Distribution Line Fault Location Method and System Based on Beidou System

Abstract

The present invention discloses a distribution line fault location method and system based on the Beidou system, which relates to the technical field of line fault analysis. It includes collecting key node data, defining node states according to the status data of distributed power sources, fault indicators and smart meters, constructing a data matrix, and calculating the principal component matrix of the status data; constructing a fault mode recognition model for the principal component matrices of different status data and performing model fusion; calculating the state probability of the node at the next moment based on historical fault data. The method of the present invention constructs a model through the principal component matrices from three perspectives of distributed power sources, fault indicators and smart meters, which can more comprehensively reflect the multi-dimensional characteristics of faults. The comprehensive fault mode vector generated through perspective fusion can more accurately represent the overall fault state of the current power system, reduce the possible biases and blind spots of single-perspective models, and thus improve the accuracy of fault location.

Description

Technical Field

[0001] The present invention relates to the technical field of line fault analysis, in particular to a method and system for fault location of distribution lines based on the Beidou system. Background Art

[0002] With the continuous expansion of the scale of the power system and the increase in the access ratio of distributed power sources, the structure of the distribution network has become increasingly complex, and the difficulty of fault location and diagnosis has increased significantly. Traditional methods for fault location of distribution lines mainly rely on a single data source of fault indicators and smart meters, such as fault current, switch status, and voltage and current changes;

[0003] However, when facing multi-point faults, instantaneous faults, and complex power grid topologies, these methods have limitations of insufficient location accuracy and slow response speed. At the same time, due to the access of distributed power sources, it is difficult for traditional fault location methods to accurately distinguish the power disturbances caused by distributed power source fluctuations from actual faults, which easily leads to misjudgment and missed judgment. Moreover, relying solely on a single or a small number of data sources for analysis is difficult to comprehensively reflect the complex relationships of various state data in the power system, reducing the accuracy of fault location and being unfavorable for the rapid recovery after faults and the stable operation of the power system. Summary of the Invention

[0004] In view of the problems existing in the above-mentioned existing method and system for fault location of distribution lines based on the Beidou system, the present invention is proposed.

[0005] Therefore, the problem to be solved by the present invention is that due to the access of distributed power sources, it is difficult for traditional fault location methods to accurately distinguish the power disturbances caused by distributed power source fluctuations from actual faults, which easily leads to misjudgment and missed judgment. Moreover, relying solely on a single or a small number of data sources for analysis is difficult to comprehensively reflect the complex relationships of various state data in the power system, reducing the accuracy of fault location and being unfavorable for the rapid recovery after faults and the stable operation of the power system.

[0006] To solve the above technical problems, the present invention provides the following technical solutions: A method for fault location of distribution lines based on the Beidou system, which includes,

[0007] Collect key node data, define node states according to the state data of distributed power sources, fault indicators, and smart meters, construct a data matrix, and calculate the principal component matrix of the state data;

[0008] Construct a fault mode recognition model for the principal component matrices of different state data and perform model fusion;

[0009] Calculate the state probability of the next moment node based on historical fault data, and calculate the state probability of the current time step through Bayes' theorem combined with the comprehensive fault mode vector;

[0010] Analyze the failure probability of regional nodes based on key nodes, perform secure encryption and transmission of data, and conduct off-site backup and verification regularly.

[0011] As a preferred solution of the distribution line fault location method based on the Beidou system according to the present invention, wherein: collecting the data of key nodes, and defining the node status according to the status data of distributed power sources, fault indicators and smart meters, including,

[0012] Define the load center node, the main line branch node, the substation node and the high-fault area node as key nodes based on the distribution line;

[0013] Deploy an integrated monitoring unit at the key nodes of the distribution line, including an integrated Beidou communication module, an AMI smart meter and a fault indicator;

[0014] Poll and detect the deployed nodes based on the AMI smart meter, and collect the status data of the AMI smart meter, including voltage, current, active power, reactive power and meter status;

[0015] Based on the obtained status data of the AMI smart meter, determine the meter status of the AMI smart meter. If the meter is online and working properly, the status value is judged to be 1. If the meter is disconnected, the status value is judged to be 0;

[0016] At the same time, determine based on the status data of the fault indicator. If the fault indicator detects a fault current, the status value of the indicator is judged to be 1. If the signal of the fault indicator is normal, the status value of the indicator is judged to be 0;

[0017] Collect the status data of the distributed power source, including real-time power generation, voltage, frequency offset and current injection.

[0018] As a preferred solution of the distribution line fault location method based on the Beidou system according to the present invention, wherein: constructing a data matrix and calculating the principal component matrix of the status data, including,

[0019] Construct data matrices based on the status data of distributed power sources, fault indicators and smart meters respectively. The rows of the matrix represent different observation samples, and the columns represent different features;

[0020] Standardize the matrix number, calculate the covariance matrix of the matrix, and perform eigenvalue decomposition on the covariance matrix to obtain the eigenvalue vector and the eigenvector matrix;

[0021] Sort the eigenvalue vectors according to their magnitudes, accumulate the sum of the eigenvalue vectors until it accounts for 95% of the sum of the total eigenvalue vectors, form a dimensionality reduction matrix, and multiply it with the standardized data matrix for data dimensionality reduction, expressed as:

[0022] ;

[0023] wherein represents the principal component matrix of the i-th state data after dimensionality reduction, represents the dimensionality reduction matrix of the i-th state data, represents the data matrix of the i-th state data.

[0024] As a preferred solution of the power distribution line fault location method based on the Beidou system described in the present invention, wherein: constructing a fault mode recognition model for the principal component matrices of different state data and performing model fusion includes,

[0025] using a multi-layer perceptron MLP and constructing fault mode recognition models respectively according to the principal component matrices from three perspectives, expressed as:

[0026] ;

[0027] wherein represents the model output of the principal component matrix of the i-th state data, represents the ReLU activation function, represents the weight of the i-th state data, represents the bias term of the i-th state data;

[0028] using the training set and calculating the error between the model prediction and the true fault category using the cross-entropy loss function, performing iterative optimization of the model parameters through the Adam optimizer and by the gradient descent method, and stopping the iteration and outputting the updated model parameters when the loss of the model no longer significantly decreases during continuous iteration;

[0029] Based on the fault mode recognition models from three perspectives, performing model fusion according to the weighting coefficients, expressed as:

[0030] ;

[0031] wherein represents the comprehensive fault mode vector of the k-th sample, represents the total number of perspectives, represents the weighting coefficient of the principal component matrix of the i-th state data;

[0032] using the training set and calculating the error between the model prediction and the true fault category using the cross-entropy loss function, calculating the gradient of the loss function with respect to each weighting coefficient, performing iterative optimization of the model parameters through the Adam optimizer and by the gradient descent method, and stopping the iteration and outputting the updated model parameters when the loss of the model no longer significantly decreases during continuous iteration.

[0033] As a preferred solution of the power distribution line fault location method based on the Beidou system of the present invention, wherein: calculating the state probability of the next moment node based on historical fault data, and calculating the state probability of the current time step through Bayes' theorem combined with the comprehensive fault mode vector, including,

[0034] Regarding the key nodes of the power distribution line as the nodes of the dynamic Bayesian network DBN, and defining whether the node is in a fault state at a certain time point according to the node state;

[0035] Determine the dependency relationship of the DBN nodes according to the topological structure of the power distribution line, draw a directed acyclic graph DAG, represent the state transition relationship between the nodes and determine the network structure of the dynamic Bayesian network;

[0036] Based on historical fault data, count the number of faults that occur at node i and the number of faults that occur at the adjacent node j in the next time step, and divide by the total number of faults at node i to calculate the single state probability between the two nodes, and construct a state transition matrix according to the state probabilities of different nodes;

[0037] At the same time, based on historical fault data, divide the total number of faults in the historical data of the node by the total number of acquisitions at all time steps of the node to calculate the prior probability of the node;

[0038] Calculate the state probability of the next moment node according to the state transition matrix and the state probability of the current moment, expressed as:

[0039] ;

[0040] Where n represents the total number of nodes, and represent the state probabilities of node j at time step t + 1 and node i at time step t respectively, represents the state of node i at time t transferred to the state of node j at time t + 1 of the transition probability;

[0041] Through Bayes' theorem, combine the state probability with the comprehensive fault mode vector to update the state probability of each node at the current time step, expressed as:

[0042] ;

[0043] Where represents the posterior fault probability based on the comprehensive fault mode vector and the node state probability at the current moment, represents the probability that the node state is when observing the comprehensive fault mode vector of node i represents the state probability of node i, Represents the total probability observed under all node states ;

[0044] Record the state probability of the next moment of the nodes in the historical data statistics nodes, and select the accurate state probability of the next moment of the nodes according to the actual fault information to calculate the sum of the average value and the standard deviation as the prediction threshold. If the calculated state probability of the next moment of the nodes is greater than or equal to the prediction threshold, node fault warning is carried out.

[0045] As a preferred scheme of the power distribution line fault location method based on the Beidou system described in the present invention, wherein: the fault probability of the regional nodes is analyzed according to the key nodes, including,

[0046] Utilize the high-precision positioning function of the Beidou system to obtain the real-time position coordinates of each monitoring node, and calculate the Euclidean distance between two nodes through the geographical coordinates of the two nodes , and based on the key nodes, select the maximum distance value between all node pairs in the region according to the geographical location and normalize it;

[0047] Based on the posterior fault probability at the current time step and based on the Euclidean distance of the nodes, calculate the fault location point probability of the regional nodes, expressed as:

[0048] ;

[0049] Where represents the fault location point probability of regional node i, represents the posterior fault probability of node j, represents the maximum Euclidean distance of the regional nodes, L represents the total number of nodes in the key node region, represents the Euclidean distance between nodes i and j;

[0050] Based on the fault location point probability of the regional nodes, calculate the sum of the average value of the fault probabilities of all regional joint nodes and twice the standard deviation as the location threshold based on historical data. If the calculated fault location point probability of the regional nodes is greater than the location threshold, it is judged as a regional fault.

[0051] As a preferred scheme of the power distribution line fault location method based on the Beidou system described in the present invention, wherein: the data is securely encrypted and transmitted, and off-site backup and verification are carried out regularly, including,

[0052] For monitoring nodes such as distributed power sources, fault indicators, and smart meters, through the Beidou system terminal device, use the TLS transport layer security protocol encryption channel to send the information to the monitoring center in real time;

[0053] Use the AES encryption algorithm for the detection data to encrypt and store the location information of the monitoring nodes, and use the role-based access control (RBAC) mechanism to limit the data access permissions;

[0054] Store the backup data in a remote server and perform data integrity verification.

[0055] Another object of the present invention is to provide a system for a distribution line fault location method based on the Beidou system, which includes,

[0056] A data acquisition module that collects data based on distributed power sources, fault indicators, and smart meters and performs preprocessing;

[0057] A fault mode recognition module that calculates the covariance matrix for the state data matrices of different data sources, performs eigenvalue decomposition, constructs a single-view fault mode recognition model, and performs model fusion;

[0058] A fault prediction module that calculates the state probability prediction for the next moment based on historical fault data and calculates the state probability for the current time step based on Bayes' theorem;

[0059] A regional node fault analysis module that calculates the fault probabilities of all nodes within each critical node region and comprehensively analyzes the fault probability distribution of the regional nodes;

[0060] A data security management module that encrypts and stores the location information and state data of the monitoring nodes, regularly backs up the data to a remote server, and performs data integrity verification.

[0061] A computer device includes: a memory and a processor; the memory stores a computer program, and when the processor executes the computer program, the steps of the above-mentioned distribution line fault location method based on the Beidou system are implemented.

[0062] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned distribution line fault location method based on the Beidou system are implemented.

[0063] The beneficial effects of the present invention are as follows: By constructing a model based on the principal component matrix from three perspectives of distributed power sources, fault indicators, and smart meters, it can more comprehensively reflect the multi-dimensional characteristics of faults. The comprehensive fault mode vector generated through perspective fusion can more accurately represent the overall fault state of the current power system, reducing the possible biases and blind spots of a single-perspective model, thereby improving the accuracy of fault location. By combining the comprehensive fault mode vector with the state probability at the current time step, it integrates the state characteristics from different perspectives and analyzes the operating state of nodes from multiple angles. By introducing the dual constraints of posterior fault probability and geographical distance, the accuracy of fault location is improved, enabling the accurate calculation of the probability of fault location points within the region and effectively identifying the fault nodes within the region. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0065] Figure 1 It is a schematic flowchart of a method for fault location of a distribution line based on the Beidou system.

[0066] Figure 2 It is a schematic structural diagram of a fault location system for a distribution line based on the Beidou system. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0067] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the drawings in the specification.

[0068] Many specific details are set forth in the following description in order to fully understand the present invention. However, the present invention can also be implemented in other ways different from those described herein. Those skilled in the art can make similar generalizations without departing from the connotation of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed below.

[0069] Secondly, the so-called "one embodiment" or "embodiment" herein refers to a specific feature, structure, or characteristic that can be included in at least one implementation manner of the present invention. The appearances of "in one embodiment" in different places in this specification do not all refer to the same embodiment, nor are they separate or selectively exclusive embodiments from other embodiments.

[0070] Embodiment 1, refer to Figure 1, which is the first embodiment of the present invention. This embodiment provides a method for fault location of distribution lines based on the Beidou system. The method for fault location of distribution lines based on the Beidou system includes:

[0071] S1, collect key node data, define node states according to the status data of distributed power sources, fault indicators and smart meters, construct a data matrix, and calculate the principal component matrix of the status data;

[0072] Preferably, collecting key node data and defining node states according to the status data of distributed power sources, fault indicators and smart meters includes:

[0073] Define the load center node, the main line branch node, the substation node and the high - fault - prone area node as key nodes based on the distribution line;

[0074] Deploy an integrated monitoring unit at the key nodes of the distribution line, including an integrated Beidou communication module, an AMI smart meter and a fault indicator;

[0075] Poll and detect the deployed nodes based on the AMI smart meter, and collect the status data of the AMI smart meter, including voltage, current, active power, reactive power and meter status;

[0076] Based on the obtained status data of the AMI smart meter, determine the meter status of the AMI smart meter. If the meter is online and working properly, the status value is judged as 1. If the meter is disconnected, the status value is judged as 0;

[0077] At the same time, determine based on the status data of the fault indicator. If the fault indicator detects a fault current, the status value of the indicator is judged as 1. If the signal of the fault indicator is normal, the status value of the indicator is judged as 0;

[0078] Collect the status data of the distributed power source, including real - time power generation, voltage, frequency offset and current injection.

[0079] By defining the load center nodes, main line branch nodes, substation nodes, and nodes in high-fault areas in the distribution line as key nodes and deploying integrated monitoring units at these key nodes, precise monitoring of key positions in the entire distribution network can be achieved. By collecting multi-dimensional status data (such as voltage, current, power, frequency, etc.) from AMI smart meters, fault indicators, and distributed power sources, and combining the working status of the meters and indicators, the operating conditions of the distribution line can be comprehensively analyzed, and the real-time operating status of each key node can be obtained promptly and accurately. The integrated monitoring unit integrates a Beidou communication module, an AMI smart meter, and a fault indicator, which can not only transmit high-precision position information but also comprehensively obtain power parameters and fault status information. By jointly analyzing the monitoring data of distributed power sources and main lines, the overall health status of the power system can be better evaluated.

[0080] Furthermore, construct a data matrix and calculate the principal component matrix of the status data, including

[0081] Construct data matrices based on the status data of distributed power sources, fault indicators, and smart meters respectively. The rows of the matrix represent different observation samples, and the columns represent different features;

[0082] Standardize the matrix numbers and calculate the covariance matrix of the matrix, denoted as:

[0083] ;

[0084] where represents the covariance matrix of the i-th status data matrix, represents the transpose calculation of the i-th status data matrix, and n represents the number of observation samples of the i-th status data standardized matrix;

[0085] Perform eigenvalue decomposition on the covariance matrix to obtain the eigenvalue vector and eigenvector matrix, denoted as:

[0086] ;

[0087] where represents the covariance matrix of the i-th status data, represents the eigenvector matrix of the i-th status data, where each column vector is an eigenvector, represents the diagonal matrix of the eigenvalues of the i-th status data, where the values on the diagonal are the respective eigenvalue vectors. The eigenvalue vectors are obtained by decomposing the covariance matrix using the eig algorithm in linear algebra in the NumPy library to obtain the eigenvalue vector and eigenvector matrix;

[0088] Sort by the eigenvalue vectors in descending order, accumulate the sum of the eigenvalue vectors until it accounts for 95% of the total sum of the eigenvalue vectors, form a dimensionality reduction matrix, and multiply it with the standardized data matrix for data dimensionality reduction, expressed as:

[0089] ;

[0090] where represents the principal component matrix of the i-th state data for dimensionality reduction, represents the dimensionality reduction matrix of the i-th state data, represents the data matrix of the i-th state data.

[0091] By standardizing different state data matrices, the dimensions of all features (such as voltage, current, power, etc.) are made consistent, thus avoiding analysis biases caused by differences in the magnitudes of different feature values. Data matrices are constructed based on the state data of distributed power sources, fault indicators, and smart meters respectively, which can effectively fuse the feature information of multiple data sources, enabling the fault detection system to comprehensively grasp the state information of each node in the power grid, form information complementarity between different data sources (such as power fluctuations of distributed power sources, abnormal voltage and current of smart meters, and fault currents of fault indicators), and thus more accurately identify fault features. By sorting the eigenvalue vectors in descending order and selecting the main features (accounting for 95%), most of the feature information related to faults can be retained, while ignoring the secondary information unrelated to faults, thereby improving the ability to identify abnormal states and faults. By calculating the covariance matrix, the correlation structure between different state features can be discovered. Combining with the use of eigenvector decomposition and dimensionality reduction techniques, the data dimension can be effectively reduced, enabling the system to still perform efficient and accurate data analysis and fault detection when facing large-scale and multi-dimensional data.

[0092] S2. Construct a fault mode recognition model for the principal component matrices of different state data and perform model fusion;

[0093] Preferably, constructing a fault mode recognition model for the principal component matrices of different state data and performing model fusion includes,

[0094] Using a multi-layer perceptron MLP and constructing fault mode recognition models respectively according to the principal component matrices from three perspectives, expressed as:

[0095] ;

[0096] where represents the model output of the principal component matrix of the i-th state data, represents the ReLU activation function, represents the weight of the i-th state data, Denotes the bias term of the i-th state data;

[0097] Use the training set and calculate the error between the model prediction and the true fault category using the cross-entropy loss function. Iteratively optimize the model parameters through the Adam optimizer and the gradient descent method. Stop the iteration and output the model parameters to update the model when the loss of the model no longer decreases significantly during consecutive iterations.

[0098] Based on the three-perspective fault mode recognition model, perform model fusion according to the weighting coefficients, expressed as:

[0099] ;

[0100] where Denotes the comprehensive fault mode vector of the k-th sample, Denotes the total number of perspectives, Denotes the weighting coefficient of the i-th state data principal component matrix;

[0101] Use the training set and calculate the error between the model prediction and the true fault category using the cross-entropy loss function. Calculate the gradient of the loss function with respect to each weighting coefficient. Iteratively optimize the model parameters through the Adam optimizer and the gradient descent method. Stop the iteration and output the model parameters to update the model when the loss of the model no longer decreases significantly during consecutive iterations.

[0102] Construct a model through the principal component matrices from three perspectives of distributed power sources, fault indicators, and smart meters. Each perspective's principal component matrix extracts the core information of different state data. At the same time, multi-perspective fusion can more comprehensively reflect the multi-dimensional characteristics of faults, improving the ability to identify complex fault modes. Among them, by introducing the distributed power source state data, it is possible to improve the judgment of the impact of the increasing distributed power sources on the operation of the distribution network. Through principal component analysis of the distributed power source data with different access rates, the operation characteristics of the power grid under different access conditions can be extracted, so as to achieve efficient fault identification under different distributed power source access rates. In scenarios with a relatively high distributed power source access rate, it is possible to better analyze the complex fault modes caused by power fluctuations. While in the case of a relatively low access rate, the model can pay more attention to the operation status of the load end and the backbone network, enhancing the system's adaptability to different scenarios;

[0103] Through the non-linear transformation of multiple hidden layers (such as the ReLU activation function), the MLP model can learn complex non-linear relationships, thus better modeling the complex relationships between various state features and fault modes in the power system. Based on the weighted fusion of models from different perspectives, it can automatically adjust the weights according to the contribution degree of different state data in fault mode recognition. The comprehensive fault mode vector generated through perspective fusion can more accurately represent the overall fault state of the current power system, reduce the possible biases and blind spots of a single perspective model, thereby improving the accuracy of fault location. The fault mode recognition model combining the multi-layer perceptron MLP with the principal component matrix can effectively improve the accuracy and efficiency of power system fault detection. At the same time, the model fusion based on the perspective weighting coefficient further enhances the robustness and flexibility of the system.

[0104] S3. Calculate the state probability of the next moment node based on historical fault data, and calculate the state probability of the current time step through Bayes' theorem combined with the comprehensive fault mode vector.

[0105] Preferably, calculating the state probability of the next moment node based on historical fault data and calculating the state probability of the current time step through Bayes' theorem combined with the comprehensive fault mode vector includes

[0106] Regarding the key nodes of the distribution line as the nodes of the dynamic Bayesian network DBN, and defining whether the node is in a fault state at a certain time point according to the node state.

[0107] Determine the dependency relationship of DBN nodes according to the topological structure of the distribution line (for example, if two nodes share a transmission line, they may have a direct state dependency relationship), draw a directed acyclic graph DAG, represent the state transition relationship between nodes and determine the network structure of the dynamic Bayesian network. Each directed edge in the directed acyclic graph DAG Indicates that the state of node i at time t will affect the state of node j at time t + 1.

[0108] Based on historical fault data, count the number of times that node j fails at the next time step when node i fails, and divide it by the total number of times node i fails to calculate the single state probability between the two nodes, and construct a state transition matrix according to the state probabilities of different nodes.

[0109] At the same time, based on historical fault data, divide the total number of faults in the historical data of the node by the total number of acquisition times of all time steps of the node to calculate the prior probability of the node.

[0110] Calculate the state probability of the next moment node according to the state transition matrix and the state probability of the current moment, expressed as:

[0111] ;

[0112] where n represents the total number of nodes, and represent the state probabilities of node j at time step t + 1 and node i at time step t respectively, represents the state of node i at time t transferring to the state of node j at time t + 1 is the transition probability;

[0113] By Bayes' theorem, combining the state probability with the comprehensive fault mode vector to update the state probability of each node at the current time step, which is expressed as:

[0114] ;

[0115] where represents the posterior fault probability based on the comprehensive fault mode vector and the node state probability at the current moment, represents that the node state is when observing the comprehensive fault mode vector of node i, represents the state probability of node i, represents under all node states observing is the total probability;

[0116] Record the historical data to statistically calculate the state probability of the next moment of the node, and select the accurate state probability of the next moment of the node according to the actual fault information to calculate the sum of the average value and the standard deviation as the prediction threshold. If the calculated state probability of the next moment of the node is greater than or equal to the prediction threshold, a node fault warning is issued.

[0117] By taking the key nodes of the distribution line as the nodes of the dynamic Bayesian network, it is possible to accurately monitor and predict the state changes of these nodes that are crucial to the operation of the power grid. By using a directed acyclic graph (DAG) to clarify the state transfer path between nodes, it is possible to effectively capture the fault propagation characteristics between different nodes in the power grid. Based on the state transition matrix calculated from historical data, it can reflect the fault transfer law between nodes. By combining the observed data (comprehensive fault mode vector) at the current moment with the historical state probability and using Bayes' theorem to dynamically update the fault probability of the node, it is possible to correct the state probability according to real-time data, making the fault detection and prediction more accurate and real-time, and effectively reducing false alarms and missed alarms.

[0118] By utilizing the status data of AMI smart meters, fault indicators, and distributed power sources and integrating them into the calculation of DBN, the system's ability to identify different fault types can be effectively improved. After real-time observation data (such as the comprehensive fault mode vector) arrives, the dynamic Bayesian network can quickly update the node state probability, thus realizing the dynamic early warning of the fault state, ensuring that the system can promptly perceive the change of the node state, quickly send out early warning signals, and improving the real-time response ability of the system;

[0119] By combining the comprehensive fault mode vector with the state probability at the current time step, the state characteristics from different perspectives are integrated, and the operating state of the node is analyzed from multiple angles. Compared with traditional fault detection that often relies on a single data source, the processing efficiency of the massive and complex data obtained based on the Beidou system and the ability to identify complex fault modes are improved. The state probability of the node can be dynamically updated when the observation data arrives, and dynamic adjustment can be carried out without relying on complete historical data, improving the flexibility and accuracy of fault diagnosis. By combining the comprehensive fault mode vector with the state probability at the current time step, fine-grained distinction can be made among multiple states, providing a basis for the location of regional faults, distinguishing short-term fluctuations caused by power grid disturbances from long-term abnormalities caused by equipment failures, and avoiding false alarms or missed alarms;

[0120] By combining the state update of real-time data and the state prediction of the next moment, the future state can be inferred based on real-time detection, so as to give early warning and handle abnormal states in a timely manner, and preventive measures can be taken before the fault occurs, thereby reducing the probability and impact of the fault.

[0121] S4. Analyze the fault probability of regional nodes according to key nodes, perform security encryption and transmission on the data, and conduct off-site backup and verification regularly;

[0122] Preferably, analyzing the fault probability of regional nodes according to key nodes includes,

[0123] Utilize the high-precision positioning function of the Beidou system to obtain the real-time position coordinates of each monitoring node, and calculate the Euclidean distance between two nodes through the geographical coordinates of the two nodes , and select the maximum distance value between all node pairs in the region based on the key nodes according to the geographical location and normalize it;

[0124] Based on the posterior fault probability at the current time step and the Euclidean distance of the node, calculate the fault location point probability of the regional node, expressed as:

[0125] ;

[0126] where represents the fault location point probability of regional node i, Represents the posterior fault probability of node j. Represents the maximum Euclidean distance of the regional nodes. L represents the total number of nodes within the critical node area. Represents the Euclidean distance between nodes i and j.

[0127] Based on the fault location point probability of the regional nodes, calculate the sum of the average fault probability and twice the standard deviation of all regional joint nodes based on historical data as the location threshold. If the calculated fault location point probability of the regional nodes is greater than the location threshold, it is determined as a regional fault.

[0128] Obtain the high-precision position coordinates of the monitoring nodes through the Beidou system and calculate the Euclidean distance between the nodes. High-precision positioning can accurately determine the geographical locations of each node, avoiding the deviation of fault location caused by position errors, thereby improving the accuracy of fault location. Through the accurate calculation of the fault location point probability within the region, the fault nodes within the region can be effectively identified, avoiding misjudging the fault as other normal regional nodes. And in combination with the judgment of the location threshold, a fault warning signal can be sent in time to avoid the expansion and spread of the fault. By accurately calculating the fault location point probability within the region, the specific location where the fault occurs can be quickly located, reducing the time and labor costs for fault troubleshooting.

[0129] By introducing the dual constraints of posterior fault probability and geographical distance, the accuracy of fault location is improved. Traditional methods only consider the weighting of geographical distance or static prior probability, ignoring the accurate description of node status by real-time data. Combining real-time posterior fault probability with geographical distance can more accurately judge the actual status of fault nodes. Through Bayesian update, the posterior fault probability can be dynamically adjusted at any time according to new observation data, making the fault location no longer rely on fixed historical patterns, enhancing the flexibility of fault location, and being able to make accurate judgments in different fault scenarios.

[0130] Furthermore, perform security encryption and transmission on the data, and regularly conduct off-site backup and verification, including

[0131] For monitoring nodes such as distributed power sources, fault indicators, and smart meters, through the Beidou system terminal device, use the TLS transport layer security protocol encryption channel to send the information to the monitoring center in real time.

[0132] Use the AES encryption algorithm to encrypt and store the location information of the monitoring nodes for the detection data, and use the role-based access control RBAC mechanism to limit the data access rights.

[0133] Store the backup data in an off-site server and conduct data integrity verification.

[0134] By using the TLS encryption protocol to encrypt the data transmission of monitoring nodes such as distributed power sources, fault indicators, and smart meters, it can effectively prevent eavesdropping, tampering, and forgery of data during the transmission process. Through the secure storage and transmission of data, the storage security of data and the defense ability of the system are improved. By using the access control mechanism of roles, it prevents unauthorized users from randomly accessing or modifying sensitive data, avoiding data abuse and illegal operations. By backing up data in different locations, the correctness and integrity of the backup data can be ensured.

[0135] Example 2, referring to Figure 2 , which is the second embodiment of the present invention. This embodiment is different from the previous one and provides a system for the fault location method of the distribution line based on the Beidou system, including

[0136] A data acquisition module that collects and preprocesses data based on distributed power sources, fault indicators, and smart meters;

[0137] A fault mode recognition module that calculates the covariance matrix for the state data matrices of different data sources, performs eigenvalue decomposition, constructs a single-view fault mode recognition model, and performs model fusion;

[0138] A fault prediction module that calculates the state probability prediction for the next moment based on historical fault data and calculates the state probability for the current time step based on Bayes' theorem;

[0139] A regional node fault analysis module that calculates the fault probabilities of all nodes within each key node region and comprehensively analyzes the fault probability distribution of the regional nodes;

[0140] A data security management module that encrypts and stores the location information and state data of the monitoring nodes, regularly backs up the data to a remote server, and performs data integrity verification.

[0141] If the above functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, mobile hard disks, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), magnetic disks, or optical discs, etc., which can store program codes.

[0142] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0143] More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion (electronic device) having one or more wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, a computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then storing it in a computer memory.

[0144] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or combinations thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0145] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A distribution line fault location method based on Beidou system, characterized in that: include, Collect key node data, define node status according to the status data of distributed power sources, fault indicators and smart meters, build a data matrix, and calculate the principal component matrix of the status data; Construct a fault mode recognition model for the principal component matrix of different state data and perform model fusion; Calculate the state probability of the node at the next moment based on historical fault data, and calculate the state probability of the current time step by combining the comprehensive fault mode vector with the Bayesian theorem; Analyze the failure probability of regional nodes based on the a posteriori failure probability of key nodes, securely encrypt and transmit data, and regularly perform off-site backup and verification; The fault mode recognition models are constructed using multi-layer perceptron MLP and principal component matrices from three perspectives. Based on the three-perspective fault mode recognition models, the models are fused according to weighted coefficients to obtain a comprehensive fault mode vector. The state probability of the node at the next moment is calculated based on the historical fault data, and the state probability of the current time step is calculated by combining the Bayesian theorem with the comprehensive fault mode vector, including: The key nodes of the distribution line are used as nodes of the dynamic Bayesian network DBN, and the node status is used to define whether the node is in a fault state at a certain time point; Determine the dependency of DBN nodes according to the topological structure of the distribution line, draw a directed acyclic graph (DAG), represent the state transition relationship between nodes and determine the network structure of the dynamic Bayesian network; Based on historical fault data, the number of faults that occur at node i and the number of faults that occur at adjacent node j in the next time step are counted and divided by the total number of faults at node i, the single state probability between the two nodes is calculated, and the state transition matrix is ​​constructed based on the state probabilities of different nodes; At the same time, based on the historical fault data, the prior probability of the node is calculated by dividing the total number of faults in the historical data of the node by the total number of acquisitions of all time steps of the node; The state probability of the node at the next moment is calculated based on the state transfer matrix and the state probability at the current moment, which is expressed as: ； Where n represents the total number of nodes, and They represent the state probabilities of node j at time step t+1 and node i at time step t, respectively. Represents the state of node i at time t The state of node j at time t+1 is transferred The transition probability of By using Bayesian theorem, the state probability is combined with the comprehensive fault mode vector to update the state probability of each node at the current time step, which is expressed as: ； in represents the node state probability based on the comprehensive failure mode vector and the current moment as the posterior failure probability, Indicates the node status is The comprehensive failure mode vector of node i is observed The probability of represents the state probability of node i, Indicates the status of all nodes Next, observe The total probability of The state probability of the node at the next moment of the historical data statistics node is recorded, and the accurate state probability of the node at the next moment is selected according to the actual fault information to calculate the sum of the average value and the standard deviation as the prediction threshold. If the state probability of the node at the next moment of the calculation node is greater than or equal to the prediction threshold, a node fault warning is issued.

2. The method for locating a distribution line fault based on the Beidou system according to claim 1, characterized in that: The collecting of key node data and defining the node status according to the status data of the distributed power source, the fault indicator and the smart meter include: Based on the distribution lines, the load center nodes, trunk line branch nodes, substation nodes and fault-prone area nodes are defined as key nodes; Deploy integrated monitoring units at key nodes of distribution lines, including integrated Beidou communication modules, AMI smart meters and fault indicators; Based on the AMI smart meter, the deployed nodes are polled and detected to collect the status data of the AMI smart meter, including voltage, current, active power, reactive power and meter status; Based on the status data of the AMI smart meter, the status of the AMI smart meter is determined. If the meter is online and working normally, the status value is 1; if the meter is disconnected, the status value is 0; At the same time, a judgment is made based on the state data of the fault indicator. If the fault indicator detects a fault current, the state value of the indicator is judged to be 1; if the signal of the fault indicator is normal, the state value of the indicator is judged to be 0; Collect status data of distributed power sources, including real-time power generation, voltage, frequency deviation and current injection.

3. The method for locating a distribution line fault based on the Beidou system as claimed in claim 2, characterized in that: The data matrix is ​​constructed, and the principal component matrix of the state data is calculated. include, A data matrix is ​​constructed based on the state data of distributed power sources, fault indicators and smart meters respectively, where the rows of the matrix represent different observation samples and the columns represent different features; Standardize the matrix number and calculate the covariance matrix of the matrix. Perform eigenvalue decomposition to obtain eigenvalue vectors and eigenvector matrices; Sort by size of eigenvalue vectors, accumulate the sum of eigenvalue vectors until it accounts for 95% of the sum of total eigenvalue vectors, form a dimension reduction matrix, and multiply it with the standardized data matrix to perform data dimension reduction, expressed as: ； in represents the principal component matrix of the ith state data of the reduced dimension, Represents the dimension reduction matrix of the i-th state data, A data matrix representing the data of the i-th state.

4. The method for locating a distribution line fault based on the Beidou system as claimed in claim 3, characterized in that: The method of constructing a fault mode recognition model for the principal component matrix of different state data and performing model fusion includes: The fault mode recognition model is constructed using a multi-layer perceptron MLP and the principal component matrices from three perspectives, which are expressed as: ； in represents the model output of the principal component matrix of the i-th state data, represents the ReLU activation function, represents the weight of the i-th state data, Represents the bias term of the i-th state data; Use the training set and the cross entropy loss function to calculate the error between the model prediction and the actual fault category, use the Adam optimizer and the gradient descent method to iteratively optimize the model parameters, and stop iterating when the model loss no longer decreases significantly during the continuous iteration process, output the model parameters and update the model; Based on the three-perspective fault mode recognition model, the model fusion is performed according to the weighted coefficient, which is expressed as: ； in represents the comprehensive failure mode vector of the kth sample, represents the total number of viewing angles, Represents the weight coefficient of the principal component matrix of the i-th state data; Use the training set and the cross entropy loss function to calculate the error between the model prediction and the actual fault category, calculate the gradient of the loss function for each weighted coefficient, and iteratively optimize the model parameters through the Adam optimizer and the gradient descent method. When the model loss no longer decreases significantly during continuous iterations, stop iterating and output the model parameters to update the model.

5. The method for locating a distribution line fault based on the Beidou system as claimed in claim 4, characterized in that: Analyzing the failure probability of regional nodes according to key nodes includes: Using the high-precision positioning function of the Beidou system, the real-time location coordinates of each monitoring node are obtained, and the Euclidean distance between the two nodes is calculated through the geographic coordinates of the two nodes. , and based on the key nodes, the maximum distance value between all pairs of nodes in the area is selected according to the geographical location and normalized; Based on the posterior fault probability of the current time step and the Euclidean distance of the nodes, the fault location point probability of the regional nodes is calculated, which is expressed as: ； in represents the probability of fault location point of regional node i, represents the posterior failure probability of node j, represents the maximum Euclidean distance of regional nodes, L represents the total number of nodes in the key node area, represents the Euclidean distance between nodes i and j; Based on the fault location point probability of the regional node, the sum of the average fault probability of all regional joint points and the double standard deviation is calculated based on historical data as the location threshold. If the calculated fault location point probability of the regional node is greater than the location threshold, it is judged as a regional fault.

6. The method for locating a distribution line fault based on the Beidou system as claimed in claim 5, characterized in that: The data is securely encrypted and transmitted, and regularly backed up and verified off-site, including: For monitoring nodes such as distributed power sources, fault indicators and smart meters, the information is sent to the monitoring center in real time through the Beidou system terminal equipment using the TLS transport layer security protocol encrypted channel; The detection data AES encryption algorithm is used to encrypt and store the location information of the monitoring node, and the role-based access control RBAC mechanism is used to limit data access rights; Store the backup data to an offsite server and verify the data integrity.

7. A system based on the Beidou system-based distribution line fault location method according to any one of claims 1 to 6, characterized in that: include, Data acquisition module, which collects data based on distributed power supply, fault indicator and smart meter and performs pre-processing; The fault mode recognition module calculates the covariance matrix of the state data matrix of different data sources, performs eigenvalue decomposition, builds a single-view fault mode recognition model, and performs model fusion; The fault prediction module calculates the state probability prediction at the next moment based on historical fault data and calculates the state probability at the current time step based on Bayesian theorem; Regional node failure analysis module calculates the failure probability of all nodes in each key node area and comprehensively analyzes the failure probability distribution of regional nodes; The data security management module encrypts and stores the location information and status data of the monitoring nodes, regularly backs up the data to an off-site server, and verifies the data integrity.

8. A computer device comprising: Memory and processor; The memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the distribution line fault location method based on the Beidou system described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for locating a distribution line fault based on the Beidou system according to any one of claims 1 to 6 are implemented.

Images