An intelligent power grid transmission line fault location system

Abstract

The present invention belongs to the technical field of power grid transmission. The present invention discloses an intelligent power grid transmission line fault location system, which includes: collecting power operation data, spatial data, and environmental data; correcting the power operation data based on the constructed power correction model to obtain corrected power data; using a neural network algorithm to establish a topology for the corrected power data and spatial data to obtain a power grid topology map, and performing graph decomposition on the power grid topology map through an ant colony algorithm to obtain B topological subgraphs; constructing a fault prediction model for predicting fault types; performing fault prediction on the B topological subgraphs through the fault prediction model, marking the topological subgraphs with predicted fault types as fault subgraphs, and locating the fault points in the fault subgraphs through a particle swarm algorithm to determine the final fault location; greatly improving the fault location speed.

Description

Technical Field

[0001] The present invention relates to the technical field of power grid transmission, and more specifically, to an intelligent power grid transmission line fault location system. Background Art

[0002] A patent with the application publication number CN118209817A discloses a method and system for identifying faults in transmission lines based on environmental characteristics; the method includes: performing real-time monitoring and data preprocessing on multiple first transmission line nodes in a power system network to obtain voltage waveform characteristics and current waveform characteristics; performing abnormal node identification to obtain multiple second transmission line nodes; performing data fusion to obtain multi-dimensional fusion data; performing feature extraction to obtain initial multi-dimensional fusion features and performing feature dimensionality reduction processing to obtain target multi-dimensional fusion features; performing transmission line fault identification through a transmission line fault identification model, and outputting the transmission line fault types of each second transmission line node; determining the initial transmission line fault treatment plan for each second transmission line node and performing global optimization analysis to obtain the global transmission line fault treatment plan for the power system network, thereby improving the accuracy of transmission line fault identification.

[0003] Current fault detection is mainly based on single electrical signals, such as current mutations, voltage fluctuations, etc.; however, in a changing natural environment, these single signals are easily interfered by external factors, resulting in false alarms or missed alarms, and there is a lack of multi-dimensional sensing data fusion to comprehensively judge the fault type and fault location; although some transmission lines have started to use digital devices for data monitoring and collection, existing data processing systems mostly adopt traditional fault analysis methods, such as Fourier transform or simple statistical analysis; these methods can often only process limited input data, and when faced with the massive data collected by a large-scale sensor network, the analysis efficiency is low, and accurate fault location judgments cannot be made in a short time. In addition, the limitations of data processing capabilities also lead to inaccurate identification of complex fault types.

[0004] In view of this, the present invention proposes an intelligent power grid transmission line fault location system to solve the above problems. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above object, the present invention provides the following technical solution: An intelligent power grid transmission line fault location system, including: a data acquisition module: collecting power operation data, spatial data, and environmental data;

[0006] A data correction module: correcting the power operation data based on a constructed power correction model to obtain corrected power data;

[0007] Graph construction module: It includes a graph construction unit and a graph decomposition unit. The graph construction unit uses a neural network algorithm to establish a topology for the corrected power data and spatial data to obtain a power grid topology graph. The graph decomposition unit decomposes the power grid topology graph to obtain B topological subgraphs;

[0008] Fault model establishment module: Collect F groups of historical power fault data and build a fault prediction model for predicting fault types based on the historical power fault data;

[0009] Fault location module: Perform fault prediction on the B topological subgraphs through the fault prediction model, mark the topological subgraphs with predicted fault types as faulty subgraphs, and locate the fault points in the faulty subgraphs through the particle swarm algorithm to determine the final fault positions.

[0010] Furthermore, the power operation data includes: electric field data, magnetic field data, mechanical stress data, and vibration data; the spatial data includes: accuracy, latitude, and proximity relationship; the environmental data includes: temperature, wind force, and humidity.

[0011] Furthermore, the acquisition method of the corrected power data includes:

[0012] Based on the constructed power correction model, using the power operation data and environmental data as the power correction model for prediction to obtain the corrected power data, which includes corrected electric field data, corrected magnetic field data, corrected mechanical stress data, and corrected vibration data.

[0013] Furthermore, the construction method of the power grid topology graph includes:

[0014] Take each corrected power data and the accuracy and latitude in the spatial data as the attributes of a node to construct power grid nodes. The proximity relationship in the spatial nodes is used as the basis for connecting nodes to construct a power grid node matrix. The values in the power grid node matrix are determined based on the proximity relationship. When the proximity relationship is "yes", the value in the power grid node matrix is 1; when the proximity relationship is "no", the value in the power grid node matrix is 1. Based on the power grid node matrix, construct the association index set of each node. The method of constructing the association index set of each node includes: Based on the power grid node JD represented by each row in the power grid node matrix, the power grid nodes with a value of 1 in each row of the matrix are the associated nodes of the power grid node JD, and the indexes of all associated nodes form the association index set. Take the attributes of each node as the input of the input layer of the graph convolutional network, use the graph convolutional network to aggregate the features of each node to obtain node embedding features, use the node embedding features as topological nodes, and use the proximity relationship in the spatial nodes as the basis for connecting nodes. Through the cosine similarity formula calculation, perform similarity calculation on the connected topological nodes to obtain node similarity values, and use the node similarity values as the weights of the edges to construct the power grid topology graph.

[0015] Furthermore, the formula for feature aggregation is as follows: where represents the node embedding feature of the v-th power grid node after convolution in the i-th convolutional layer, and N v is the associated index set of the v-th power grid node, and d v represents the degree of the v-th power grid node, and d u represents the degree of the u-th power grid node, W l represents the weight matrix of the i-th convolutional layer, b i represents the bias term of the i-th convolutional layer, and σ represents the non-linear activation function.

[0016] Furthermore, the method for graph decomposition of the power grid topology graph includes:

[0017] Starting from each topological node in the power grid topology graph, use the breadth-first algorithm to search for the shortest path between each pair of topological nodes in the power grid topology graph; based on all the shortest paths, initialize G ants, where the value of G is the total number of shortest paths, assign one shortest path to each ant, each ant represents the execution of one shortest path, and assign each ant to different topological nodes in the power grid topology graph; initialize pheromone: initialize the pheromone concentration for each edge, and assign an initial pheromone concentration to each ant; each ant starts from the starting topological node and moves in the power grid topology graph based on the assigned shortest path, and updates the pheromone concentration for all the edges in the shortest path according to the sum of the weights of all the edges in the shortest path corresponding to each ant. The formula for pheromone concentration update is: where τ m,n represents the updated value of the pheromone concentration of the edge connecting the m-th node and the n-th node in the power grid topology graph, ρ represents the pheromone evaporation coefficient, S represents the initial pheromone concentration assigned to the ant, which is used to adjust the pheromone increment, and Sum represents the sum of the weights of all the edges passed by the selected path; when all the ants reach the end of the path, select the edge with the highest pheromone concentration in the power grid topology graph as the intermediate edge, delete the intermediate edge from the power grid topology graph to form a new power grid topology graph. When the power grid topology graph is divided into N unconnected topological subgraphs, perform module evaluation on each topological subgraph to obtain the modularity of each topological subgraph, and repeat until the value of the modularity of each topological subgraph no longer changes, and obtain B topological subgraphs.

[0018] Furthermore, the formula for module evaluation is as follows: where Q j represents the modularity of the j-th topological subgraph, a j represents the total number of edges in the j-th topological subgraph, Represents the degree of the c-th topological node in the j-th topological subgraph. Represents the degree of the d-th topological node in the j-th topological subgraph, q represents the weight of the edge connecting the c-th topological node and the d-th topological node, SIM j Represents all combinations of topological nodes with connected edges in the j-th topological subgraph. c and d represent the indices of topological nodes in the j-th topological subgraph, and c ≠ d.

[0019] Furthermore, the construction method of the fault prediction model includes:

[0020] Collect F groups of historical power fault data. The historical power fault data includes fault categories, electric field fault data, magnetic field fault data, mechanical stress fault data, and vibration fault data. Based on the fault categories, the historical power fault data with the same fault category is divided into one group, obtaining R groups of fault data groups. Perform anomaly detection on the data in each group of fault data groups. The method of performing anomaly detection includes: forming a C×1 fault vector from the electric field fault data, magnetic field fault data, mechanical stress fault data, and vibration fault data in each group of data in the fault data group, where C represents the total number of fault data categories; calculating the similarity value of each fault vector in the fault vector with other fault vectors through the similarity calculation formula, performing an average calculation on the calculated similarity values to obtain the average similarity of each fault vector, calculating the dispersion of each fault vector through the zero-score algorithm based on the average similarity, presetting a dispersion threshold, marking the power fault data with a dispersion greater than or equal to the dispersion threshold as abnormal data, and removing the historical power fault data marked as abnormal data in each group of fault data groups to obtain effective fault data groups; using the effective fault data groups as the training sample set, using the GRU model as the initial model, training the GRU model with the training sample set, using the electric field fault data, magnetic field fault data, mechanical stress fault data, and vibration fault data in the effective fault data groups as the input data of the fault prediction model, and using the predicted fault category as the output data of the fault prediction model; taking the minimization of the error between the actual fault category and the predicted fault category as the training objective, taking the recall rate function as the loss function of the stability prediction model, and stopping the training to obtain the fault prediction model when the loss function converges.

[0021] Furthermore, the determination method of the fault subgraph includes:

[0022] According to the power operation data of each topological node in the topological subgraph, calculate the global representative data of each topological subgraph. The global representative data includes the mean value of the electric field data, the mean value of the magnetic field data, the mean value of the mechanical stress data, and the mean value of the vibration data. Use the global representative data as the input of the fault prediction model for prediction to obtain the predicted fault category, mark the topological subgraph corresponding to the global representative data as the predicted fault category, and record all the topological subgraphs with marks as fault subgraphs.

[0023] Further, the method for locating the fault points in the fault subgraph includes:

[0024] Regard each topological node in the fault subgraph as a particle, and each particle represents a possible fault position. Based on the number of topological nodes in the fault subgraph, initialize the number of particle swarms to half of the total number of topological nodes, initialize the number of particles in each particle swarm to num, initialize the initial velocity of each particle, and the initial velocity is a very small positive value to ensure a certain degree of randomness in the initial state. Define the fitness function formula of the topological node representing the fitness value of the th topological node, The calculation formula of which is: where μ, α, β, and γ represent weight coefficients, and the weight coefficients satisfy the weight condition μ + α + β + γ = 1. representing the corrected electric field data of the th topological node, representing the corrected magnetic field data of the th topological node, representing the corrected mechanical stress data of the th topological node, representing the corrected vibration data of the th topological node. Calculate the fitness values of all the connected topological nodes at the position of the particle based on the fitness value formula, and record the topological node with the largest fitness value among all the connected topological nodes as the updated node. Update the position of the particle based on the updated node. When the fitness value of the updated node is greater than the fitness value of the current position of the particle, update the position of the particle to the updated node. When the fitness value of the updated node is less than or equal to the fitness value of the current position of the particle, the particle stops moving. When the fitness values of all the positions of the particles no longer change, the topological node corresponding to the most concentrated position of the particles is the fault position.

[0025] The technical effects and advantages of the present invention:

[0026] The present invention trains based on the differences between historical measured and predicted data, and optimizes the obtained power correction model by minimizing the loss function, thereby improving the correction accuracy of power operation data. The graph decomposition process uses the ant colony algorithm, effectively reducing the computational complexity of complex power grids and facilitating efficient fault analysis of local subgraphs. The GRU model is used to predict fault categories, capturing the time dependence in historical fault data, and thus improving the accuracy of fault prediction. By performing anomaly detection and elimination on historical power fault data, the data used for model training is ensured to be more effective and accurate, improving the robustness and reliability of the fault prediction model. By simulating the swarm intelligence behavior in nature through the particle swarm algorithm, efficient search and positioning are carried out in the power grid topology subgraph, ensuring the accuracy of the final positioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 FIG. is a schematic diagram of a power grid intelligent transmission line fault location system of the present invention;

[0028] Figure 2 FIG. is a schematic diagram of a power grid intelligent transmission line fault location method of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0030] Embodiment 1

[0031] Please refer to Figure 1 As shown in the figure, a power grid intelligent transmission line fault location system in this embodiment includes: a data acquisition module: collecting power operation data, spatial data, and environmental data;

[0032] A data correction module: correcting the power operation data based on the constructed power correction model to obtain corrected power data;

[0033] A graph construction module: including a graph construction unit and a graph decomposition unit. The graph construction unit uses a neural network algorithm to perform topology establishment on the corrected power data and spatial data to obtain a power grid topology graph. The graph decomposition unit decomposes the power grid topology graph to obtain B topological subgraphs;

[0034] A fault model establishment module: collecting F groups of historical power fault data and constructing a fault prediction model for predicting fault types based on the historical power fault data;

[0035] Fault location module: Perform fault prediction on B topological subgraphs through a fault prediction model, mark the topological subgraphs with predicted fault types as faulty subgraphs, and locate the fault points in the faulty subgraphs through the particle swarm algorithm to determine the final fault positions;

[0036] Each module is connected by wired and / or wireless means to achieve data transmission between modules;

[0037] Power operation data includes: electric field data, magnetic field data, mechanical stress data, and vibration data; spatial data includes: accuracy, latitude, and proximity relationship; environmental data includes: temperature, wind force, and humidity; among them, electric field data is the field generated by charged bodies or currents, and its distribution changes with the line state and environmental conditions. Electric field data is obtained through electric field sensors, usually capacitive or inductive sensors, and the sensors are placed at a certain distance to avoid damage to the equipment caused by high voltage; magnetic field data is generated by currents, and the magnetic field intensity reflects the current situation in the line. Currents with different phases form different magnetic field distributions, and magnetic field data is obtained through Hall effect sensors; mechanical stress data refers to the internal force acting on an object, and mechanical stress is mainly caused by the self-weight of the wire, wind force, electromagnetic force, etc. Mechanical stress data is obtained through fiber Bragg grating sensors. Fiber Bragg grating sensors can measure stress using the fiber Bragg grating effect and have strong anti-interference ability and are suitable for harsh environments; vibration data is obtained through acceleration sensors, usually piezoelectric accelerometers, which can sense vibration frequency and amplitude; the proximity relationship is the connection relationship between each substation, which is a definite value, "yes" or "no", "yes" means the substations are connected, and "no" means the substations are not connected;

[0038] Collect A sets of historical power operation data and historical environmental data. The historical power operation data includes: historical measured power data and historical predicted power data. Based on the historical power operation data and historical environmental data, use the historical predicted power data and historical environmental data as the input of the power correction model, use the predicted corrected power data as the output of the power correction model, and use the historical measured power data as the prediction target of the power correction model for model training to construct a power correction model. The power correction model is a linear regression model, and the target loss function of the power correction model is the mean square error function. Take minimizing the value of the target loss function as the training goal to obtain a soil viscosity prediction model with the minimum value of the target loss function;

[0039] Based on the constructed power correction model, use the power operation data and environmental data as the power correction model for prediction to obtain corrected power data, which includes corrected electric field data, corrected magnetic field data, corrected mechanical stress data, and corrected vibration data;

[0040] The data categories included in the historical measured power data and the historical predicted power data are the same, both being historical electric field data, historical magnetic field data, historical mechanical stress data, and historical vibration data, and are the same as the data categories and types included in the power operation data; the data categories and types included in the historical environmental data and the environmental data are the same, and the historical environmental data includes historical temperature, historical wind force, and historical humidity;

[0041] The historical predicted power data are the data obtained under normal temperature, normal wind force, and normal humidity, and the historical measured power data are the data obtained under abnormal temperature, abnormal wind force, and abnormal humidity. Both the historical predicted power data and the historical measured power data are obtained through electric field sensors, Hall effect sensors, fiber Bragg grating sensors, and acceleration sensors respectively;

[0042] Construct power grid nodes by taking the accuracy and latitude in each corrected power data and spatial data as the attributes of a node, and use the proximity relationship in the spatial nodes as the basis for connecting between nodes to construct a power grid node matrix. The values in the power grid node matrix are determined based on the proximity relationship. When the proximity relationship is "yes", the value in the power grid node matrix is 1, and when the proximity relationship is "no", the value in the power grid node matrix is 1; construct the association index set for each node based on the power grid node matrix. The method for constructing the association index set for each node includes: based on the power grid node JD represented by each row in the power grid node matrix, the power grid nodes with a value of 1 in each row of the matrix are the associated nodes of the power grid node JD, and the indexes of all associated nodes constitute the association index set; take the attributes of each node as the input of the input layer of the graph convolutional network, and use the graph convolutional network to perform feature aggregation on each node. The formula for performing feature aggregation is:

[0043] Among them, represents the node embedding feature of the v-th power grid node after being convolved by the i-th convolutional layer. The node embedding feature is a low-dimensional vector obtained by mapping the power grid node through the convolutional layer, which is convenient for efficient calculation and analysis. N v is the association index set of the v-th power grid node, and d v represents the degree of the v-th power grid node (i.e., the number of edges connected to the v-th power grid node), and d u represents the degree of the u-th power grid node, and W l represents the weight matrix of the i-th convolutional layer, which is used to extract the local features of the input data, and b i represents the bias term of the i-th convolutional layer, which is used to adjust the output of the convolutional operation. σ represents the non-linear activation function, such as the ReLU function, represents the input feature of the i-th convolutional layer, and i represents the index of the convolutional layer;

[0044] Using the node embedding features as topological nodes and the proximity relationships in the spatial nodes as the basis for connections between nodes, calculating the similarity of connected topological nodes through the cosine similarity formula to obtain node similarity values, constructing a power grid topological graph with the node similarity values as the edge weights. The number of vertices in the power grid topological graph is the number of point embedding features, and the number of connected edges in the power grid topological graph is the number of values of 1 in the power grid node matrix;

[0045] Perform graph decomposition on the power grid topological graph. The ways to perform graph decomposition include: starting from each topological node in the power grid topological graph, using the breadth-first algorithm to search for the shortest paths between every pair of topological nodes in the power grid topological graph; based on all the shortest paths, initialize G ants, where the G value is the total number of shortest paths, assign one shortest path to each ant, and each ant represents the execution of one shortest path and is assigned to different topological nodes in the power grid topological graph; initialize pheromones: initialize the pheromone concentration for each edge and assign an initial pheromone concentration to each ant; each ant starts from the starting topological node and moves in the power grid topological graph based on the assigned shortest path, and update the pheromone concentration for all the edges in the shortest path according to the sum of the weights of all the edges in the shortest path corresponding to each ant. The formula for updating the pheromone concentration is: where, τ m,n represents the updated value of the pheromone concentration of the edge connecting the m-th node and the n-th node in the power grid topological graph, ρ represents the pheromone evaporation coefficient, S represents the initial pheromone concentration assigned to the ant, used to adjust the pheromone increment, and Sum represents the sum of the weights of all the edges passed by the selected path; when all the ants reach the end of the path, select the edge with the highest pheromone concentration in the power grid topological graph as the intermediate edge, delete the intermediate edge from the power grid topological graph to form a new power grid topological graph. When the power grid topological graph is divided into N non-connected topological subgraphs, perform module evaluation on each topological subgraph, N≥2; the formula for performing module evaluation is: where, Q j represents the modularity of the j-th topological subgraph, a j represents the total number of edges in the j-th topological subgraph, represents the degree of the c-th topological node in the j-th topological subgraph, represents the degree of the d-th topological node in the j-th topological subgraph, q represents the weight of the edge connecting the c-th topological node and the d-th topological node, SIM j represents all combinations of topological nodes with connected edges in the j-th topological subgraph, c and d represent the indices of topological nodes in the j-th topological subgraph, and c≠d; repeat until the value of the modularity of each topological subgraph no longer changes, and obtain B topological subgraphs;

[0046] Collect historical power failure data of group F. The historical power failure data includes fault categories, electric field fault data, magnetic field fault data, mechanical stress fault data, and vibration fault data. Based on the fault categories, group the historical power failure data with the same fault category into one group to obtain R groups of fault data groups. Perform anomaly detection on the data in each group of fault data groups. The methods for anomaly detection include: forming a C×1 fault vector from the electric field fault data, magnetic field fault data, mechanical stress fault data, and vibration fault data in each group of data in the fault data group, where C represents the total number of fault data categories; calculating the similarity values between each fault vector in the fault vector and other fault vectors through a similarity calculation formula, performing an average calculation on the calculated similarity values to obtain the average similarity of each fault vector, calculating the dispersion of each fault vector through a zero score algorithm based on the average similarity, presetting a dispersion threshold, marking the power failure data with a dispersion greater than or equal to the dispersion threshold as abnormal data, removing the historical power failure data marked as abnormal data in each group of fault data groups to obtain effective fault data groups; using the effective fault data groups as the training sample set, taking the GRU model as the initial model, training the GRU model using the training sample set, using the electric field fault data, magnetic field fault data, mechanical stress fault data, and vibration fault data in the effective fault data groups as the input data of the fault prediction model, and taking the predicted fault category as the output data of the fault prediction model; taking minimizing the error between the actual fault category and the predicted fault category as the training objective, taking the recall rate function as the loss function of the stability prediction model, and stopping the training to obtain the fault prediction model when the loss function converges;

[0047] The fault categories are the currently known fault situations in the power grid transmission line. For example: Short - circuit fault: refers to the conductive connection between phases or between a phase and the ground, resulting in a rapid increase in current. Common short - circuit faults include: single - phase grounding short - circuit (single - phase - to - ground fault), two - phase short - circuit (two - phase short - circuit), two - phase grounding short - circuit, three - phase short - circuit, and three - phase grounding short - circuit; Open - circuit fault: refers to the disconnection of the wire in the transmission line, usually caused by mechanical damage or equipment aging, resulting in abnormal current flow; Grounding fault: refers to the accidental contact between the wire and the ground, usually accompanied by high - current leakage to the ground, which may cause a fire or equipment damage in severe cases;

[0048] According to the power operation data of each topological node in the topological sub - graph, calculate the global representative data of each topological sub - graph. The global representative data includes the mean of the electric field data, the mean of the magnetic field data, the mean of the mechanical stress data, and the mean of the vibration data. Use the global representative data as the input of the fault prediction model for prediction to obtain the predicted fault category, and mark the topological sub - graph corresponding to the global representative data as the predicted fault category. Record all the topological sub - graphs with marks as fault sub - graphs;

[0049] Regarding each topological node in the fault sub - graph as a particle, where each particle represents a possible fault location, based on the number of topological nodes in the fault sub - graph, the number of particle swarms is initialized to half of the total number of topological nodes. The number of particles in each particle swarm is initialized as num, and the initial velocity of each particle is initialized. The initial velocity is a very small positive value to ensure a certain degree of randomness in the initial state. Define the fitness function formula of the topological node representing the fitness value of the th topological node, The calculation formula of is: where μ, α, β, and γ represent weight coefficients, and the weight coefficients satisfy the weight condition μ + α+β + γ = 1, representing the corrected electric field data of the th topological node, representing the corrected magnetic field data of the th topological node, representing the corrected mechanical stress data of the th topological node, representing the corrected vibration data of the th topological node. Based on the fitness value formula, calculate the fitness values of all connected topological nodes at the position of the particle. Denote the topological node with the maximum fitness value among all connected topological nodes as the updated node. Update the position of the particle based on the updated node. When the fitness value of the updated node is greater than the fitness value of the current position of the particle, update the position of the particle to the updated node. When the fitness value of the updated node is less than or equal to the fitness value of the current position of the particle, the particle stops moving. When the fitness values at the positions of all particles no longer change, the topological node corresponding to the position where the particles are most concentrated is the fault location; the number of particles in the particle swarm is determined by multiplying the data dimension of the topological nodes in the fault sub - graph by a particle constant. The particle constant is a positive integer, and the value range of the particle constant is [1, 5];

[0050] In this embodiment, by training based on the difference between historical measured and predicted data and optimizing the obtained power correction model by minimizing the loss function, the correction accuracy of power operation data is improved; the graph decomposition process uses the ant colony algorithm, effectively reducing the computational complexity of complex power grids and facilitating efficient fault analysis of local sub - graphs; predicting the fault category through the GRU model, capturing the time dependence in historical fault data, and thus improving the accuracy of fault prediction; by performing anomaly detection and elimination on historical power fault data, ensuring that the data used for model training is more effective and accurate, and improving the robustness and reliability of the fault prediction model; by simulating the swarm intelligence behavior in nature through the particle swarm algorithm, performing efficient search and positioning in the power grid topological sub - graph, and ensuring the accuracy of the final positioning.

[0051] Example 2

[0052] Please refer to Figure 2 as shown. For the parts not described in detail in this embodiment, refer to the description in Embodiment 1. A method for fault location of an intelligent power transmission line of a power grid is provided, including:

[0053] S1. Collect power operation data, spatial data, and environmental data;

[0054] S2. Correct the power operation data based on the constructed power correction model to obtain corrected power data;

[0055] S3. Use the neural network algorithm to establish a topology for the corrected power data and spatial data to obtain a power grid topology map, and decompose the power grid topology map to obtain B topological subgraphs;

[0056] S4. Collect F groups of historical power fault data, and construct a fault prediction model for predicting fault types based on the historical power fault data;

[0057] S5. Perform fault prediction on the B topological subgraphs through the fault prediction model, mark the topological subgraphs with predicted fault types as fault subgraphs, and locate the fault points in the fault subgraphs through the particle swarm algorithm to determine the final fault positions.

[0058] Example 3

[0059] This embodiment publicly provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the above-provided fault location system for an intelligent power transmission line of a power grid.

[0060] Since the electronic device introduced in this embodiment is the electronic device used to implement a fault location system for an intelligent power transmission line of a power grid in an embodiment of the present application, based on the fault location system for an intelligent power transmission line of a power grid introduced in an embodiment of the present application, those skilled in the art can understand the specific implementation manners and various variations of the electronic device in this embodiment. Therefore, the specific implementation of how this electronic device implements the method in an embodiment of the present application will not be described in detail here. As long as those skilled in the art implement the electronic device used in an embodiment of the present application for a fault location system for an intelligent power transmission line of a power grid, it falls within the scope of protection of the present application.

[0061] The above formulas are all dimensionless and take their numerical values for calculation. The formulas are obtained by software simulation of a large amount of collected data to obtain a formula closest to the actual situation. The preset parameters and threshold selection in the formulas are set by those skilled in the art according to the actual situation.

[0062] The above are only the preferred embodiments of the present invention. The protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the concept of the present invention belong to the protection scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements should also be regarded as within the protection scope of the present invention.

Claims

1. An intelligent power grid transmission line fault location system, characterized in that, Including: Data acquisition module: Acquire power operation data, spatial data, and environmental data; Data correction module: Correct the power operation data based on the constructed power correction model to obtain corrected power data; Graph construction module: It includes a graph construction unit and a graph decomposition unit. The graph construction unit uses a neural network algorithm to establish a topology for the corrected power data and spatial data to obtain a power grid topology graph. The graph decomposition unit decomposes the power grid topology graph to obtain topological subgraphs; Fault model establishment module: Acquire F groups of historical power fault data and construct a fault prediction model for predicting fault types based on the historical power fault data; Fault location module: Through the fault prediction model, fault prediction is carried out on topological subgraphs, the topological subgraphs with predicted fault types are marked as faulty subgraphs, and the fault points in the faulty subgraphs are located by the particle swarm optimization algorithm to determine the final fault positions; The determination method of the fault subgraph includes: According to the power operation data of each topological node in the topological subgraph, calculate the global representative data of each topological subgraph. The global representative data includes the mean of the electric field data, the mean of the magnetic field data, the mean of the mechanical stress data, and the mean of the vibration data. Use the global representative data as the input of the fault prediction model for prediction to obtain the predicted fault category, and mark the topological subgraph corresponding to the global representative data as the predicted fault category. Denote all the topological subgraphs with marks as fault subgraphs; The method for locating the fault points in the fault subgraph includes: Regard each topological node in the fault subgraph as a particle, where each particle represents a possible fault location. Based on the number of topological nodes in the fault subgraph, initialize the number of particle swarms to half of the total number of topological nodes, and initialize the number of particles in each particle swarm to , initialize the initial velocity of each particle, where the initial velocity is a very small positive value, and define the fitness function formula for the topological node , represents the fitness value of the th topological node, and the calculation formula of is: ; where , , and represent weight coefficients, and the weight coefficients satisfy the weight condition: , represents the corrected electric field data of the th topological node, represents the corrected magnetic field data of the th topological node, represents the corrected mechanical stress data of the th topological node, represents the corrected vibration data of the th topological node. Calculate the fitness values of all connected topological nodes at the position of the particle based on the fitness value formula, and record the topological node with the largest fitness value among all connected topological nodes as the updated node. Update the position of the particle based on the updated node. When the fitness value of the updated node is greater than the fitness value of the current position of the particle, update the position of the particle to the updated node. When the fitness value of the updated node is less than or equal to the fitness value of the current position of the particle, the particle stops moving. When the fitness values at the positions of all particles no longer change, the topological node corresponding to the position where the particles are most concentrated is the fault location.

2. The intelligent power grid transmission line fault location system according to claim 1, characterized in that, The power operation data includes: electric field data, magnetic field data, mechanical stress data, and vibration data; the spatial data includes: accuracy, latitude, and proximity relationship; the environmental data includes: temperature, wind force, and humidity.

3. The intelligent power grid transmission line fault location system according to claim 2, wherein The acquisition method of the corrected power data includes: Based on the constructed power correction model, use the power operation data and environmental data as the input of the power correction model for prediction to obtain the corrected power data. The corrected power data includes corrected electric field data, corrected magnetic field data, corrected mechanical stress data, and corrected vibration data.

4. An intelligent power grid transmission line fault location system according to claim 3, characterized in that, The construction method of the power grid topology map includes: Construct power grid nodes by taking the precision and latitude in each corrected power data and spatial data as attributes of a node. Use the proximity relationship in the spatial nodes as the basis for connecting nodes, and construct a power grid node matrix. The values in the power grid node matrix are determined based on the proximity relationship. When the proximity relationship is "yes", the value in the power grid node matrix is 1; when the proximity relationship is "no", the value in the power grid node matrix is 1. Construct the associated index set for each node based on the power grid node matrix. The method of constructing the associated index set for each node includes: based on the power grid nodes represented by each row in the power grid node matrix , and the power grid nodes with a value of 1 in each row of the matrix are the associated nodes of the power grid node . The indices of all associated nodes form the associated index set. Use the attributes of each node as the input to the input layer of the graph convolutional network. Employ the graph convolutional network to aggregate features for each node to obtain node embedding features. Use the node embedding features as topological nodes and the proximity relationship in the spatial nodes as the basis for connecting nodes. Calculate the similarity of the connected topological nodes through the cosine similarity formula to obtain node similarity values. Construct a power grid topology graph with the node similarity values as the weights of the edges.

5. The intelligent power grid transmission line fault location system according to claim 4, characterized in that, The formula for feature aggregation is as follows: ; where represents the node embedding feature of the -th power grid node after convolution by the -th convolutional layer, is the association index set of the -th power grid node, represents the degree of the -th power grid node, represents the degree of the -th power grid node, represents the weight matrix of the -th convolutional layer, represents the bias term of the -th convolutional layer, represents the non-linear activation function.

6. The intelligent power grid transmission line fault location system according to claim 5, characterized in that, The method for decomposing the power grid topology map includes: Starting from each topological node in the power grid topology diagram, use the breadth-first algorithm to search for the shortest paths between every pair of topological nodes in the power grid topology diagram; based on all the shortest paths, initialize G ants, where the value of G is the total number of shortest paths. Assign one shortest path to each ant, and each ant represents the execution of one shortest path and is assigned to different topological nodes in the power grid topology diagram; initialize pheromones: initialize the pheromone concentration for each edge and assign an initial pheromone concentration to each ant; each ant starts from the starting topological node and moves in the power grid topology diagram based on the assigned shortest path as the basis for each ant's movement. Update the pheromone concentration for all the edges in the shortest path according to the sum of the weights of all the edges in the shortest path corresponding to each ant. The formula for updating the pheromone concentration is: ; where represents the updated value of the pheromone concentration of the edge connecting the th node and the th node in the power grid topology diagram, represents the pheromone evaporation coefficient, represents the initial pheromone concentration assigned to the ant, which is used to adjust the pheromone increment, represents the sum of the weights of all the edges passed by the selected path; when all the ants reach the end point of the path, select the edge with the highest pheromone concentration as the intermediate edge according to the pheromone concentration of each edge in the power grid topology diagram, and delete the intermediate edge from the power grid topology diagram to form a new power grid topology diagram. When the power grid topology diagram is divided into disconnected topological subgraphs, perform module evaluation on each topological subgraph to obtain the modularity of each topological subgraph, and repeat until the value of the modularity of each topological subgraph no longer changes, and obtain topological subgraphs.

7. The intelligent power grid transmission line fault location system according to claim 6, wherein The formula for module evaluation is as follows: ; where represents the modularity of the th topological subgraph, represents the total number of edges in the th topological subgraph, represents the degree of the th topological node in the th topological subgraph, represents the degree of the th topological node in the th topological subgraph, represents the weight of the edge connecting the th topological node and the th topological node, represents all combinations of topological nodes with connected edges in the th topological subgraph, and represent the indices of the topological nodes in the th topological subgraph, and .

8. The intelligent power grid transmission line fault location system according to claim 7, characterized in that, The construction method of the fault prediction model includes: Collect historical power failure data in group F. The historical power failure data includes fault categories, electric field fault data, magnetic field fault data, mechanical stress fault data, and vibration fault data. Based on the fault categories, the historical power failure data with the same fault category is divided into one group to obtain R groups of fault data groups. Perform anomaly detection on the data in each group of fault data groups. The methods for anomaly detection include: forming an fault vector with the electric field fault data, magnetic field fault data, mechanical stress fault data, and vibration fault data in each group of data in the fault data group, where represents the total number of fault data categories; calculate the similarity values between each fault vector in the fault vector and other fault vectors through the similarity calculation formula, perform average calculation on the calculated similarity values to obtain the average similarity of each fault vector, calculate the dispersion of each fault vector through the zero score algorithm based on the average similarity, preset the dispersion threshold, mark the power failure data with a dispersion greater than or equal to the dispersion threshold as abnormal data, and remove the historical power failure data marked as abnormal data in each group of fault data groups to obtain effective fault data groups; use the effective fault data groups as the training sample set, use the GRU model as the initial model, train the GRU model with the training sample set, use the electric field fault data, magnetic field fault data, mechanical stress fault data, and vibration fault data in the effective fault data groups as the input data of the fault prediction model, and use the predicted fault category as the output data of the fault prediction model; use minimizing the error between the actual fault category and the predicted fault category as the training objective, use the recall rate function as the loss function of the stability prediction model, and stop training to obtain the fault prediction model when the loss function converges.

Images