A power grid fault warning system based on topological change scenarios

Abstract

The present invention discloses a power grid fault early warning system based on a topological change scenario, which relates to the technical field of power grid fault early warning; a topological graph construction module constructs a network topological graph; an anomaly location module determines whether each element in the network topological graph is abnormal according to the comparison parameters and the actual parameters; a topological reconstruction module obtains a set topological graph based on the element anomalies, and substitutes the set topological graph into a Bayesian compressive sensing model to obtain a target network topological graph; a fault probability calculation module calculates the fault probability of each node; a setting value determination module determines the setting value of each node according to the fault probability of each node. By performing network topology reconstruction through the Bayesian compressive sensing model, the adaptability and flexibility of the early warning system are improved. Then, by calculating the fault probability of each node, the fault risks of each node in the power grid are quantified, the efficiency and reliability of power grid fault early warning are improved, and the safe and stable operation of the power system is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of power grid fault warning, and particularly relates to a power grid fault warning system based on topological change scenarios. Background Art

[0002] As the cornerstone of the operation of modern society, the stability and security of the power system are directly related to economic development and social stability. However, the occurrence of power grid faults is inevitable. Once they occur, they will not only cause huge economic losses but also may trigger serious social impacts. Therefore, the warning of power grid faults is particularly important.

[0003] A power distribution network fault detection system proposed in the invention patent publication number CN113376479B includes fault indicators, a master station, and a mobile terminal. Each time a new fault occurs, the fault recording module searches and compares the fault records within a previously set time period A to find out whether a fault has occurred in the same area. If a fault has occurred in the same area within the previous time period A, a warning message will be displayed on the display module so as to quickly locate the fault area of the power grid. However, this method cannot effectively respond when the power grid topology changes. This method mainly relies on static monitoring data, and this method can only play a better role when the power grid topology is stable. However, the current power grid topology is becoming more and more complex and often changes, which makes the fault detection method based on static data difficult to cope with the changing power grid environment, resulting in low efficiency and reliability of the fault warning system. Summary of the Invention

[0004] The purpose of the present invention is to solve the problem that the traditional fault detection method based on static data is difficult to cope with the changing power grid environment, resulting in low efficiency and reliability of the fault warning system, and to propose a power grid fault warning system based on topological change scenarios.

[0005] A power grid fault warning system based on topological change scenarios is proposed in the embodiment of the present invention. The system includes:

[0006] A topology graph construction module, configured to construct a network topology graph according to the power grid structure of the target power grid area, and obtain an adjacency matrix according to the network topology graph;

[0007] Anomaly location module, which is used to obtain the environmental data corresponding to each element in the network topology diagram in real time, substitute all the environmental data into a preset model to obtain the corresponding reference parameters for each element, and determine whether each element is abnormal according to the reference parameters and actual parameters; each element in the network topology diagram is any node or any edge in the network topology diagram; the environmental data is the voltage amplitude, voltage phase, current, phase angle of each node in the network topology diagram, and the resistance and reactance of each line; the actual parameter is the environmental data of this element.

[0008] Topology reconstruction module, which is used to remove the abnormal elements in the network topology diagram to obtain a rectified topology diagram, and substitute the rectified topology diagram into the Bayesian compressive sensing model to perform network topology reconstruction to obtain a target network topology diagram.

[0009] Fault probability calculation module, which is used to calculate the fault probability of each node in the target network topology diagram.

[0010] Setting value determination module, which is used to determine the setting value of each node in the target network topology diagram according to the fault probability of each node.

[0011] Optionally, the anomaly location module includes:

[0012] Data preprocessing module, which is used to detect the environmental data of each element in real time and preprocess all the environmental data to obtain a target data set.

[0013] Admittance calculation module, which is used to calculate the admittance of each line according to the resistance and reactance of each line, and obtain the conductivity matrix and susceptance matrix according to the admittance of each line.

[0014] Equation establishment module, which is used to establish an active power equation, a reactive power equation, a current equation and a phase angle equation according to the target data set, the conductivity matrix and the susceptance matrix.

[0015] Preset model generation module, which is used to obtain the preset model by replacing the corresponding formulas in the power flow calculation model according to the active power equation, the reactive power equation, the current equation and the phase angle equation.

[0016] Optionally, the equation establishment module includes:

[0017] Active power equation establishment module, which is used to obtain the active power through the formula

[0018] where,

[0019] is the active power between node i and adjacent node j, g is the conductivity between node i and node j, b ij is the susceptance between node i and node j, andij is the susceptance rate between node i and node j, v i and v j are the voltage amplitudes of node i and node j respectively, θ i and θ j are the voltage phase angles of node i and node j respectively.

[0020] Optionally, the equation establishment module further includes:

[0021] A reactive power equation establishment module, which is used to obtain reactive power through the formula

[0022] ;

[0023] where is the reactive power of node i between adjacent nodes j, g ij is the conductivity between node i and node j, b ij is the susceptance rate between node i and node j, v i and v j are the voltage amplitudes of node i and node j respectively, θ i and θ j are the voltage phase angles of node i and node j respectively.

[0024] Optionally, the equation establishment module further includes:

[0025] A current equation establishment module, which is used to obtain current through the formula ;

[0026] where is the current of node i between adjacent nodes j, v i and v j are the voltage amplitudes of node i and node j respectively, z ij is the resistance between node i and node j.

[0027] Optionally, the equation establishment module further includes:

[0028] A phase angle equation module, which is used to obtain the phase angle through the formula σ i = θ i - θ0;

[0029] where σ i is the phase angle of node i, θ i and θ0 are the phase angles of node i and the reference waveform of node i respectively.

[0030] Optionally, the anomaly location module further includes:

[0031] A difference calculation and judgment module, configured to calculate the difference between the actual parameter and the reference parameter corresponding to each environmental parameter type of each element to obtain a judgment difference;

[0032] An abnormal element determination module, configured to, if the judgment differences corresponding to all environmental parameter types of an element are greater than a preset threshold, mark the element as abnormal; the preset thresholds for the environmental parameter types corresponding to each element are different.

[0033] Optionally, the topology reconstruction module includes:

[0034] An initialization module, configured to generate an adjacency matrix of the set topology graph, denoted as a set adjacency matrix, and substitute the set adjacency matrix and the environmental data corresponding to all elements in the set topology graph into a Bayesian compressive sensing model for initialization to obtain initialization parameters;

[0035] A sparse signal determination module, configured to construct an observation model according to the initialization parameters and the set topology graph, and perform parameter estimation and signal reconstruction through a Markov chain Monte Carlo method to obtain a sparse signal;

[0036] A target network topology graph determination module, configured to determine the connection relationship between nodes according to the non-zero elements in the sparse signal, and generate the target network topology graph.

[0037] Optionally, the fault probability calculation module includes:

[0038] A topology sub-graph determination module, configured to determine the number of power-off area topology connection sub-graphs of each node through a graph theory algorithm according to the target network topology graph;

[0039] A first feature extraction module, configured to extract features from all power-off area topology connection sub-graphs of a node to obtain a first feature for each node;

[0040] A second feature extraction module, configured to obtain the alarm information feature corresponding to the node in the historical fault data, denoted as a second feature;

[0041] A fault probability determination module, configured to substitute the first feature and the second feature into a preset fault probability model to obtain the fault probability of the node.

[0042] Advantages of the present invention:

[0043] The present invention provides a power grid fault early warning system based on a topological change scenario, including a topological graph construction module for constructing a network topological graph according to the power grid structure of a target power grid area and obtaining an adjacency matrix based on the network topological graph; an anomaly location module for, for each element in the network topological graph, obtaining the corresponding environmental data of the element in real time, substituting all the environmental data into a preset model to obtain the corresponding reference parameter for each element, and determining whether the element is abnormal for each element according to the reference parameter and the actual parameter; a topological reconstruction module for removing the abnormal positions of the elements in the network topological graph to obtain a rectified topological graph, substituting the rectified topological graph into a Bayesian compressive sensing model to perform network topological reconstruction to obtain a target network topological graph; a fault probability calculation module for calculating the fault probability of each node in the target network topological graph; and a setting value determination module for determining the setting value of each node in the target network topological graph according to the fault probability of each node. Through the topological graph construction module and the anomaly location module, the system can obtain the environmental data of each element of the power grid in real time and dynamically update the power grid topological structure. This real-time monitoring and dynamic analysis greatly improve the accuracy and timeliness of fault detection, and avoid the lag and inaccuracy of traditional static monitoring methods in dealing with a frequently changing power grid environment. After detecting abnormal elements in the power grid, through the Bayesian compressive sensing model for network topological reconstruction, the system can effectively cope with the complex and changeable power grid environment, improve the adaptability and flexibility of the early warning system, and then by calculating the fault probability of each node, quantify the fault risk of each node in the power grid, and determine the setting value of each node according to the fault probability to enable the power grid to perform adaptive adjustment according to the actual operating state, improve the efficiency and reliability of power grid fault early warning, and ensure the safe and stable operation of the power system. Brief Description of the Drawings

[0044] The present invention will be further described below with reference to the accompanying drawings.

[0045] Figure 1 The framework diagram of a power grid fault early warning system based on a topological change scenario is provided for an embodiment of the present invention. Detailed Embodiments

[0046] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with 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. The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, in the present invention, descriptions such as "first" and "second" are only for descriptive purposes, and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include at least one such feature. In addition, the technical solutions between various embodiments can be combined with each other, but it must be based on the fact that those skilled in the art can implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0047] Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0048] The embodiments of the present invention provide a power grid fault warning system based on a topological change scenario. See Figure 1 , Figure 1 is a framework diagram of a power grid fault warning system based on a topological change scenario provided by an embodiment of the present invention. The system includes the following steps:

[0049] A topology graph construction module, configured to construct a network topology graph according to the power grid structure of the target power grid area, and obtain an adjacency matrix according to the network topology graph;

[0050] An anomaly location module, configured to, for each element in the network topology graph, obtain the environmental data corresponding to the element in real time, substitute all the environmental data into a preset model to obtain the corresponding comparison parameter for each element, and determine whether the element is abnormal according to the comparison parameter and the actual parameter for each element;

[0051] A topology reconstruction module, configured to remove the abnormal elements from the network topology graph to obtain a set-topology graph, and substitute the set-topology graph into a Bayesian compressive sensing model to perform network topology reconstruction to obtain a target network topology graph;

[0052] A fault probability calculation module, configured to calculate the fault probability of each node in the target network topology graph;

[0053] A setting value determination module, configured to determine the setting value of each node in the target network topology graph according to the fault probability of each node.

[0054] Among them, each element in the network topology diagram is any node or any edge in the network topology diagram; the environmental data is the voltage amplitude, voltage phase, current, phase angle of each node in the network topology diagram and the resistance and reactance of each line; the actual parameter is the environmental data of the element.

[0055] Based on a power grid fault early warning system based on a topology change scenario provided by an embodiment of the present invention, the system can obtain the environmental data of each element of the power grid in real time and dynamically update the power grid topology structure through a topology map construction module and an abnormality positioning module. This real-time monitoring and dynamic analysis greatly improves the accuracy and timeliness of fault detection, avoids the lag and inaccuracy of traditional static monitoring methods when dealing with frequently changing power grid environments, and after detecting abnormal elements in the power grid, reconstructs the network topology through a Bayesian compressed sensing model. The system can effectively cope with complex and changeable power grid environments, improves the adaptability and flexibility of the early warning system, and then quantifies the failure risk of each node in the power grid by calculating the failure probability of each node, and determines the setting value of each node through the failure probability so that the power grid can be adaptively adjusted according to the actual operating status, thereby improving the efficiency and reliability of power grid fault early warning and ensuring the safe and stable operation of the power system.

[0056] In one implementation, the nodes of the network topology diagram are composed of distribution transformers, switch stations, power plants, substations, power users, etc., and the edges are determined by actual connection relationships.

[0057] In one implementation, the preset model is constructed by first preprocessing the environmental data of each element in the power grid, including voltage amplitude, voltage phase, current, phase angle, and resistance and reactance of each line, to remove noise and outliers, and organize them into a target data set in a unified format; then calculate the admittance of each line based on the resistance and reactance of each line. This step converts each line in the power grid into a representation of its conductive performance, summarizes the admittance information of all lines, generates a conductivity matrix and an susceptance matrix, and establishes basic equations related to power grid operation based on the preprocessed target data set and the admittance matrix, including active power equation, reactive power equation, current equation and phase angle equation. These equations are used to describe the electrical relationship between each node and line in the power grid, and the above basic equations are integrated together to form an overall preset model. This model is used to replace the traditional power flow calculation model and reflect the dynamic electrical relationship between each node and line in the power grid.

[0058] In one implementation, by constructing and updating a network topology map and an adjacency matrix, the system can reflect changes in the grid structure in real time, thereby improving the speed and accuracy of response to changes in the grid state.

[0059] In one implementation, the environmental data of each element of the power grid is obtained and processed in real time, which can quickly identify and locate faults when the power grid environment changes dynamically, enhancing the sensitivity and early warning ability of the system.

[0060] In one implementation, by substituting the environmental data into a preset model to obtain comparison parameters and comparing them with the actual parameters, it can accurately judge whether the power grid elements are abnormal, thus improving the accuracy of fault detection.

[0061] In one implementation, after detecting abnormal elements in the power grid, it can quickly remove the faulty elements and reconstruct the network topology diagram, and use the Bayesian compressive sensing model for reconstruction to ensure that the system can still be quickly adjusted and restored in case of faults, maintaining the continuity and stability of the power grid.

[0062] In one implementation, according to the fault probability of each node, its setting value is dynamically adjusted to ensure the stable operation of the power grid in a dynamically changing environment, so as to perform dynamic adjustment. The system can adapt to the actual operating state of the power grid, improving the operating efficiency and reliability.

[0063] In one embodiment, the anomaly location module includes:

[0064] A data preprocessing module, which is used to detect the environmental data of each element in real time and preprocess all the environmental data to obtain a target data set;

[0065] An admittance calculation module, which is used to calculate the admittance of each line according to the resistance and reactance of each line, and obtain the conductivity matrix and susceptance matrix according to the admittance of each line;

[0066] An equation establishment module, which is used to establish an active power equation, a reactive power equation, a current equation and a phase angle equation according to the target data set, the conductivity matrix and the susceptance matrix;

[0067] A preset model generation module, which is used to obtain a preset model by replacing the corresponding formulas in the power flow calculation model according to the active power equation, the reactive power equation, the current equation and the phase angle equation.

[0068] In one implementation, the preprocessing includes noise removal, outlier detection, data correction, and missing value processing, and all of these preprocessings use existing technologies.

[0069] In one implementation, the admittance is calculated according to the resistance and reactance of each line, and the conductivity matrix and susceptance matrix are generated. These matrices accurately describe the conductive characteristics of each line in the power grid and are important inputs for subsequent equation establishment and preset model generation.

[0070] In one implementation, using the target data set and the admittance matrix, an active power equation, a reactive power equation, a current equation and a phase angle equation are established to comprehensively reflect the electrical characteristics of the power grid.

[0071] In one implementation, according to the established active power equation, reactive power equation, current equation and phase angle equation, the traditional power flow calculation model is replaced to generate a preset model with more accurate and dynamic response, so as to perform fault detection and early warning more efficiently, adapt to the dynamic changes of the power grid, and improve the efficiency of calculation and analysis.

[0072] In one embodiment, the equation establishment module includes:

[0073] The active power equation establishment module is used to obtain the active power through the formula

[0074] ;

[0075] where is the active power between node i and adjacent node j, g ij is the conductivity between node i and node j, b ij is the susceptance rate between node i and node j, v i and v j are the voltage amplitudes of node i and node j respectively, θ i and θ j are the voltage phase angles of node i and node j respectively.

[0076] In one embodiment, the equation establishment module further includes:

[0077] The reactive power equation establishment module is used to obtain the reactive power through the formula

[0078] ;

[0079] where is the reactive power between node i and adjacent node j, g ij is the conductivity between node i and node j, b ij is the susceptance rate between node i and node j, v i and v j are the voltage amplitudes of node i and node j respectively, θ i and θ j are the voltage phase angles of node i and node j respectively.

[0080] In one embodiment, the equation establishment module further includes:

[0081] The current equation establishment module is used to obtain the current through the formula ;

[0082] where is the current between node i and adjacent node j, v i and v jare the voltage amplitudes of nodes i and j respectively, and z ij is the resistance between nodes i and j.

[0083] In one embodiment, the equation establishing module further includes:

[0084] a phase angle equation module, configured to obtain a phase angle through the formula σ i = θ i - θ0;

[0085] wherein, σ i is the phase angle of node i, and θ i and θ0 are the phase angles of node i and the reference waveform of node i respectively.

[0086] In one embodiment, the abnormal location module further includes:

[0087] a judgment difference calculation module, configured to calculate the difference between the actual parameter and the control parameter corresponding to each environmental parameter type for each element to obtain a judgment difference;

[0088] an abnormal element determination module, configured to mark the element as abnormal if the judgment differences corresponding to all environmental parameter types are greater than a preset threshold; the preset thresholds for each environmental parameter type corresponding to each element are different.

[0089] In one implementation, by calculating the difference, the system can quantify the deviation degree of environmental data and provide a more objective and quantitative basis for abnormal judgment.

[0090] In one implementation, by judging the judgment differences of all environmental parameter types, the system can perform multi-dimensional comprehensive analysis to avoid misjudgment caused by a single parameter. Each environmental parameter type has a different preset threshold, and the system can set different sensitivities for different types of parameters to improve the flexibility and accuracy of detection.

[0091] In one implementation, the fact that the preset thresholds for each environmental parameter type corresponding to each element are different specifically means that, according to the environmental parameters, the types of each environmental parameter are classified, such as voltage amplitude, voltage phase, current, phase angle, resistance of each line, reactance of each line. The data of different environmental parameters are extracted from the historical operation records, and statistical analysis is performed on each parameter to determine its statistical characteristics such as mean, standard deviation, maximum value, minimum value, etc. Corresponding thresholds are set for each statistical characteristic, and all thresholds are mapped to the same length in the corresponding data, and the average value of the thresholds corresponding to each statistical characteristic is calculated to obtain the preset threshold for the environmental parameter type corresponding to the element; the corresponding thresholds for each statistical characteristic are determined by technical personnel. For example, the set threshold can be a data point outside the range of the mean ± 2 standard deviations, and the set threshold can be 80% of the maximum value, etc.

[0092] In one implementation, only when the judgment difference of all environmental parameter types is greater than a preset threshold value, the element is recorded as an anomaly, thereby reducing the probability of false alarms and improving the accuracy of anomaly detection.

[0093] In one embodiment, the topology reconstruction module includes:

[0094] An initialization module is used to generate an adjacency matrix of a tuning topology map, which is recorded as a tuning adjacency matrix. The environment data corresponding to all elements in the tuning adjacency matrix and the tuning topology map are substituted into a Bayesian compressed sensing model for initialization to obtain initialization parameters.

[0095] The sparse signal determination module is used to construct an observation model according to the initialization parameters and the tuning topology map, and obtain the sparse signal by performing parameter estimation and signal reconstruction through the Markov chain Monte Carlo method;

[0096] The target network topology map determination module is used to determine the connection relationship between nodes according to the non-zero elements in the sparse signal and generate the target network topology map.

[0097] In one implementation, by generating an adjacency matrix of a set topology graph, the system can accurately reflect the connection relationship of the current power grid and provide accurate initial data for the subsequent Bayesian compressed sensing model.

[0098] In one implementation, the environmental data is combined with the set adjacency matrix to initialize the model, so that the model can more comprehensively reflect the actual operating status of the power grid and ensure the accuracy of the initialization parameters.

[0099] In one implementation, parameter estimation and signal reconstruction are performed using the Markov Chain Monte Carlo method, which can handle complex probability distributions, provide high-quality parameter estimation results, and improve the accuracy of the model.

[0100] In one implementation, a topology reconstruction method based on sparse signals can quickly determine the actual connection structure of the power grid, ensuring that the system can respond and adjust in a timely manner when the power grid structure changes.

[0101] In one embodiment, the failure probability calculation module includes:

[0102] A topology subgraph determination module is used to determine the number of topology connection subgraphs of the power outage area of ​​each node through a graph theory algorithm according to the target network topology graph;

[0103] A first feature extraction module is used to extract features from all power outage area topology connection subgraphs of each node to obtain a first feature;

[0104] The second feature extraction module is used to obtain the alarm information features corresponding to the node in the historical fault data, denoted as the second feature;

[0105] The fault probability determination module is used to substitute the first feature and the second feature into a preset fault probability model to obtain the fault probability of the node.

[0106] In one implementation, by using a graph theory algorithm to determine the number of topological connection subgraphs in the power-off area of each node, it is possible to accurately identify the areas in the power grid that may be affected, ensuring rapid location of the fault point and the scope of influence.

[0107] In one implementation, for each node, extracting the features of all topological connection subgraphs in the power-off area of the node can comprehensively analyze the operating status and potential fault risks of each node in the power grid.

[0108] In one implementation, the historical fault data records various faults and their detailed information that occurred during the past operation of the power grid, including the fault occurrence time, fault location, fault type, environmental data, fault duration, fault influence scope, fault cause analysis, etc.; the fault information is a further analysis and refinement of the historical fault data, including alarm information, recovery measures, fault consequences, fault statistics, etc.

[0109] In one implementation, by analyzing the alarm information features in the historical fault data, the system can identify common fault patterns in the power grid, enhancing the prediction ability.

[0110] In one implementation, substituting the first feature and the second feature into a preset fault probability model can comprehensively evaluate the fault probability of each node, provide a more comprehensive fault prediction result, and calculate personalized fault probabilities according to the characteristics and historical data of different nodes, improving the accuracy and pertinence of the prediction.

[0111] In one implementation, the generation of the fault probability model is specifically as follows: collect the past fault data and alarm information of the power grid, remove noise and outliers from the fault data and alarm information, fill in missing values, and perform normalization processing to obtain a historical data set. Divide the historical data set into a validation set and a training set, train the training set through a random forest algorithm to obtain a training model, verify it with the validation set, and then through cross-validation and hyperparameter tuning, adjust the model until the difference between the results obtained from the validation set and the training set is within a preset range to obtain the fault probability model; the preset range is determined by technical personnel.

[0112] The above has described an embodiment of the present invention in detail, but the content is only a preferred embodiment of the present invention and cannot be considered as limiting the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the application of the present invention should still fall within the scope covered by the patent of the present invention.

Claims

1. A power grid fault early warning system based on a topological change scenario, characterized in that, The system comprises: A topology map construction module is used to construct a network topology map according to the power grid structure of the target power grid area, and obtain an adjacency matrix according to the network topology map; The abnormality positioning module is used to obtain the environmental data corresponding to each element in the network topology diagram in real time, substitute all the environmental data into the preset model to obtain the control parameters corresponding to each element, and determine whether the element is abnormal according to the control parameters and actual parameters for each element; each element is any node or any edge in the network topology diagram; the environmental data is the voltage amplitude, voltage phase, current, phase angle of each node and the resistance and reactance of each line; the actual parameters are the environmental data of the element; The preset model is constructed as follows: first, the environmental data of each element in the power grid, including voltage amplitude, voltage phase, current, phase angle, and resistance and reactance of each line, are preprocessed to remove noise and outliers and organized into a target data set in a unified format; According to the resistance and reactance of each line, the admittance of each line is calculated. This step converts each line in the power grid into a representation of its conductive performance, summarizes the admittance information of all lines, generates a conductivity matrix and a susceptance matrix, and establishes basic equations related to power grid operation based on the preprocessed target data set and the admittance matrix, including active power equation, reactive power equation, current equation and phase angle equation; A topology reconstruction module is used to remove abnormal elements in the network topology map to obtain a set topology map, and substitute the set topology map into a Bayesian compressed sensing model to perform network topology reconstruction to obtain a target network topology map; A failure probability calculation module, used to calculate the failure probability of each node in the target network topology diagram; A setting value determination module, used to determine the setting value of each node in the target network topology diagram according to the failure probability of each node; The topology reconstruction module includes: An initialization module is used to generate an adjacency matrix of a tuning topology map, which is recorded as a tuning adjacency matrix, and substitute the tuning adjacency matrix and the environmental data corresponding to all elements in the tuning topology map into a Bayesian compressed sensing model for initialization to obtain initialization parameters; A sparse signal determination module, used for constructing an observation model according to the initialization parameters and the tuning topology map, and performing parameter estimation and signal reconstruction through a Markov chain Monte Carlo method to obtain a sparse signal; The target network topology map determining module is used to determine the connection relationship between nodes according to the non-zero elements in the sparse signal to generate the target network topology map.

2. The power grid fault early warning system based on the topological change scenario according to claim 1, wherein The abnormality positioning module includes: The data preprocessing module is used to detect the environmental data of each element in real time and preprocess all environmental data to obtain the target data set; An admittance calculation module is used to calculate the admittance of each line according to the resistance and reactance of each line, and obtain a conductivity matrix and a susceptance matrix according to the admittance of each line; An equation building module, used to build an active power equation, a reactive power equation, a current equation and a phase angle equation according to the target data set, the conductivity matrix and the susceptance matrix; A preset model generation module, configured to obtain the preset model by replacing corresponding formulas in the power flow calculation model according to the active power equation, the reactive power equation, the current equation, and the phase angle equation.

3. The power grid fault warning system based on the topological change scenario according to claim 2, characterized in that, The equation establishment module includes: An active power equation establishment module, configured to use the formula Obtain the active power; Among them, is the active power between node i and adjacent node j, g ij is the conductivity between node i and node j, b ij is the susceptance rate between node i and node j, v i and v j are the voltage amplitudes of node i and node j respectively, θ i and θ j are the voltage phase angles of node i and node j respectively.

4. A power grid fault warning system based on a topological change scenario according to claim 2, characterized in that The equation establishment module further includes: A reactive power equation establishment module, configured to use the formula Obtain reactive power; Among them, is the reactive power between node i and adjacent node j, g ij is the conductivity between node i and node j, b ij is the susceptance rate between node i and node j, v i and v j are the voltage amplitudes of node i and node j respectively, θ i and θ j are the voltage phase angles of node i and node j respectively.

5. The power grid fault early warning system based on a topological change scenario according to claim 2, wherein, The equation establishment module further includes: The current equation establishment module is used to obtain the current through the formula ; Among them, is the current between node i and adjacent node j, v i and v j are the voltage amplitudes of node i and node j respectively, and z ij is the resistance between node i and node j.

6. The power grid fault warning system based on a topological change scenario according to claim 2, characterized in that, The equation establishment module further includes: Phase angle equation module, which is used to obtain the phase angle through the formula σ i = θ i - θ0 where, σ i is the phase angle of node i, and θ i and θ0 are the phase angle of node i and the phase angle of the reference waveform of node i, respectively.

7. The power grid fault early warning system based on the topological change scenario according to claim 2, characterized in that, The anomaly location module further includes: A judgment difference calculation module, configured to calculate the difference between the actual parameter and the reference parameter corresponding to each environmental parameter type for each element to obtain a judgment difference; An anomaly element determination module, configured to mark the element as an anomaly if the judgment differences corresponding to all environmental parameter types are greater than a preset threshold; the preset thresholds for each environmental parameter type corresponding to each element are different.

8. The power grid fault warning system based on a topological change scenario according to claim 1, characterized in that The fault probability calculation module includes: A topology subgraph determination module, configured to determine the number of topology connection subgraphs in the power-off area of each node according to the target network topology graph through a graph theory algorithm; A first feature extraction module, configured to extract features from all topology connection subgraphs in the power-off area of each node to obtain a first feature for each node; A second feature extraction module, configured to obtain the alarm information feature corresponding to the node in the historical fault data and record it as the second feature; A fault probability determination module, configured to substitute the first feature and the second feature into a preset fault probability model to obtain the fault probability of the node.

Images