Fault type identification model construction method, fault type identification method and system

Abstract

The invention provides a construction method of a fault type identification model and a fault type identification method and system.The construction method comprises the steps that traveling wave data collection is conducted on a plurality of power transmission lines with different faults in a power transmission network, and a fault traveling wave data set is obtained; analyzing the fault traveling wave data set by adopting a power grid topology connection relation table of the power transmission network to obtain a space-time dynamic fault graph; training a space-time diagram neural network by adopting the space-time dynamic fault diagram and the fault type label corresponding to each node in the space-time dynamic fault diagram to obtain an initial recognition model; and under a constraint condition corresponding to a physical law of a circuit system, optimizing model parameters of the initial identification model based on a swarm intelligent search mechanism of a wolf pack algorithm to obtain a fault type identification model of the power transmission line. The recognition accuracy and efficiency of the fault type recognition model of the power transmission line can be improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence optimization algorithm and power system fault identification, and particularly relates to a construction method of a fault type identification model, a fault type identification method and system. BACKGROUND

[0002] With the continuous development of society, the demand for power supply is expanding, and the total length of transmission lines is expanding. This trend makes the stable operation of transmission lines have more and more obvious influence on the power grid. In order to ensure power supply, it is necessary to accurately, quickly and effectively diagnose and identify the type of line fault when overhead transmission line fault occurs. However, manual inspection has the problems of low accuracy and low inspection efficiency. In order to solve this problem, an automatic diagnosis method for overhead transmission line fault needs to be developed.

[0003] In the related art, due to the diversity and complexity of the fault type of the transmission line, the fault type identification method of the transmission line usually has the defects of difficulty in collecting traveling wave data and low accuracy of the algorithm model. Therefore, how to overcome the defects of the existing technology to improve the accuracy and efficiency of the fault type identification of the transmission line becomes a problem to be solved at present. SUMMARY

[0004] In order to solve the problems of the prior art, the present application provides a construction method of a fault type identification model, a fault type identification method and system, which aims to improve the accuracy and efficiency of the fault type identification model of the transmission line.

[0005] The purpose of the present application is achieved by using the following technical solutions:

[0006] On the one hand, the present application provides a construction method of a fault type identification model of a transmission line, which comprises:

[0007] Collecting traveling wave data of a plurality of transmission lines with different faults in the power grid to obtain a fault traveling wave data set;

[0008] Analyzing the fault traveling wave data set by using a power grid topology connection relationship table of the power grid to obtain a time-space dynamic fault graph;

[0009] Training a time-space graph neural network by using the time-space dynamic fault graph and a fault type label corresponding to each node in the time-space dynamic fault graph to obtain an initial identification model;

[0010] Under the constraint condition corresponding to the physical law of the circuit system, optimizing the model parameters of the initial identification model based on the group intelligence search mechanism of the wolf swarm algorithm to obtain a fault type identification model of the transmission line.

[0011] Optionally, the power grid topology connection relationship table of the power transmission network is adopted to analyze the fault traveling wave data set, and a time-space dynamic fault graph is obtained, including:

[0012] Based on each fault traveling wave data in the fault traveling wave data set, a sliding window algorithm is used for time-domain feature extraction at multiple moments, and dynamic feature data of the fault traveling wave data is obtained.

[0013] According to the power grid topology connection relationship table, a time-space dynamic fault graph is established, taking the monitoring points of the multiple power transmission lines as multiple nodes, taking the dynamic feature data of the fault traveling wave data as the node features of the corresponding nodes, and taking the line impedance connection parameters corresponding to the multiple power transmission lines as the edge weights.

[0014] Optionally, the time-space dynamic fault graph and the fault type label corresponding to each node in the time-space dynamic fault graph are used to train a time-space graph neural network, and an initial identification model is obtained, including:

[0015] Based on each node in the time-space dynamic fault graph, the node feature of the node is input into the time-space graph neural network for fault type prediction, and a predicted fault type corresponding to the node is obtained.

[0016] Based on each node, a multi-task loss function is used to determine the loss between the predicted fault type corresponding to the node and the fault type label corresponding to the node, and a loss value of the node is obtained.

[0017] According to the loss of the node in the time-space dynamic fault graph, the model parameters of the time-space graph neural network are adjusted until the output loss meets the convergence condition, and the initial identification model is obtained.

[0018] Optionally, the multi-task loss function can be represented as:

[0019]

[0020] wherein, L cls is the classification loss of the i-th node in the time-space dynamic fault graph; N is the number of nodes in the time-space dynamic fault graph; C is the number of fault types corresponding to the time-space dynamic fault graph; y i,c is the fault type label c corresponding to the i-th node in the time-space dynamic fault graph; p i,c is the probability that the i-th node in the time-space dynamic fault graph belongs to the fault category c; L graph is the time-space consistency loss of the i-th node in the time-space dynamic fault graph; ε is the edge set corresponding to the power grid topology connection relationship table of the multiple power transmission lines; h i is the hidden feature of the i-th node in the time-space dynamic fault graph; hj is a hidden feature of the jth node in the spatiotemporal dynamic fault graph; ω ij is an edge weight of a power transmission line between the ith node and the jth node in the spatiotemporal dynamic fault graph; L total is a loss value of the ith node in the spatiotemporal dynamic fault graph; α and β are both hyperparameters of the spatiotemporal graph neural network.

[0021] Optionally, under the constraint condition corresponding to the physical law of the circuit system, the model parameters of the initial identification model are optimized based on a group intelligence search mechanism of a wolf swarm algorithm to obtain the fault type identification model of the power transmission line, including:

[0022] An initial value range of each model parameter of the initial identification model under the constraint condition corresponding to the physical law of the circuit system is obtained.

[0023] Based on each model parameter, the initial value range of each model parameter is searched and optimized based on the group intelligence search mechanism of the wolf swarm algorithm to obtain an optimal value of the model parameter.

[0024] The optimal values of the plurality of model parameters are combined to obtain an optimal parameter combination of the initial identification model.

[0025] The initial identification model is trained using the optimal parameter combination to obtain the fault type identification model.

[0026] Optionally, the constraint corresponding to the physical law of the circuit system includes a line impedance constraint and an electromagnetic transient characteristic constraint; and the initial value range of each model parameter of the initial identification model under the constraint corresponding to the physical law of the circuit system is obtained, including:

[0027] Based on each model parameter, a value range of the model parameter under the line impedance constraint and a value range of the model parameter under the electromagnetic transient characteristic constraint are respectively obtained.

[0028] Based on each model parameter, an intersection between the value range of the model parameter under the line impedance constraint and the value range of the model parameter under the electromagnetic transient characteristic constraint is taken as an initial value range of the model parameter.

[0029] Correspondingly, an embodiment of the present application provides a system for constructing a fault type identification model of a power transmission line, including:

[0030] A data acquisition module is configured to acquire traveling wave data of a plurality of power transmission lines with different faults in a power transmission network to obtain a fault traveling wave data set.

[0031] an analysis module configured to analyze the fault traveling wave data set by using a power grid topology connection relationship table of the power transmission grid to obtain a time-space dynamic fault graph;

[0032] a training module configured to train a time-space graph neural network by using the time-space dynamic fault graph and a fault type label corresponding to each node in the time-space dynamic fault graph to obtain an initial identification model;

[0033] an optimization module configured to optimize model parameters of the initial identification model based on a swarm intelligence search mechanism of a wolf swarm algorithm under a constraint condition corresponding to a physical law of a circuit system to obtain a fault type identification model of the power transmission line.

[0034] Optionally, the analysis module comprises:

[0035] a feature extraction unit configured to perform multi-time time-domain feature extraction on each fault traveling wave data in the fault traveling wave data set by using a sliding window algorithm to obtain dynamic feature data of the fault traveling wave data;

[0036] a construction analysis unit configured to establish, according to the power grid topology connection relationship table, a time-space dynamic fault graph with monitoring points of the plurality of power transmission lines as a plurality of nodes, with the dynamic feature data of the fault traveling wave data as node features of the corresponding nodes, and with line impedance connection parameters of the plurality of power transmission lines as edge weights.

[0037] Optionally, the training module comprises:

[0038] a fault type prediction unit configured to input node features of each node in the time-space dynamic fault graph into the time-space graph neural network to perform fault type prediction to obtain a predicted fault type corresponding to the node;

[0039] a loss determination unit configured to determine, based on each node, a loss between the predicted fault type corresponding to the node and a fault type label corresponding to the node by using a multi-task loss function to obtain a loss value of the node;

[0040] an adjustment unit configured to adjust model parameters of the time-space graph neural network according to the loss of the node in the time-space dynamic fault graph until an output loss meets a convergence condition to obtain the initial identification model.

[0041] Optionally, the multi-task loss function can be represented as:

[0042]

[0043] wherein, L clsis the classification loss of the ith node in the spatiotemporal dynamic fault graph; N is the number of nodes in the spatiotemporal dynamic fault graph; C is the number of fault types corresponding to the spatiotemporal dynamic fault graph; y i,c is the fault type label c corresponding to the ith node in the spatiotemporal dynamic fault graph; p i,c is the probability that the ith node in the spatiotemporal dynamic fault graph belongs to the fault category c; L graph is the spatiotemporal consistency loss of the ith node in the spatiotemporal dynamic fault graph; ε is an edge set corresponding to a power grid topology connection relationship table of the plurality of power transmission lines; h i is the hidden feature of the ith node in the spatiotemporal dynamic fault graph; h j is the hidden feature of the jth node in the spatiotemporal dynamic fault graph; ω ij is the edge weight of the power transmission line between the ith node and the jth node in the spatiotemporal dynamic fault graph; L total is the loss value of the ith node in the spatiotemporal dynamic fault graph; α and β are both hyperparameters of the spatiotemporal graph neural network.

[0044] Optionally, the optimization module comprises:

[0045] An acquisition unit is configured to acquire an initial value range of each model parameter of the initial identification model under a constraint condition corresponding to a physical law of the circuit system;

[0046] A search unit is configured to search and optimize the initial value range of each model parameter based on the initial value range of each model parameter by using a swarm intelligence search mechanism of the wolf pack algorithm to obtain an optimal value of the model parameter.

[0047] A combination unit is configured to combine the optimal values of the plurality of model parameters to obtain an optimal parameter combination of the initial identification model.

[0048] A training unit is configured to train the initial identification model by using the optimal parameter combination to obtain the fault type identification model.

[0049] Optionally, the constraint corresponding to the physical law of the circuit system comprises a line impedance constraint and an electromagnetic transient characteristic constraint; the acquisition unit is specifically configured to acquire, based on each model parameter, a value range of the model parameter under the line impedance constraint and a value range of the model parameter under the electromagnetic transient characteristic constraint; and acquire, based on each model parameter, an intersection between the value range of the model parameter under the line impedance constraint and the value range of the model parameter under the electromagnetic transient characteristic constraint as the initial value range of the model parameter.

[0050] In another aspect, the embodiment of the present application also provides a fault type identification method of a power transmission line, the method comprising:

[0051] Collecting multi-time wave propagation data of the power transmission line currently having a fault in the power transmission network to obtain real-time fault wave propagation data;

[0052] Analyzing the real-time fault wave propagation data by using a power grid topology connection relationship table of the power transmission network to obtain a real-time dynamic fault graph;

[0053] Inputting the real-time dynamic fault graph into the fault type identification model of the power transmission line constructed by the method to identify the fault type, and obtaining the fault type of the power transmission line currently having a fault.

[0054] Correspondingly, the embodiment of the present application also provides a fault type identification system of a power transmission line, the system comprising:

[0055] A real-time collection module configured to collect multi-time wave propagation data of the power transmission line currently having a fault in the power transmission network to obtain real-time fault wave propagation data;

[0056] A real-time analysis module configured to analyze the real-time fault wave propagation data by using a power grid topology connection relationship table of the power transmission network to obtain a real-time dynamic fault graph;

[0057] A fault identification module configured to input the real-time dynamic fault graph into the fault type identification model of the power transmission line constructed by the method to identify the fault type, and obtain the fault type of the power transmission line currently having a fault.

[0058] In still another aspect, the embodiment of the present application also provides an electronic device comprising at least one processor and a memory;

[0059] The memory is configured to store one or more programs;

[0060] When the one or more programs are executed by the at least one processor, the construction method of the fault type identification model of the power transmission line or the construction method of the fault type identification model of the power transmission line is implemented.

[0061] Correspondingly, the embodiment of the present application also provides a readable storage medium having an execution program stored thereon, and the execution program, when executed, implements the construction method of the fault type identification model of the power transmission line or the construction method of the fault type identification model of the power transmission line.

[0062] Compared with the prior art, the embodiment of the present application has the following beneficial effects:

[0063] The application provides a method and system for constructing a fault type identification model of a power transmission line. In the method, first, wave propagation data of different faults of multiple power transmission lines in a power transmission network is collected to obtain a fault wave propagation data set; second, the fault wave propagation data set is analyzed by using a power grid topology connection relationship table of the power transmission network to obtain a time-space dynamic fault graph; third, the time-space graph neural network is trained by using the time-space dynamic fault graph and a fault type label corresponding to each node in the time-space dynamic fault graph to obtain an initial identification model; and finally, the model parameters of the initial identification model are optimized based on a group intelligence search mechanism of a wolf swarm algorithm under a constraint condition corresponding to a law of circuit system physics to obtain the fault type identification model of the power transmission line. In this way, on the one hand, the initial identification model is constructed by using the time-space dynamic fault graph and the time-space graph neural network, so that the initial identification model obtained can capture the time-space correlation of the fault signal, thereby improving the identification accuracy of the initial identification model; on the other hand, the constraint condition corresponding to the law of circuit system physics and the group intelligence search mechanism of the wolf swarm algorithm can avoid the model learning from violating the false features of the physical law, and the leader-follower cooperative search can avoid the gradient descent of the model from falling into a local optimum, so that the generalization ability of the obtained model is obviously better than that of the traditional grid search, thereby improving the accuracy and efficiency of the fault type identification model of the power transmission line.

[0064] The application provides a method and system for constructing a fault type identification model of a power transmission line. In the method, first, wave propagation data of different faults of multiple power transmission lines in a power transmission network is collected to obtain a fault wave propagation data set; second, the fault wave propagation data set is analyzed by using a power grid topology connection relationship table of the power transmission network to obtain a time-space dynamic fault graph; third, the time-space graph neural network is trained by using the time-space dynamic fault graph and a fault type label corresponding to each node in the time-space dynamic fault graph to obtain an initial identification model; and finally, the model parameters of the initial identification model are optimized based on a group intelligence search mechanism of a wolf swarm algorithm under a constraint condition corresponding to a law of circuit system physics to obtain the fault type identification model of the power transmission line. In this way, on the one hand, the initial identification model is constructed by using the time-space dynamic fault graph and the time-space graph neural network, so that the initial identification model obtained can capture the time-space correlation of the fault signal, thereby improving the identification accuracy of the initial identification model; on the other hand, the constraint condition corresponding to the law of circuit system physics and the group intelligence search mechanism of the wolf swarm algorithm can avoid the model learning from violating the false features of the physical law, and the leader-follower cooperative search can avoid the gradient descent of the model from falling into a local optimum, so that the generalization ability of the obtained model is obviously better than that of the traditional grid search, thereby improving the accuracy and efficiency of the fault type identification model of the power transmission line.

[0065] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, but not limiting the technical solutions provided by the embodiments of the application. BRIEF DESCRIPTION OF DRAWINGS

[0066] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein:

[0067] Figure 1 A flowchart illustrating a method for constructing a fault type identification model for transmission lines, provided in an embodiment of the present invention;

[0068] Figure 2 This is a schematic diagram of a process for acquiring multiple current traveling wave signals and multiple voltage traveling wave signals according to an embodiment of the present invention;

[0069] Figure 3 A flowchart illustrating a method for identifying fault types in power transmission lines according to an embodiment of the present invention;

[0070] Figure 4 A schematic diagram illustrating the composition of a system for constructing a fault type identification model for transmission lines, as provided in an embodiment of the present invention.

[0071] Figure 5 A schematic diagram illustrating the composition of a fault type identification system for power transmission lines provided in an embodiment of the present invention;

[0072] Figure 6 A schematic diagram illustrating the composition of an electronic device provided in an embodiment of the present invention; Detailed Implementation

[0073] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0074] In the following description, references are made to “some embodiments,” which describe a subset of all possible embodiments. However, it is understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0075] In the following description, the terms "first, second, third" are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that "first, second, third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of the invention described herein can be implemented in an order other than that illustrated or described herein.

[0076] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which embodiments of the invention pertain. The terminology used herein is for descriptive purposes only and is not intended to limit the scope of the invention.

[0077] Example 1:

[0078] See Figure 1 The diagram shown is a flowchart illustrating a method for constructing a fault type identification model for transmission lines according to an embodiment of the present invention. The following is a summary of the process. Figure 1 The steps shown are explained below:

[0079] Step 101: Collect traveling wave data from multiple transmission lines experiencing different faults within the power grid to obtain a fault traveling wave dataset.

[0080] In some embodiments of the present invention, the power transmission network mainly undertakes the long-distance transmission and distribution of electricity; wherein, the positional relationship between multiple transmission lines that have different faults in the power transmission network can be determined according to actual needs, and the faults that occur on multiple transmission lines can be partially the same or completely different.

[0081] For example, a short-circuit fault occurs on transmission line 1 within the power transmission network, a line break fault occurs on transmission line 2, and a ground fault occurs on transmission line 3. The time corresponding to different faults occurring on multiple transmission lines within the power transmission network can be different.

[0082] In some embodiments of the present invention, traveling wave data can be collected at their respective monitoring points for multiple transmission lines that have experienced different faults in the power grid, thereby obtaining a fault traveling wave dataset; wherein, each fault traveling wave data in the fault traveling wave dataset is the traveling wave data of a faulty transmission line.

[0083] Here, specially customized wideband current and voltage transformers can be used to collect data on each faulty transmission line. First, weak signals are reinforced and isolated using the principle of electromagnetic induction to ensure electrical purity along the signal transmission path, protecting the back-end precision processing circuits from strong electrical shocks and stray electromagnetic interference. Second, low-noise, high-bandwidth operational amplifiers are selected to form bandpass filters. Their extremely low input noise voltage ensures that even weak traveling wave initiation signals at the millivolt or even microvolt level can be accurately amplified. At the same time, the bandpass filter and signal differential circuit design prevent interference information from obscuring signal characteristics. Finally, the bandwidth of up to several megahertz can completely preserve the high-frequency components in the traveling wave signal, allowing subsequent analysis to obtain rich signal details, accurately reconstruct the original waveform of the traveling wave, and provide a solid data starting point for subsequent fault analysis.

[0084] It should be noted that each fault traveling wave data in the fault traveling wave data set provided in the embodiments of the present invention can be: three-phase and zero-sequence current and voltage waveforms, that is, including: four current waveform data composed of IA, IB, IC, and I0, and four voltage waveform data composed of UA, UB, UC, and U0.

[0085] refer to Figure 2 As shown, for the faulty transmission line 201, a broadband instrument transformer group 202 deployed on it can be used, which can be divided into current transformers and voltage transformers. Thus, the traveling wave data of the faulty transmission line 201 can be acquired using the split current transformers and voltage transformers respectively, resulting in the following signals shown in 203: current traveling wave signals (IA, IB, IC, I0) and voltage traveling wave signals (UA, UB, UC, U0), where I0 is the zero-sequence current and U0 is the zero-sequence voltage; the frequency range of the current and voltage traveling wave signals is 1Hz-10MHz, and the duration is in the microsecond range. Furthermore, the current traveling wave signals (IA, IB, IC, I0) and voltage traveling wave signals (UA, UB, UC, U0) acquired by 203 can be processed by the current channel processing and voltage channel processing shown in 204, which are composed of bandpass filtering and differential circuits respectively, to obtain the corresponding fault traveling wave data (i.e., the fault traveling wave data of the transmission line that has experienced a fault).

[0086] Step 102: Using the power grid topology connection table of the transmission network, analyze the fault traveling wave dataset to obtain a spatiotemporal dynamic fault diagram.

[0087] In some embodiments of the present invention, the collected fault traveling wave dataset can be analyzed using the power grid topology connection table of the power transmission network, and the spatiotemporal dynamic fault map corresponding to the power transmission network can be constructed using the analysis results.

[0088] It should be noted that the power grid topology connection table is a data structure that describes the connection relationships between various transmission lines within the power grid.

[0089] In some embodiments of the present invention, step 102 can be implemented by the following steps 1021 and 1022 (not shown in the figure):

[0090] Step 1021: Based on each fault traveling wave data in the fault traveling wave dataset, use the sliding window algorithm to extract time-domain features at multiple times to obtain the dynamic feature data of the fault traveling wave data.

[0091] In some embodiments of the present invention, for each fault traveling wave data in the fault traveling wave dataset, a sliding window algorithm can be used to extract time-domain features at multiple time points (e.g., 10s or 30s) to obtain the dynamic feature data of the corresponding fault traveling wave data; wherein, the dynamic feature data can refer to the traveling wave signal attribute information of the fault traveling wave data in real time changes, including but not limited to: the propagation dynamic characteristics of the fault traveling wave data, signal morphology characteristics, etc.

[0092] It should be noted that the sliding window algorithm is a common technique for handling array or string problems. It efficiently solves problems related to subarrays or substrings by maintaining a dynamic window.

[0093] Step 1022: Based on the power grid topology connection table, establish a spatiotemporal dynamic fault diagram with the monitoring points corresponding to each of the multiple transmission lines as multiple nodes, the dynamic feature data of the fault traveling wave data as the node features of the corresponding nodes, and the line impedance connection parameters corresponding to the multiple transmission lines as edge weights.

[0094] In some embodiments of the present invention, after obtaining the dynamic feature data of each fault traveling wave data (here, each fault traveling wave data can correspond to a transmission line that has caused a fault), a spatiotemporal dynamic fault map corresponding to the power grid can be constructed by taking the monitoring points corresponding to multiple transmission lines as multiple nodes, taking the dynamic feature data of the fault traveling wave data corresponding to the transmission line as the node feature of the monitoring point (node) corresponding to the transmission line, and taking the line impedance connection parameters corresponding to multiple transmission lines as the edge weight of the edge formed by the corresponding two nodes.

[0095] Step 103: Using the spatiotemporal dynamic fault diagram and the fault type label corresponding to each node in the spatiotemporal dynamic fault diagram, train the spatiotemporal graph neural network to obtain the initial recognition model.

[0096] In some embodiments of the present invention, a Spatial-Temporal Graph Neural Network (STGNN) is a deep learning model specifically designed to process data that combines spatial topological relationships with temporal dynamic changes. It typically contains three main modules: a spatial modeling module, a temporal modeling module, and a spatiotemporal fusion module.

[0097] In some embodiments of the present invention, step 103 provided in the above embodiments can be implemented by the following steps 1031 to 1033 (not shown in the figures):

[0098] Step 1031: Based on each node in the spatiotemporal dynamic fault graph, input the node features of the node into the spatiotemporal graph neural network to predict the fault type and obtain the predicted fault type corresponding to the node.

[0099] Step 1032: Based on each node, use a multi-task loss function to determine the loss between the predicted fault type corresponding to the node and the fault type label corresponding to the node, and obtain the loss value of the node.

[0100] In some embodiments of the present invention, the multi-task loss function may include classification loss and spatiotemporal consistency loss. Correspondingly, the multi-task loss function can be expressed by the following formula (1):

[0101]

[0102] Among them, L cls y is the classification loss of the i-th node in the spatiotemporal dynamic fault graph; N is the number of nodes in the spatiotemporal dynamic fault graph; C is the number of fault types corresponding to the spatiotemporal dynamic fault graph; i,c c is the fault type label c corresponding to the i-th node in the spatiotemporal dynamic fault graph; p i,c L represents the probability that the i-th node in the spatiotemporal dynamic fault graph belongs to fault category c; graph ε represents the spatiotemporal consistency loss of the i-th node in the spatiotemporal dynamic fault graph; ε represents the edge set corresponding to the grid topology connection table of the multiple transmission lines; h i h is the hidden feature of the i-th node in the spatiotemporal dynamic fault graph; j ω is the hidden feature of the j-th node in the spatiotemporal dynamic fault graph; ij L represents the edge weight of the transmission line between the i-th node and the j-th node in the spatiotemporal dynamic fault graph; total α is the loss value of the i-th node in the spatiotemporal dynamic fault graph; α and β are both hyperparameters of the spatiotemporal graph neural network.

[0103] Step 1033: Adjust the model parameters of the spatiotemporal graph neural network according to the loss of the nodes in the spatiotemporal dynamic fault graph until the output loss meets the convergence condition, and obtain the initial recognition model.

[0104] In some embodiments of the present invention, the model parameters of the obtained spatiotemporal graph neural network (such as normalization layer parameters, graph convolution parameters, temporal convolution or pooling parameters in time modeling, etc.) can be adjusted according to the loss of all nodes in the spatiotemporal dynamic fault graph until a model whose corresponding output loss satisfies the convergence condition is obtained, thereby using the model as the initial recognition model.

[0105] Step 104: Under the constraints of the physical laws of the circuit system, the model parameters of the initial identification model are optimized based on the swarm intelligence search mechanism of the wolf pack algorithm to obtain the fault type identification model of the transmission line.

[0106] In some embodiments of the present invention, the physical laws of the circuit system may include: current constraint laws, voltage constraint laws, and component characteristic laws; correspondingly, the constraint conditions corresponding to the physical laws of the circuit system are the constraint conditions under the current constraint laws, voltage constraint laws, and component-specific laws. Here, the constraint conditions may be: line impedance constraints and electromagnetic transient characteristic constraints.

[0107] In some embodiments of this invention, a swarm intelligence search mechanism based on wolf pack algorithms is introduced to optimize the model parameters of the initial identification model. By simulating the predation behavior of a wolf pack, and utilizing the swarm intelligence search mechanism (i.e., intelligent cooperation and information sharing mechanism) of the wolf pack, the optimal parameter combination is searched within a search range determined under the constraints corresponding to the physical laws of the circuit system, thereby improving the derivation accuracy of the initial identification model. Specifically, the intelligent behaviors of the wolf pack algorithm, such as roaming, summoning, and surrounding, can be used to optimize the parameters in the initial identification model.

[0108] It should be noted that the wolf pack algorithm, with its unique intelligent behaviors such as roaming, summoning, and encirclement, provides a new optimization approach. This wolf pack algorithm comprises three main phases:

[0109] Wandering Phase: The wolf pack algorithm simulates the dispersed search behavior of a wolf pack, extensively searching for possible high-quality solutions in the parameter space, and initially locating the parameter regions that have a significant impact on model performance.

[0110] Summoning Phase: Individual wolves in the pack exchange and share information, bringing together the high-quality solutions they have found, and guiding the pack to converge towards a region with better parameters.

[0111] Siege phase: The wolf pack concentrates its strength to conduct a fine search and adjustment of the locked high-quality area, thereby finding the optimal combination of model parameters.

[0112] By applying the wolf pack algorithm to optimize the parameters of the initial identification model, model complexity and computational load can be effectively reduced, while improving the model's generalization ability. This intelligent optimization method not only enhances the accuracy of the initial identification model but also enables it to better adapt to the complex and ever-changing working environment of the power system, providing a strong guarantee for the stable operation of the power system.

[0113] In some embodiments of the present invention, step 104 above can be implemented by the following steps 1041 to 1044 (not shown in the figure):

[0114] Step 1041: Obtain the initial value range of each model parameter among the multiple model parameters of the initial identification model under the constraint conditions corresponding to the physical laws of the circuit system.

[0115] In some embodiments of the present invention, the constraints corresponding to the physical laws of the circuit system may include: line impedance constraints and electromagnetic transient characteristic constraints.

[0116] In some embodiments of the present invention, the initial recognition model includes, but is not limited to, the number of neural network layers, learning rate, etc.

[0117] In some embodiments of the present invention, where the constraints corresponding to the physical laws of the circuit system may include line impedance constraints and electromagnetic transient characteristic constraints, the above step 1041 can be implemented by the following steps A1 and A2:

[0118] Step A1: Based on each of the model parameters, obtain the range of values ​​of the model parameter under the line impedance constraint and the range of values ​​of the model parameter under the electromagnetic transient characteristic constraint.

[0119] Step A2: Based on each of the model parameters, the intersection between the range of values ​​of the model parameter under the line impedance constraint and the range of values ​​under the electromagnetic transient characteristic constraint is taken as the initial range of values ​​of the model parameter.

[0120] Step 1042: Based on each model parameter, the swarm intelligence search mechanism of the wolf pack algorithm is used to search and optimize the initial value range of each model parameter to obtain the optimal value of the model parameter.

[0121] Step 1043: Combine the optimal values ​​of the multiple model parameters to obtain the optimal parameter combination of the initial recognition model.

[0122] Step 1044: Using the optimal parameter combination, train the initial identification model to obtain the fault type identification model.

[0123] In some embodiments of the present invention, the optimal parameter combination is used to train the initial identification model to obtain the fault type identification model. The specific implementation steps may include the following:

[0124] The first step is to train the initial recognition model using the optimal parameter combination to obtain the recognition model to be adjusted.

[0125] The second step involves using a spatiotemporal dynamic fault map and the fault type label corresponding to each node in the spatiotemporal dynamic fault map to evaluate the performance of the identification model to be adjusted, and obtaining the evaluation results.

[0126] The third step is to adjust the model structure and parameters of the identification model to be adjusted based on the evaluation results, so as to obtain the fault type identification model.

[0127] The performance evaluation of the model to be adjusted can include assessing its accuracy, recall, F1 score, etc.

[0128] In some embodiments of the present invention, the fault type identification model can also be obtained in the following ways:

[0129] 1. Initialize the initial identification model using the optimal parameter combination, and conduct supervised training using the training nodes of the spatiotemporal dynamic fault map and the corresponding fault type labels to obtain the identification model to be adjusted;

[0130] 2. Performance Evaluation: The spatiotemporal dynamic fault map is divided into a training set and a test set in a 7:3 ratio. The test set retains the nodes with the latest timestamps. The prediction accuracy, recall, and macro F1 score of the model for fault type labels on the test set are calculated, and a confusion matrix is ​​generated.

[0131] 3. Model Adjustment: If the macro F1 score is < 0.85, try the following optimizations in sequence:

[0132] a) Increase the number of layers in the graph neural network to 3, and add residual connections between layers;

[0133] b) Introduce class weights into the loss function to address the imbalance problem;

[0134] If the accuracy is <0.9, adjust the learning rate decay strategy or introduce gradient clipping.

[0135] 4. Repeat steps 2-3 until the evaluation indicators reach the preset threshold, and finally output the fault type identification model.

[0136] The present invention provides a method for constructing a fault type identification model for transmission lines. In the execution of this method, firstly, traveling wave data is collected from multiple transmission lines experiencing different faults within the transmission network to obtain a fault traveling wave dataset; secondly, the fault traveling wave dataset is analyzed using the power grid topology connection table of the transmission network to obtain a spatiotemporal dynamic fault diagram; then, the spatiotemporal dynamic fault diagram and the fault type label corresponding to each node in the spatiotemporal dynamic fault diagram are used to train a spatiotemporal graph neural network to obtain an initial identification model; finally, under the constraints corresponding to the physical laws of the circuit system, the model parameters of the initial identification model are optimized based on a swarm intelligence search mechanism using the wolf pack algorithm to obtain the fault type identification model for the transmission lines. Thus, on the one hand, an initial identification model is constructed using spatiotemporal dynamic fault maps and spatiotemporal graph neural networks. This allows the constructed initial identification model to simultaneously capture the spatiotemporal correlation of fault signals, thereby improving the identification accuracy of the initial identification model. On the other hand, by utilizing the constraints corresponding to the physical laws of the circuit system and the swarm intelligence search mechanism of the wolf pack algorithm, it is possible to avoid the model learning false features that violate physical laws. Furthermore, by using leader-follower collaborative search, the gradient descent of the model is prevented from getting trapped in local optima, resulting in a model whose generalization ability is significantly better than that of traditional grid search. This improves the accuracy and efficiency of the fault type identification model for transmission lines.

[0137] Example 2:

[0138] See Figure 3 The diagram shown is a flowchart illustrating a method for identifying fault types in power transmission lines according to an embodiment of the present invention. The following is a summary of the process. Figure 3 The steps shown are explained below:

[0139] Step 301: Collect traveling wave data at multiple times for the transmission lines currently experiencing faults within the transmission network to obtain real-time fault traveling wave data;

[0140] Step 302: Using the power grid topology connection table of the transmission network, analyze the real-time fault traveling wave data to obtain a real-time dynamic fault diagram;

[0141] Step 303: Input the real-time dynamic fault diagram into the fault type identification model of the transmission line constructed by the method described in any of the above methods to identify the fault type and obtain the fault type of the transmission line that is currently experiencing a fault.

[0142] In some embodiments of the present invention, a broadband current and voltage transformer deployed on the power transmission network can be used to collect data on the transmission line that is currently experiencing a fault in real time to obtain real-time fault traveling wave data. The real-time fault traveling wave data may correspond to four channels of real-time current waveform data consisting of Ia, Ib, Ic, and I0, and four channels of real-time voltage waveform data consisting of Ua, Ub, Uc, and U0.

[0143] In some embodiments of the present invention, the power grid topology connection table of the transmission network is used to analyze the real-time fault traveling wave data to obtain the real-time dynamic fault diagram. For the specific implementation of this method, please refer to the description of step 102, as well as steps 1021 and 1022 in the above embodiments, which will not be repeated here.

[0144] In some embodiments of the present invention, the fault type identification model for transmission lines constructed in any of the above embodiments can be used to identify the fault type of the real-time dynamic fault map to obtain the fault type of the transmission line currently experiencing a fault. Here, the fault type includes, but is not limited to, short-circuit faults, grounding faults, etc. The present invention does not impose any limitations on this.

[0145] Thus, based on the constraints corresponding to the physical laws of the circuit system mentioned above, the swarm intelligence search mechanism of the wolf pack algorithm, and the spatiotemporal dynamic fault map, the fault type identification model of the transmission line obtained by improving the spatiotemporal graph neural network can identify the fault type of the transmission line in real time by performing real-time fault type identification on the real-time dynamic fault map. This model can identify the fault type of the transmission line that has experienced a fault more accurately and efficiently.

[0146] Example 3:

[0147] Based on the same inventive concept, embodiments of the present invention also provide a system for constructing a fault type identification model for transmission lines, such as... Figure 4 The diagram shown is a schematic representation of a system for constructing a fault type identification model for transmission lines according to an embodiment of the present invention. The system 400 includes:

[0148] Data acquisition module 401 is used to acquire traveling wave data from multiple transmission lines with different faults in the power transmission network to obtain fault traveling wave dataset;

[0149] Analysis module 402 is used to analyze the fault traveling wave dataset using the power grid topology connection table of the transmission network to obtain a spatiotemporal dynamic fault diagram;

[0150] Training module 403 is used to train the spatiotemporal graph neural network using the spatiotemporal dynamic fault graph and the fault type label corresponding to each node in the spatiotemporal dynamic fault graph to obtain an initial recognition model;

[0151] The optimization module 404 is used to optimize the model parameters of the initial identification model under the constraints corresponding to the physical laws of the circuit system, based on the swarm intelligence search mechanism of the wolf pack algorithm, to obtain the fault type identification model of the transmission line.

[0152] Optionally, the analysis module 402 includes:

[0153] The feature extraction unit is used to extract time-domain features at multiple times based on each fault traveling wave data in the fault traveling wave dataset using a sliding window algorithm, so as to obtain the dynamic feature data of the fault traveling wave data.

[0154] An analysis unit is constructed to establish a spatiotemporal dynamic fault graph based on the power grid topology connection table. The graph uses the monitoring points corresponding to each of the multiple transmission lines as multiple nodes, the dynamic feature data of the fault traveling wave data as the node features of the corresponding nodes, and the line impedance connection parameters corresponding to the multiple transmission lines as edge weights.

[0155] Optionally, the training module 403 includes:

[0156] The fault type prediction unit is used to input the node features of each node into the spatiotemporal graph neural network to predict the fault type based on each node in the spatiotemporal dynamic fault graph, so as to obtain the predicted fault type corresponding to the node.

[0157] The loss determination unit is used to determine the loss between the predicted fault type and the fault type label corresponding to the node based on each node using a multi-task loss function, and obtain the loss value of the node.

[0158] The adjustment unit is used to adjust the model parameters of the spatiotemporal graph neural network according to the loss of the nodes in the spatiotemporal dynamic fault graph until the output loss meets the convergence condition, thereby obtaining the initial recognition model.

[0159] Optionally, the multi-task loss function can be expressed as:

[0160]

[0161] Among them, L cls y is the classification loss of the i-th node in the spatiotemporal dynamic fault graph; N is the number of nodes in the spatiotemporal dynamic fault graph; C is the number of fault types corresponding to the spatiotemporal dynamic fault graph; i,c c is the fault type label c corresponding to the i-th node in the spatiotemporal dynamic fault graph; p i,c L represents the probability that the i-th node in the spatiotemporal dynamic fault graph belongs to fault category c; graphε represents the spatiotemporal consistency loss of the i-th node in the spatiotemporal dynamic fault graph; ε represents the edge set corresponding to the grid topology connection table of the multiple transmission lines; h i h is the hidden feature of the i-th node in the spatiotemporal dynamic fault graph; j ω is the hidden feature of the j-th node in the spatiotemporal dynamic fault graph; ij L represents the edge weight of the transmission line between the i-th node and the j-th node in the spatiotemporal dynamic fault graph; total α is the loss value of the i-th node in the spatiotemporal dynamic fault graph; α and β are both hyperparameters of the spatiotemporal graph neural network.

[0162] Optionally, the optimization module 404 includes:

[0163] The acquisition unit is used to acquire the initial value range of each model parameter among the multiple model parameters of the initial identification model under the constraint conditions corresponding to the physical laws of the circuit system.

[0164] The search unit is used to search and optimize the initial value range of each model parameter based on each model parameter using the swarm intelligence search mechanism of the wolf pack algorithm, so as to obtain the optimal value of the model parameter.

[0165] The combination unit is used to combine the optimal values ​​of the multiple model parameters to obtain the optimal parameter combination of the initial recognition model;

[0166] The training unit is used to train the initial identification model using the optimal parameter combination to obtain the fault type identification model.

[0167] Optionally, the constraints corresponding to the physical laws of the circuit system include: line impedance constraints and electromagnetic transient characteristic constraints; the acquisition unit is specifically used to acquire, based on each model parameter, the value range of the model parameter under the line impedance constraint and the value range of the model parameter under the electromagnetic transient characteristic constraint; based on each model parameter, the intersection between the value range of the model parameter under the line impedance constraint and the value range under the electromagnetic transient characteristic constraint is used as the initial value range of the model parameter.

[0168] It should be noted that the description of the system for constructing the fault type identification model for transmission lines is similar to the description of the above-described method embodiment for constructing the fault type identification model for transmission lines, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the system embodiments of this invention, please refer to the description of the method embodiments of this invention for understanding.

[0169] Example 4:

[0170] Correspondingly, embodiments of the present invention also provide a fault type identification system for transmission lines, such as... Figure 5 The diagram shown is a schematic representation of a fault type identification system for transmission lines provided in an embodiment of the present invention. The system 500 includes:

[0171] The real-time acquisition module 501 is used to acquire traveling wave data of the transmission line currently experiencing a fault in the power transmission network at multiple times to obtain real-time fault traveling wave data.

[0172] The real-time analysis module 502 is used to analyze the real-time fault traveling wave data using the power grid topology connection table of the transmission network to obtain a real-time dynamic fault diagram.

[0173] The fault identification module 503 is used to input the real-time dynamic fault diagram into the fault type identification model of the transmission line constructed by the method described in any of the above methods to identify the fault type and obtain the fault type of the transmission line that is currently experiencing a fault.

[0174] It should be noted that the description of the embodiment of the fault type identification system for transmission lines is similar to the description of the embodiment of the fault type identification method for transmission lines described above, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the system embodiment of the present invention, please refer to the description of the method embodiment of the present invention for understanding.

[0175] Example 5:

[0176] Based on the same inventive concept, such as Figure 6 As shown, the present invention also provides an electronic device, which may be a computer device, a microcontroller device, a smart mobile device, etc. The electronic device in this embodiment may include a processor 610, a memory 620, a transceiver component 630, etc. The processor 610, the memory 620, and the transceiver component 630 are connected via a bus 640; the memory 620 can be used to store executable programs, and an exemplary executable program may include instructions; the processor 610 is used to execute the instructions stored in the memory. The memory 620 can also be used to store data, which can be accessed and / or modified when instructions are executed.

[0177] The processor may be a Central Processing Unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. It is the computing core and control core of the terminal, and it is suitable for implementing one or more instructions. Specifically, it is suitable for loading and executing one or more instructions in the storage medium to realize the corresponding method flow or corresponding function, so as to realize the method for constructing the fault type identification model of the transmission line in the above embodiment, or the steps of the method for constructing the fault type identification model of the transmission line.

[0178] Example 6:

[0179] Based on the same inventive concept, this invention also provides a readable storage medium, specifically an electronic device readable storage medium (Memory). This readable storage medium is a memory device within an electronic device used to store programs and data. It is understood that the storage medium here can include both built-in storage media within the electronic device and extended storage media supported by the electronic device. The storage medium provides storage space, which stores the terminal's operating system. Furthermore, this storage space also stores one or more instructions suitable for loading and execution by a processor. These instructions can be one or more executable programs (including program code). It should be noted that the storage medium here can be high-speed RAM or non-volatile memory, such as at least one disk storage device. The processor loads and executes one or more instructions stored in the storage medium to implement the method for constructing a fault type identification model for transmission lines in the above embodiments, or the steps of the method for constructing a fault type identification model for transmission lines.

[0180] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0181] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0182] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0183] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0184] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for constructing a fault type identification model for transmission lines, characterized in that, The method includes: Traveling wave data were collected from multiple transmission lines experiencing different faults within the power grid to obtain a fault traveling wave dataset; Using the power grid topology connection table of the transmission network, the fault traveling wave dataset is analyzed to obtain a spatiotemporal dynamic fault diagram; The spatiotemporal dynamic fault diagram and the fault type label corresponding to each node in the spatiotemporal dynamic fault diagram are used to train the spatiotemporal graph neural network to obtain an initial recognition model. Under the constraints of the physical laws of the circuit system, the model parameters of the initial identification model are optimized using a swarm intelligence search mechanism based on the wolf pack algorithm to obtain the fault type identification model of the transmission line.

2. The method according to claim 1, characterized in that, The method of using the power grid topology connection table of the transmission network to analyze the fault traveling wave dataset and obtain a spatiotemporal dynamic fault diagram includes: Based on each fault traveling wave data in the fault traveling wave dataset, the sliding window algorithm is used to extract time-domain features at multiple times to obtain the dynamic feature data of the fault traveling wave data. Based on the power grid topology connection table, a spatiotemporal dynamic fault graph is established, with the monitoring points corresponding to each of the multiple transmission lines as multiple nodes, the dynamic characteristic data of the fault traveling wave data as the node characteristics of the corresponding nodes, and the line impedance connection parameters corresponding to the multiple transmission lines as edge weights.

3. The method according to claim 2, characterized in that, The process involves training a spatiotemporal graph neural network using the spatiotemporal dynamic fault graph and the fault type label corresponding to each node in the spatiotemporal dynamic fault graph to obtain an initial recognition model, including: Based on each node in the spatiotemporal dynamic fault graph, the node features of the node are input into the spatiotemporal graph neural network to predict the fault type, thereby obtaining the predicted fault type corresponding to the node. Based on each node, a multi-task loss function is used to determine the loss between the predicted fault type corresponding to the node and the fault type label corresponding to the node, thereby obtaining the loss value of the node. Based on the loss of the nodes in the spatiotemporal dynamic fault graph, the model parameters of the spatiotemporal graph neural network are adjusted until the output loss meets the convergence condition, thus obtaining the initial recognition model.

4. The method according to claim 3, characterized in that, The multi-task loss function can be expressed as: Among them, L cls y is the classification loss of the i-th node in the spatiotemporal dynamic fault graph; N is the number of nodes in the spatiotemporal dynamic fault graph; C is the number of fault types corresponding to the spatiotemporal dynamic fault graph; i,c c is the fault type label c corresponding to the i-th node in the spatiotemporal dynamic fault graph; p i,c L represents the probability that the i-th node in the spatiotemporal dynamic fault graph belongs to fault category c; graph ε represents the spatiotemporal consistency loss of the i-th node in the spatiotemporal dynamic fault graph; ε represents the edge set corresponding to the grid topology connection table of the multiple transmission lines; h i h is the hidden feature of the i-th node in the spatiotemporal dynamic fault graph; j ω is the hidden feature of the j-th node in the spatiotemporal dynamic fault graph; ij L represents the edge weight of the transmission line between the i-th node and the j-th node in the spatiotemporal dynamic fault graph; total α is the loss value of the i-th node in the spatiotemporal dynamic fault graph; α and β are both hyperparameters of the spatiotemporal graph neural network.

5. The method according to claim 1, characterized in that, Under the constraints corresponding to the physical laws of the circuit system, the swarm intelligence search mechanism based on the wolf pack algorithm optimizes the model parameters of the initial identification model to obtain the fault type identification model of the transmission line, including: Under the constraints corresponding to the physical laws of the circuit system, obtain the initial value range of each model parameter among the multiple model parameters of the initial identification model; Based on each model parameter, the swarm intelligence search mechanism of the wolf pack algorithm is used to search and optimize the initial value range of each model parameter to obtain the optimal value of the model parameter. The optimal values ​​of the multiple model parameters are combined to obtain the optimal parameter combination of the initial recognition model; The initial identification model is trained using the optimal parameter combination to obtain the fault type identification model.

6. The method according to claim 5, characterized in that, The constraints corresponding to the physical laws of the circuit system include: line impedance constraints and electromagnetic transient characteristic constraints; obtaining the initial value range of each model parameter among the multiple model parameters of the initial identification model under the constraints corresponding to the physical laws of the circuit system includes: Based on each of the model parameters, the value range of the model parameter under the line impedance constraint and the value range of the model parameter under the electromagnetic transient characteristic constraint are obtained respectively. Based on each of the model parameters, the intersection between the range of values ​​of the model parameter under the line impedance constraint and the range of values ​​under the electromagnetic transient characteristic constraint is taken as the initial range of values ​​of the model parameter.

7. A system for constructing a fault type identification model for transmission lines, characterized in that, The system includes: The data acquisition module is used to collect traveling wave data from multiple transmission lines experiencing different faults within the power transmission network, and obtain a fault traveling wave dataset. The analysis module is used to analyze the fault traveling wave dataset using the power grid topology connection table of the transmission network to obtain a spatiotemporal dynamic fault diagram. The training module is used to train the spatiotemporal graph neural network using the spatiotemporal dynamic fault graph and the fault type label corresponding to each node in the spatiotemporal dynamic fault graph to obtain an initial recognition model. The optimization module is used to optimize the model parameters of the initial identification model under the constraints corresponding to the physical laws of the circuit system, based on the swarm intelligence search mechanism of the wolf pack algorithm, so as to obtain the fault type identification model of the transmission line.

8. The system according to claim 7, characterized in that, The analysis module includes: The feature extraction unit is used to extract time-domain features at multiple times based on each fault traveling wave data in the fault traveling wave dataset using a sliding window algorithm, so as to obtain the dynamic feature data of the fault traveling wave data. An analysis unit is constructed to establish a spatiotemporal dynamic fault graph based on the power grid topology connection table. The graph uses the monitoring points corresponding to each of the multiple transmission lines as multiple nodes, the dynamic feature data of the fault traveling wave data as the node features of the corresponding nodes, and the line impedance connection parameters corresponding to the multiple transmission lines as edge weights.

9. A method for identifying fault types in transmission lines, characterized in that, The method includes: Multi-moment acquisition of traveling wave data of the transmission line currently experiencing a fault within the power transmission network yields real-time fault traveling wave data; The real-time fault traveling wave data is analyzed using the power grid topology connection table of the transmission network to obtain a real-time dynamic fault diagram; The real-time dynamic fault diagram is input into the fault type identification model of the transmission line constructed by the method described in any one of claims 1 to 6 to identify the fault type and obtain the fault type of the transmission line that is currently experiencing a fault.

10. A fault type identification system for transmission lines, characterized in that, The system includes: The real-time acquisition module is used to acquire traveling wave data at multiple moments from the transmission lines currently experiencing faults within the power grid, thereby obtaining real-time fault traveling wave data. The real-time analysis module is used to analyze the real-time fault traveling wave data using the power grid topology connection table of the transmission network to obtain a real-time dynamic fault diagram. The fault identification module is used to input the real-time dynamic fault diagram into the fault type identification model of the transmission line constructed by the method described in any one of claims 1 to 6 to identify the fault type and obtain the fault type of the transmission line that is currently experiencing a fault.

Images