which is used to run a program, wherein when the program is run, the aforementioned method for locating a power transmission line fault based on ground potential terminal transient information perception is executed.

[0138] The technical effect of this embodiment is the same as the technical effect of the transmission line fault location method based on ground potential terminal transient information perception in the embodiment, and will not be repeated here.

[0139] The processor in this embodiment may be a central processing unit (CPU), a controller, a microcontroller, or other data processing chips.

[0140] Those skilled in the art will appreciate that the units of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition of each example has been generally described in terms of function in the above description. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0141] In the embodiments provided by the present invention, it should be understood that the division of units is merely a logical function division, and there may be other division methods in actual implementation, for example, multiple units can be combined into one unit, one unit can be split into multiple units, or some features can be ignored, etc.

[0142] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing unit, each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0143] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-0nly Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk, etc., various media that can store program code.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention, and they should all be included in the scope of the specification of the present invention.

Claims

1. A method for locating transmission line faults based on ground potential transient information perception, characterized in that: include: S1. Acquire fault transient information detected by a sensor array at the trunk and branch nodes of the transmission line deployed at the ground potential end, including voltage traveling wave signals and current mutation rate signals of sensors at each node; S2. Perform variational modal decomposition on the voltage traveling wave signal to obtain traveling wave components of different modes, and perform polarity judgment and amplitude measurement on the current mutation rate signal to obtain a polarity-amplitude calibration result of the current mutation rate signal; S3. Based on the traveling wave components and polarity-amplitude calibration results of different modes, combined with the topological parameters of the transmission line and the spatial distribution of the ground potential end sensors, the edge weights in the graph structure are adjusted to obtain the composite feature matrix and edge weight matrix of each node, including: The initial graph structure is constructed based on the transmission line topology parameters, where nodes represent sensor locations and edges represent electrical connection relationships between adjacent nodes. Dynamically adjust edge weights based on the energy proportion and propagation delay of the traveling wave component; The polarity-amplitude characteristic vector of the current mutation rate signal is spliced ​​with the traveling wave energy characteristic to generate a composite characteristic matrix of the node; S4. Input the composite feature matrix and edge weight matrix of each node into the graph neural network positioning model, and output the transmission line fault location results, including: The graph embedding layer embeds the composite feature matrix and edge weight matrix of each node into the graph structure, and outputs the node embedding vector that integrates the local topological information; The spatiotemporal dynamic modeling of node embedding vectors is performed through the spatiotemporal correlation layer, and the attention mechanism is used to adaptively assign weights at different times and nodes to output a global feature matrix with enhanced spatiotemporal correlation. The global feature matrix is ​​mapped to the fault location space through the fault mapping layer, and the mapping functions of fault section identification and fault distance estimation are learned respectively through a two-branch residual network, and the final output is a multidimensional vector result including the fault probability distribution and the fault point location coordinates.

2. The method for locating a power transmission line fault based on ground potential terminal transient information perception according to claim 1, characterized in that: The acquisition of fault transient information detected by a sensor array at a trunk and branch node deployed in a ground potential end transmission line includes: The voltage traveling wave signal and the current mutation rate signal when a fault occurs are synchronously collected by the sensor array, and time synchronization calibration is performed on the collected voltage traveling wave signal.

3. The method for locating a power transmission line fault based on ground potential terminal transient information perception according to claim 1, characterized in that: The variational modal decomposition of the voltage traveling wave signal to obtain traveling wave components of different modes includes: Filtering the voltage traveling wave signal; For the filtered voltage traveling wave signal, the center frequency of each modal component is solved through iterative optimization based on the set mode number and penalty factor of variational modal decomposition; According to the center frequency of each modal component, a preset number of main modal components are extracted as the traveling wave propagation characteristics, and the energy proportion and time-frequency distribution of each component are calculated.

4. The method for locating a power transmission line fault based on ground potential terminal transient information perception according to claim 1, characterized in that: After performing variational modal decomposition on the voltage traveling wave signal to obtain traveling wave components of different modes, the method further includes: Calculate the wavelet entropy value of the traveling wave component. If the wavelet entropy value is greater than a preset threshold, remove the corresponding traveling wave component.

5. The method for locating power transmission line faults based on ground potential terminal transient information perception according to claim 3, characterized in that: The polarity determination and amplitude measurement of the current mutation rate signal to obtain the polarity-amplitude calibration result of the current mutation rate signal includes: The sliding window difference method is used to calculate the current mutation rate; Compare the current mutation rate with the preset threshold to determine the mutation polarity; The mutation polarity is normalized to generate a scalarized polarity-amplitude feature vector.

6. The method for locating power transmission line faults based on ground potential terminal transient information perception according to claim 1, characterized in that: The global feature matrix is ​​mapped to the fault location space through the fault mapping layer, and the mapping functions of fault section identification and fault distance estimation are learned respectively through the two-branch residual network. The final output includes a multi-dimensional vector of fault probability distribution and fault point location coordinates. The results include: A two-branch residual network is used to handle different tasks. The fully connected layer and softmax of the first branch output the probability distribution of the fault section, and the regression layer of the second branch outputs the estimated distance between the fault point and the nearest sensor. The outputs of the two-branch residual network are weightedly fused to generate the final multi-dimensional fault location vector.

7. A transmission line fault location system based on ground potential terminal transient information perception, characterized in that: The method for locating a transmission line fault based on ground potential terminal transient information perception according to any one of claims 1 to 6 comprises: A signal acquisition unit is used to acquire fault transient information detected by a sensor array at the trunk and branch nodes of the transmission line deployed at the ground potential end, including voltage traveling wave signals and current mutation rate signals of sensors at each node; The signal processing unit is used to perform variational modal decomposition on the voltage traveling wave signal to obtain traveling wave components of different modes, and to perform polarity judgment and amplitude measurement on the current mutation rate signal to obtain the polarity-amplitude calibration result of the current mutation rate signal; The spatiotemporal graph structure processing unit is used to adjust the edge weights or connection relationships in the graph structure based on the traveling wave components and polarity-amplitude calibration results of different modes, combined with the topological parameters of the transmission line and the spatial distribution of the ground potential end sensors, to obtain the composite feature matrix and edge weight matrix of each node; The fault location output unit is used to input the composite feature matrix and edge weight matrix of each node into the graph neural network location model and output the transmission line fault location results, including: The graph embedding layer embeds the composite feature matrix and edge weight matrix of each node into the graph structure, and outputs the node embedding vector that integrates the local topological information; The spatiotemporal dynamic modeling of node embedding vectors is performed through the spatiotemporal correlation layer, and the attention mechanism is used to adaptively assign weights at different times and nodes to output a global feature matrix with enhanced spatiotemporal correlation. The global feature matrix is ​​mapped to the fault location space through the fault mapping layer. The mapping functions of fault section identification, fault distance estimation and fault type discrimination are learned respectively through a two-branch residual network. The final output is a multidimensional vector including fault probability distribution, fault point location coordinates and fault type confidence.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute the transmission line fault location method based on ground potential terminal transient information perception according to any one of claims 1 to 6.

9. A processor, characterized in that: The processor is used to run a program, wherein the program, when running, executes the transmission line fault location method based on ground potential terminal transient information perception according to any one of claims 1 to 6.

Images