Distribution network line fault positioning method based on edge calculation and multi-region cooperation

Abstract

The invention provides a distribution network line fault positioning method based on edge calculation and multi-region cooperation, and relates to the technical field of fault positioning methods, and the method comprises the steps: deploying an edge sensing node based on a distribution network line, collecting distribution network data through the edge sensing node, and carrying out the preprocessing of the distribution network data at the edge sensing node, a decision tree model is embedded in an edge sensing node, an edge calculation function is realized based on the decision tree model, acquired distribution network data are analyzed in real time, and potential faults are detected. According to the method, edge sensing nodes are deployed in each region of a distribution network line, a decision tree model is introduced to realize fault detection, so that fault information can be analyzed and processed in real time at the edge nodes, multi-region cooperative processing is carried out by using a weighted voting algorithm, and when a fault is detected in a certain region, the fault detection efficiency is improved. And the edge sensing nodes and the edge sensing nodes in other areas jointly participate in comparison and verification of fault information, so that the accuracy of fault detection is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of fault location methods, and in particular to a distribution network line fault location method based on edge computing and multi-region collaboration. Background Art

[0002] Fault location of distribution network lines is an important part of power system operation and maintenance, which is related to the stability and safety of power supply. Traditional fault location methods mostly rely on transmitting the collected distribution network data to the central server for analysis and processing; As the scale and complexity of distribution networks increase, centralized processing methods have exposed many deficiencies in data transmission delay, network bandwidth consumption, and processing timeliness. In order to improve the efficiency of fault location of distribution lines, some technical solutions attempt to introduce edge computing technology. By deploying edge sensing nodes at key nodes of distribution lines, local processing and preliminary analysis of some data can be achieved. However, the existing edge computing technology has the following main problems when applied to distribution network fault location: First, the processing capacity of edge nodes in a single area is limited, making it difficult to fully analyze cross-region fault characteristics, resulting in limited positioning accuracy; Secondly, the coordination ability between edge nodes is weak, and it is impossible to effectively integrate multi-regional data resources, which easily leads to inconsistent judgment results; In addition, the existing solutions lack an efficient multi-region collaborative processing mechanism and cannot accurately compare and verify fault information between multiple edge nodes; Therefore, it is urgent to propose a distribution network line fault location method based on edge computing and multi-regional collaboration. Summary of the invention

[0003] The present invention provides a distribution network line fault location method based on edge computing and multi-region collaboration, aiming to solve the technical problem of limited positioning accuracy in related technologies, avoid the use situation where inconsistent judgment results are likely to occur, and solve the subsequent problem of being unable to accurately compare and verify fault information between multiple edge nodes. By deploying edge perception nodes in various areas of the distribution network line and introducing a decision tree model to realize fault detection, the fault information can be analyzed and processed in real time at the edge nodes. At the same time, a weighted voting algorithm is used for multi-region collaborative processing. When a fault is detected in a certain area, the edge perception nodes in other areas jointly participate in the comparison and verification of the fault information, thereby improving the accuracy of fault detection. In addition, the method is based on the collaboration of multiple nodes, and uses signal strength measurement and triangulation to locate the fault position, so as to quickly determine the precise location of the fault point.

[0004] In order to achieve the above purpose, the embodiments of the present application are implemented in the following ways: The embodiment of the present application provides a distribution network line fault location method based on edge computing and multi-region collaboration, including the following: Step S100: deploy edge sensing nodes based on distribution network lines, collect distribution network data through edge sensing nodes, and pre-process the distribution network data at the edge sensing nodes; Step S200, fault detection, embedding a decision tree model in the edge sensing node, implementing edge computing functions based on the decision tree model, performing real-time analysis on the collected distribution network data, and detecting potential faults; Step S300, multi-region collaboration: when an edge sensing node detects a possible fault, it performs collaborative processing with edge sensing nodes in other regions, and the collaborative processing compares and verifies the fault information through a weighted voting algorithm; Step S400, fault location, based on the collaboration of edge sensing nodes in multiple areas, determines the fault location in the distribution network line.

[0005] The present application provides a distribution network line fault location method based on edge computing and multi-regional collaboration, which is based on the distribution network line, deploys edge sensing nodes, collects distribution network data through edge sensing nodes, and pre-processes the distribution network data at the edge sensing nodes, embeds a decision tree model in the edge sensing node, implements edge computing functions based on the decision tree model, performs real-time analysis on the collected distribution network data, detects potential faults, and when the edge sensing node detects a possible fault, performs collaborative processing with edge sensing nodes in other regions, and the collaborative processing compares and verifies the fault information through a weighted voting algorithm, and determines the fault location in the distribution network line based on the collaboration of edge sensing nodes in multiple regions. The method deploys edge sensing nodes in various regions of the distribution network line and introduces a decision tree model to realize fault detection, so that the fault information can be analyzed and processed in real time at the edge node. At the same time, a weighted voting algorithm is used for multi-region collaborative processing. When a fault is detected in a certain area, the edge sensing nodes in other areas jointly participate in the comparison and verification of the fault information, thereby improving the accuracy of fault detection. In addition, the method is based on the collaboration of multiple nodes, and uses signal strength measurement and triangulation to locate the fault location, which can quickly determine the precise location of the fault point.

[0006] In some specific implementations, edge sensing nodes are deployed based on distribution network lines, distribution network data is collected through edge sensing nodes, and the distribution network data is preliminarily screened and preprocessed at the edge sensing nodes, specifically including: Step S101: deploy edge sensing nodes based on the topological structure of the distribution network lines and the load conditions at key locations; Step S102: Collect distribution network data through deployed edge sensing nodes, where the distribution network data includes current, voltage, temperature and humidity data; Step S103: preprocessing the distribution network data. Preprocessing the obtained distribution network data includes normalization, noise suppression and feature extraction.

[0007] In some specific implementations, the distribution network data is preprocessed, and the obtained distribution network data is preprocessed, and the preprocessing includes normalization, noise suppression and feature extraction, specifically including: Step S103.1, normalization, converting the distribution network data collected by the edge sensing nodes of different dimensions into the same dimension, specifically:

[0008] In the formula, is the original distribution network data, and are the minimum and maximum values ​​in the data set, respectively. is the normalized distribution network data; Step S103.2: noise suppression, reducing random noise in the distribution network data collected by the edge sensing node, specifically:

[0009] In the formula, For the Smooth output at all times, For the The original input at time, is the window size, Indicates the position offset of the network configuration data in the current window; Step S103.3, feature extraction, using principal component analysis to map the distribution network data collected by the edge sensing node to a new coordinate system, specifically:

[0010] In the formula, is the original distribution network data matrix, is the eigenvector matrix, The specific value of is obtained by eigenvalue decomposition of the covariance matrix. is the distribution network data matrix after dimension reduction.

[0011] In some specific implementations, fault detection embeds a decision tree model in the edge sensing node, implements edge computing functions based on the decision tree model, performs real-time analysis on the collected distribution network data, and detects potential faults, specifically including: Step S201, initializing the model, loading the decision tree model in the edge sensing node, and initializing the model parameters; Step S202: embedding processing, embedding the decision tree model in the edge sensing node to realize the edge computing function. The decision tree model creates decision nodes based on the features in the distribution network data. Each decision node corresponds to a threshold judgment of a feature, and a decision tree is constructed. Each time the distribution network data is segmented, the information gain is used to calculate the decision tree. Select the best features , specifically:

[0012] In the formula, is the distribution network dataset, Characterized by Features The value of The corresponding sub-distribution network dataset, For the distribution network dataset Entropy of Step S203, feature input, input the preprocessed features into the decision tree model in real time, and the model will judge layer by layer along the tree structure based on the input feature values ​​until it reaches the leaf node, and output the fault type or normal state; Among them, each time a new feature vector is input, the feature vector specifically refers to the distribution network data type, and the distribution network data type includes the current values ​​of voltage, current, temperature and humidity. The model will make judgments on the nodes of the tree in turn, select the corresponding branches, and output the final results based on the branches.

[0013] Step S204, feature matching: the decision tree model matches different fault types based on the features of the input data or determines that the data is in a normal state. If the feature value matches a certain fault mode, the model will give the corresponding fault category at the leaf node; Among them, taking the voltage in the distribution network data as an example, if the voltage is higher than a certain threshold, it enters the "voltage fault" branch, specifically:

[0014] In the formula, Indicates voltage, Indicates the preset voltage threshold, Indicates the judgment condition. The judgment condition means that if the currently measured voltage is higher than the set threshold, the model considers the voltage to be abnormal and enters the "voltage fault" branch of the decision tree. It means that when the condition is met, the decision tree will continue to execute along the branch of voltage fault; Step S205, fault type judgment: the decision tree model determines the fault type according to the judgment result of the current feature. The fault types include "short circuit fault", "overload fault" and "no fault". The decision tree model outputs the judgment result through the leaf node and gives the specific fault type in combination with the current input feature. If the leaf node is a fault, the corresponding type is matched and marked; if it is normal, the "no fault" state is output.

[0015] In some specific implementations, multi-region collaboration, when an edge sensing node detects a possible fault, performs collaborative processing with edge sensing nodes in other regions, and the collaborative processing compares and verifies the fault information through a weighted voting algorithm, specifically including: Step S301, receiving and transmitting fault information. When an edge sensing node in a certain area detects a possible fault, the edge sensing node transmits the fault information to other edge sensing nodes in the relevant area. Step S302: Weighted allocation: Different weights are allocated to the edge sensing nodes in each region based on their geographical location, historical accuracy, and performance factors. The higher the weight value, the higher the credibility of the edge sensing nodes in the region in fault judgment. The specific weight allocation is:

[0016] In the formula, Indicates The weight basis of edge sensing nodes, Indicates the number of edge sensing nodes participating in the voting, Indicates The normalized weights of edge sensing nodes, Indicates participation in voting The weight basis of the edge sensing nodes.

[0017] Step S303: Preliminary voting and comparison: Each edge sensing node compares the received fault information with the fault information in its own area. If similar fault features are detected in multiple areas, the reliability of the fault information will be marked and enhanced. Specifically: For each region’s edge sensing node , calculate its fault feature matching degree , specifically:

[0018] In the formula, represents the number of fault features that match other regions, Represents the number of all features detected by the edge sensing nodes in the area; Step S304: Based on the weighted sum of the voting results, weighted voting is performed based on the fault feature matching degree and weight of each edge sensing node to determine whether the fault is established. The result of the weighted voting is the credibility of each regional edge sensing node for the fault, specifically:

[0019] In the formula, is the total score of weighted voting, For the The normalized weights of edge sensing nodes, For the The fault feature matching degree of each edge sensing node.

[0020] Step S305: Fault verification, based on the result of weighted voting, if the total score of weighted voting Exceeding the preset threshold , then confirm that the fault exists and send the fault information to the control center, specifically:

[0021] In the formula, is the total score of weighted voting, It is the preset minimum fault confirmation standard; Among them, the fault verification rule is: Greater than or equal to ; It will confirm that the detected fault is established, that is, the fault is verified by multiple regional edge sensing nodes and is considered to be a real fault, triggering a fault alarm and taking subsequent maintenance and repair measures. Less than , the fault information is considered invalid and is judged to be a false fault caused by noise or error. No alarm is triggered and the fault will not be recorded as a real fault.

[0022] Here, for example, a set of experimental data is substituted as a demonstration to verify the above fault verification rules, specifically: There are currently three edge sensing nodes , , Participate in fault detection, and their weights are , , , during detection: node Fault feature matching , the data indicates a complete match; node Fault feature matching , this data indicates a higher match; node Fault feature matching , the data indicates a lower match; Weighted total score The calculation is:

[0023] Threshold is set to 0.75, and If it is greater than 0.75, the fault will be confirmed and the alarm will be triggered.

[0024] In some specific implementations, fault location, based on the collaboration of sensing nodes in multiple regions, determines the fault location in the distribution network line, specifically including: Step S401, measuring the strength of the fault signal. After detecting a fault, each edge sensing node participating in the collaboration measures the signal strength from the fault point to the edge sensing node. The strength of the signal is proportional to the distance to the fault. The stronger the signal or the shorter the delay, the closer the distance to the fault point. Each edge sensing node records the signal arrival time and the strength of the fault signal through its monitoring device as a reference indicator of the distance to the fault point. Step S402: estimate the distance. According to the signal strength of each edge sensing node, calculate the distance between the fault point and each edge sensing node, and use the signal attenuation model to estimate the distance. Specifically,

[0025] In the formula, The fault point and The distance of the edge sensing nodes, is the signal strength at the reference point, For the The signal strength measured by the edge sensing node; Step S403: Data fusion: The distance information measured by multiple edge sensing nodes is fused, and the location of the fault point is calculated based on the distance information of at least three edge sensing nodes based on the triangulation method, specifically: The triangulation method solves the coordinates of the fault point through the following equations: :

[0026] In the formula, are the coordinates of the three edge sensing nodes, The distance from the fault point measured by the edge sensing node; Step S404: determine the fault point and calculate the coordinates of the fault point based on the equations solved by triangulation. , that is, the actual fault location in the distribution network line. The coordinates of the fault point are determined by solving the intersection of multiple edge sensing node distance equations. The calculation result will be directly output as the geographical location of the fault point; Step S405, verify the fault location result. After the fault location is determined, the result is fed back to the control center and compared with the actual topological structure of the distribution network line to verify the accuracy of the fault location. The location result will be cross-compared with the topological database of the distribution network line. If there is an error, it will be recalculated.

[0027] In some specific implementations, verifying the fault location result specifically includes: Step S405.1, result feedback, when the coordinates of the fault point After confirmation, the result is fed back to the control center and a preliminary fault location report is generated. After receiving the result, the control center starts the verification process and compares it with the topology of the distribution network line; Step S405.2: Cross-comparison: In the actual topology database of the distribution network, the nearest neighbor matching method is used to find the fault point. As a benchmark, find the edge sensing node closest to it and calculate the distance:

[0028] In the formula, represents the coordinates of the fault point, Indicates the first The coordinates of the edge sensing nodes, Indicates the minimum distance between the fault point and the nearest edge sensing node. The specific location of the fault point in the distribution network line is determined by the nearest neighbor matching method, and its matching with the actual edge sensing node is checked; Step S405.3: Error calculation and threshold judgment, calculate the fault point With the nearest edge sensing node Positioning error between and with the preset threshold Compare and determine whether the positioning is accurate, specifically:

[0029] like , the positioning result is considered accurate and the fault location is confirmed. , it is considered that there is an error in positioning; Step S405.4: Position correction: when the positioning error When the preset threshold is exceeded, the recalculation process is started. The recalculation is based on the new number of edge sensing nodes to improve the accuracy. Specifically: Increase the number of edge sensing nodes participating in the collaboration , and repeat the verification process from step S405.2 to step S405.3:

[0030] In the formula, Indicates the number of edge sensing nodes participating after recalculation, and Respectively represent the updated Normalized weights of edge sensing nodes and fault feature matching, Represents the recalculated weighted total score.

[0031] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 A flow chart of a distribution network line fault location method based on edge computing and multi-region collaboration provided in an embodiment of the present application. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0035] See also Figure 1 , Figure 1 A flowchart of a distribution network line fault location method based on edge computing and multi-region collaboration is provided for an embodiment of the present application.

[0036] In this embodiment, the distribution network line fault location method based on edge computing and multi-region collaboration may include step S100, step S200, step S300 and step S400.

[0037] Step S100: deploy edge perception nodes based on distribution network lines, collect distribution network data through edge perception nodes, and pre-process the distribution network data at the edge perception nodes.

[0038] Here, specifically include: Step S101: deploy edge sensing nodes based on the topological structure of the distribution network lines and the load conditions at key locations; Step S102: Collect distribution network data through deployed edge sensing nodes, where the distribution network data includes current, voltage, temperature and humidity data; Step S103: preprocessing the distribution network data. Preprocessing the obtained distribution network data includes normalization, noise suppression and feature extraction.

[0039] Step S103.1, normalization, converting the distribution network data collected by the edge sensing nodes of different dimensions into the same dimension, specifically:

[0040] In the formula, is the original distribution network data, and are the minimum and maximum values ​​in the data set, respectively. is the normalized distribution network data; Step S103.2: noise suppression, reducing random noise in the distribution network data collected by the edge sensing node, specifically:

[0041] In the formula, For the Smooth output at all times, For the The original input at time, is the window size, Indicates the position offset of the network configuration data in the current window; Step S103.3, feature extraction, using principal component analysis to map the distribution network data collected by the edge sensing node to a new coordinate system, specifically:

[0042] In the formula, is the original distribution network data matrix, is the eigenvector matrix, The specific value of is obtained by eigenvalue decomposition of the covariance matrix. is the distribution network data matrix after dimension reduction.

[0043] Step S200, fault detection, embeds a decision tree model in the edge perception node, implements edge computing function based on the decision tree model, performs real-time analysis on the collected distribution network data, and detects potential faults.

[0044] Here, specifically include: Step S201, initializing the model, loading the decision tree model in the edge sensing node, and initializing the model parameters; Step S202: embedding processing, embedding the decision tree model in the edge sensing node to realize the edge computing function. The decision tree model creates decision nodes based on the features in the distribution network data. Each decision node corresponds to a threshold judgment of a feature, and a decision tree is constructed. Each time the distribution network data is segmented, the information gain is used to calculate the decision tree. Select the best features , specifically:

[0045] In the formula, is the distribution network dataset, Characterized by Features The value of The corresponding sub-distribution network dataset, For the distribution network dataset Entropy of Step S203, feature input, input the preprocessed features into the decision tree model in real time, and the model will judge layer by layer along the tree structure based on the input feature values ​​until it reaches the leaf node, and output the fault type or normal state; Step S204, feature matching: the decision tree model matches different fault types based on the features of the input data or determines that the data is in a normal state. If the feature value matches a certain fault mode, the model will give the corresponding fault category at the leaf node; Step S205, fault type judgment: the decision tree model determines the fault type according to the judgment result of the current feature. The fault types include "short circuit fault", "overload fault" and "no fault". The decision tree model outputs the judgment result through the leaf node and gives the specific fault type in combination with the current input feature. If the leaf node is a fault, the corresponding type is matched and marked; if it is normal, the "no fault" state is output.

[0046] Step S300, multi-region collaboration, when the edge sensing node detects a possible fault, it collaborates with the edge sensing nodes in other regions to compare and verify the fault information through a weighted voting algorithm.

[0047] Here, specifically include: Step S301, receiving and transmitting fault information. When an edge sensing node in a certain area detects a possible fault, the edge sensing node transmits the fault information to other edge sensing nodes in the relevant area. Step S302: Weighted allocation: Different weights are allocated to the edge sensing nodes in each region based on their geographical location, historical accuracy, and performance factors. The higher the weight value, the higher the credibility of the edge sensing nodes in the region in fault judgment. The specific weight allocation is:

[0048] In the formula, Indicates The weight basis of edge sensing nodes, Indicates the number of edge sensing nodes participating in the voting, Indicates The normalized weights of edge sensing nodes, Indicates participation in voting The weight basis of the edge sensing nodes.

[0049] Step S303: Preliminary voting and comparison: Each edge sensing node compares the received fault information with the fault information in its own area. If similar fault features are detected in multiple areas, the reliability of the fault information will be marked and enhanced. Specifically: For each region’s edge sensing node , calculate its fault feature matching degree , specifically:

[0050] In the formula, represents the number of fault features that match other regions, Represents the number of all features detected by the edge sensing nodes in the area; Step S304: Based on the weighted sum of the voting results, weighted voting is performed based on the fault feature matching degree and weight of each edge sensing node to determine whether the fault is established. The result of the weighted voting is the credibility of each regional edge sensing node for the fault, specifically:

[0051] In the formula, is the total score of weighted voting, For the The normalized weights of edge sensing nodes, For the The fault feature matching degree of each edge sensing node.

[0052] Step S305: Fault verification, based on the result of weighted voting, if the total score of weighted voting Exceeding the preset threshold , then confirm that the fault exists and send the fault information to the control center, specifically:

[0053] In the formula, is the total score of weighted voting, It is the preset minimum fault confirmation standard.

[0054] Step S400, fault location, based on the collaboration of edge sensing nodes in multiple areas, determines the fault location in the distribution network line.

[0055] Here, specifically include: Step S401, measuring the strength of the fault signal. After detecting a fault, each edge sensing node participating in the collaboration measures the signal strength from the fault point to the edge sensing node. The strength of the signal is proportional to the distance to the fault. The stronger the signal or the shorter the delay, the closer the distance to the fault point. Each edge sensing node records the signal arrival time and the strength of the fault signal through its monitoring device as a reference indicator of the distance to the fault point. Step S402: estimate the distance. According to the signal strength of each edge sensing node, calculate the distance between the fault point and each edge sensing node, and use the signal attenuation model to estimate the distance. Specifically,

[0056] In the formula, The fault point and The distance of the edge sensing nodes, is the signal strength at the reference point, For the The signal strength measured by the edge sensing node; Step S403: Data fusion: The distance information measured by multiple edge sensing nodes is fused, and the location of the fault point is calculated based on the distance information of at least three edge sensing nodes based on the triangulation method, specifically: The triangulation method solves the coordinates of the fault point through the following equations: :

[0057] In the formula, are the coordinates of the three edge sensing nodes, The distance from the fault point measured by the edge sensing node; Step S404: determine the fault point and calculate the coordinates of the fault point based on the equations solved by triangulation. , that is, the actual fault location in the distribution network line. The coordinates of the fault point are determined by solving the intersection of multiple edge sensing node distance equations. The calculation result will be directly output as the geographical location of the fault point; Step S405: Verify the fault location result. After the fault location is determined, the result is fed back to the control center and compared with the actual topological structure of the distribution network line to verify the accuracy of the fault location. The location result will be cross-compared with the topological database of the distribution network line. If there is an error, it will be recalculated; Step S405.1, result feedback, when the coordinates of the fault point After confirmation, the result is fed back to the control center and a preliminary fault location report is generated. After receiving the result, the control center starts the verification process and compares it with the topology of the distribution network line; Step S405.2: Cross-comparison: In the actual topology database of the distribution network, the nearest neighbor matching method is used to find the fault point. As a benchmark, find the edge sensing node closest to it and calculate the distance:

[0058] In the formula, represents the coordinates of the fault point, Indicates the first The coordinates of the edge sensing nodes, Indicates the minimum distance between the fault point and the nearest edge sensing node. The specific location of the fault point in the distribution network line is determined by the nearest neighbor matching method, and its matching with the actual edge sensing node is checked; Step S405.3: Error calculation and threshold judgment, calculate the fault point With the nearest edge sensing node Positioning error between and with the preset threshold Compare and determine whether the positioning is accurate, specifically:

[0059] like , the positioning result is considered accurate and the fault location is confirmed. , it is considered that there is an error in positioning; Step S405.4: Position correction: when the positioning error When the preset threshold is exceeded, the recalculation process is started. The recalculation is based on the new number of edge sensing nodes to improve the accuracy. Specifically: Increase the number of edge sensing nodes participating in the collaboration , and repeat the verification process from step S405.2 to step S405.3:

[0060] In the formula, Indicates the number of edge sensing nodes participating after recalculation, and Respectively represent the updated Normalized weights of edge sensing nodes and fault feature matching, Represents the recalculated weighted total score.

[0061] When this method is actually put into use, first, edge sensing nodes are deployed at key locations of distribution network lines according to the topology and load conditions to ensure that each node can collect current, voltage, temperature and humidity data in real time. The deployed edge sensing nodes collect data on the distribution network lines and transmit the real-time current, voltage, temperature and humidity data to the local edge computing. The collected data is preprocessed at the edge sensing nodes, including normalization, noise suppression and feature extraction. Subsequently, the edge computing function is embedded in the edge sensing node, and the decision tree model is loaded and initialized. The model is used to perform real-time analysis on the distribution network data. The preprocessed data is input into the decision tree model. The model makes decisions based on the current, voltage and other characteristics to determine the type of fault. The model receives the feature data input in real time, and gradually identifies the fault type through the conditional judgment of the decision tree. If the feature data matches a certain fault mode, the fault type is output, otherwise the "no fault" state is output. The decision tree model finally determines the type of fault (such as short circuit fault, overload fault, etc.) and generates a preliminary judgment result at the edge node. It should be noted that the fault information at this time is marked and ready to be shared with other regional nodes; Then, when an edge sensing node detects a possible fault, it will broadcast the fault information to the sensing nodes in the adjacent area so that other nodes can participate in the collaborative verification. The sensing nodes in each area assign weights according to various factors, and compare the received fault information with the data in their own area to calculate the fault feature matching degree. Through the weighted voting algorithm, the fault judgment results of each node are weighted and summed to obtain the total fault credibility score. If the weighted total score exceeds the preset threshold, the fault is confirmed to exist, an alarm is generated and the information is sent to the control center; otherwise, the fault information is considered invalid and no alarm is triggered. After that, after detecting a fault, each sensing node participating in the collaboration measures the signal strength from the fault point to the node, using this as a reference indicator of the distance to the fault point. The signal attenuation model is used to calculate the distance between the fault point and each sensing node. Based on the distance data of at least three sensing nodes, the triangulation method is used to calculate the specific coordinates of the fault point, determine the actual fault location in the distribution network line, and obtain the precise location of the fault point by solving the triangulation equation group, and output the geographical coordinates of the location. Finally, the fault location result is fed back to the control center and a fault report is generated. The coordinates of the fault point are compared with the nodes in the topological structure of the distribution network line using the nearest neighbor matching method to calculate the location error. If the error is less than the preset threshold, the location result is considered accurate. Otherwise, the recalculation process is started to increase the number of collaborative sensing nodes to improve the location accuracy. After recalculation, the new coordinates of the fault point are obtained and the error calculation process is repeated until the location error meets the requirements. It should be noted that the confirmed fault location and type will be stored in the fault database, and the control center will generate a detailed fault report and notify the maintenance personnel to conduct on-site inspection and repair. The fault report includes the following: Fault occurrence time: Record the specific timestamp of fault detection to facilitate tracking the timeline of the event; Detection location: including the area where the fault occurs and the location information of the edge sensing nodes involved in fault detection and coordination; Summary of collected data: including the values ​​of current, voltage, temperature, humidity and other data collected when the fault occurs, as well as the characteristics of the data after preprocessing; Fault type: describes in detail the fault type identified by the decision tree and the characteristic basis that leads to the judgment. For example, "the voltage exceeds the threshold and is judged as an overload fault"; Weighted voting results: Displays the weight, feature matching degree, and total voting score of each participating node in the multi-regional collaboration process; Fault location coordinates: geographical coordinates of the fault point calculated by triangulation; Information about sensing nodes involved in fault location: lists all sensing nodes involved in fault location, including node coordinates and distance measurements; Error analysis: Displays the positioning error between the fault location and the nearest distribution network node, and whether it meets the preset threshold. If recalculated, it also includes the new coordinates after recalculation and the adjustment process; Comparison of historical fault data: Compare the current fault with similar faults that occurred at the same location in the past to analyze whether there are repetitive or similar problems; Report generation time: Record the specific time when the report was generated to record the timeline of troubleshooting.

[0062] It should be noted that the control center refers to the central management platform in the distribution network method used to receive, analyze, store and coordinate the processing of data and fault information from various edge sensing nodes. It plays the role of data aggregation, analysis verification, command issuance and coordination control in the entire fault detection and location process, and is responsible for centralized management and decision-making of the detection data and analysis results of edge nodes.

[0063] It should be understood that the "one embodiment" or "an embodiment" or "a feasible implementation" or "some implementations" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the embodiments of the present invention. Therefore, "in one embodiment" or "in an embodiment" or "a feasible implementation" or "some implementations" appearing in various places throughout the specification do not necessarily refer to the same embodiment. In addition, those skilled in the art should also be aware that the embodiments described in the specification are all optional embodiments, and the steps and modules involved are not necessarily required for the embodiments of the present invention.

[0064] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of the processes and / or blocks in the flowchart and / or block diagram, can be implemented by instructions. These instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the process in the flowchart. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0065] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0066] These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0067] The above is only a specific implementation of the present invention, but the protection scope of the embodiments of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any technician familiar with the technical field within the technical scope disclosed in the embodiments of the present invention should be included in the protection scope of the embodiments of the present invention. Therefore, the protection scope of the embodiments of the present invention shall be based on the protection scope of the claims.

Claims

1. A distribution network line fault location method based on edge computing and multi-region collaboration, characterized in that: These include: S100, deploying edge perception nodes based on distribution network lines, collecting distribution network data through the edge perception nodes, and preprocessing the distribution network data at the edge perception nodes; S200, fault detection, embedding a decision tree model in the edge sensing node, realizing edge computing function based on the decision tree model, performing real-time analysis on the collected distribution network data, and detecting potential faults; S300, multi-region collaboration, when an edge sensing node detects a possible fault, it performs collaborative processing with edge sensing nodes in other regions, and the collaborative processing compares and verifies the fault information through a weighted voting algorithm; S400, fault location, based on the collaboration of edge sensing nodes in multiple areas, determines the fault location in the distribution network line.

2. The distribution network line fault location method based on edge computing and multi-region collaboration as claimed in claim 1, characterized in that: Based on the distribution network lines, edge perception nodes are deployed to collect distribution network data through edge perception nodes, and the distribution network data is preliminarily screened and preprocessed at the edge perception nodes, including: S101, deploy edge sensing nodes, and deploy edge sensing nodes based on the topological structure of the distribution network lines and the load conditions at key locations; S102, collecting distribution network data, collecting data through deployed edge sensing nodes, the distribution network data including current, voltage, temperature and humidity data; S103, preprocessing the distribution network data, preprocessing the obtained distribution network data, the preprocessing includes normalization, noise suppression and feature extraction.

3. The distribution network line fault location method based on edge computing and multi-region collaboration as claimed in claim 2, characterized in that: Distribution network data preprocessing: The obtained distribution network data is preprocessed. The preprocessing includes normalization, noise suppression and feature extraction. Specifically: S103.

1. Normalization: convert the distribution network data collected by edge sensing nodes of different dimensions into the same dimension, specifically: ； In the formula, is the original distribution network data, and are the minimum and maximum values ​​in the data set, respectively. is the normalized distribution network data; S103.2, noise suppression, reducing random noise in the distribution network data collected by the edge sensing node, specifically: ； In the formula, For the Smooth output at all times, For the The original input at time, is the window size, Indicates the position offset of the network configuration data in the current window; S103.3, feature extraction, using principal component analysis to map the distribution network data collected by the edge sensing node to a new coordinate system, specifically: ； In the formula, is the original distribution network data matrix, is the eigenvector matrix, The specific value of is obtained by eigenvalue decomposition of the covariance matrix. is the distribution network data matrix after dimension reduction.

4. The distribution network line fault location method based on edge computing and multi-region collaboration as claimed in claim 1, characterized in that: Fault detection: embed a decision tree model in the edge sensing node, implement edge computing based on the decision tree model, analyze the collected distribution network data in real time, and detect potential faults, including: S201, initializing the model, loading the decision tree model in the edge sensing node, and initializing the model parameters; S202, embedding processing, embedding the decision tree model in the edge perception node to realize the edge computing function, the decision tree model creates decision nodes based on the features in the distribution network data, each decision node corresponds to a threshold judgment of a feature, constructs a decision tree, and each time the distribution network data is segmented, the information gain is used to Select the best features , specifically: ； In the formula, is the distribution network dataset, Characterized by Features The value of The corresponding sub-distribution network dataset, For the distribution network dataset Entropy of S203, feature input, input the preprocessed features into the decision tree model in real time, the model will judge layer by layer along the tree structure based on the input feature values ​​until it reaches the leaf node, and output the fault type or normal state; S204, feature matching: the decision tree model matches different fault types or determines that the data is in a normal state based on the features of the input data. If the feature value matches a certain fault mode, the model will give the corresponding fault category at the leaf node; S205, fault type judgment: the decision tree model determines the fault type according to the judgment result of the current feature, and the fault types include "short circuit fault", "overload fault" and "no fault". The decision tree model outputs the judgment result through the leaf node, and gives the specific fault type in combination with the current input feature. If the leaf node is a fault, the corresponding type is matched and marked; if it is normal, the "no fault" state is output.

5. The distribution network line fault location method based on edge computing and multi-region collaboration as claimed in claim 1, characterized in that: Multi-region collaboration: When an edge sensing node detects a possible fault, it collaborates with edge sensing nodes in other regions to compare and verify the fault information through a weighted voting algorithm, specifically including: S301, receiving and transmitting fault information. When an edge sensing node in a certain area detects a possible fault, the edge sensing node transmits the fault information to other edge sensing nodes in the relevant area. S302: Weighted allocation: Different weights are allocated to the edge sensing nodes in each region based on their geographical location, historical accuracy, and performance factors. The higher the weight value, the higher the credibility of the edge sensing nodes in the region in fault judgment. The specific weight allocation is: ； In the formula, Indicates The weight basis of edge sensing nodes, Indicates the number of edge sensing nodes participating in the voting, Indicates The normalized weights of edge sensing nodes, Indicates participation in voting The weight benchmark of edge sensing nodes; S303, preliminary voting and comparison: each edge sensing node compares the received fault information with the fault information in its own area. If similar fault features are detected in multiple areas, the reliability of the fault information will be marked and enhanced. Specifically: For each region’s edge sensing node , calculate its fault feature matching degree , specifically: ； In the formula, represents the number of fault features that match other regions, Represents the number of all features detected by the edge sensing nodes in the area; S304: Based on the weighted sum of the voting results, weighted voting is performed based on the fault feature matching degree and weight of each edge sensing node to determine whether the fault is established. The result of the weighted voting is the credibility of each regional edge sensing node for the fault, specifically: ； In the formula, is the total score of weighted voting, Indicates The normalized weights of edge sensing nodes, For the The fault feature matching degree of edge sensing nodes; S305, fault verification, based on the result of weighted voting, if the total score of weighted voting Exceeding the preset threshold , then confirm that the fault exists and send the fault information to the control center, specifically: ； In the formula, is the total score of weighted voting, It is the preset minimum fault confirmation standard.

6. The distribution network line fault location method based on edge computing and multi-region collaboration according to claim 1, characterized in that: Fault location, based on the collaboration of sensing nodes in multiple areas, determines the fault location in the distribution network line, including: S401, measuring fault signal strength. After detecting a fault, each edge sensing node participating in the collaboration measures the signal strength from the fault point to the edge sensing node. The strength of the signal is proportional to the fault distance. The stronger the signal or the shorter the delay, the closer the distance to the fault point. Each edge sensing node records the signal arrival time and the strength of the fault signal through its monitoring device as a reference indicator of the distance to the fault point. S402: Estimating the distance: According to the signal strength of each edge sensing node, the distance between the fault point and each edge sensing node is calculated, and the signal attenuation model is used to estimate the distance, specifically: ； In the formula, The fault point and The distance of the edge sensing nodes, is the signal strength at the reference point, For the The signal strength measured by the edge sensing node; S403, data fusion, fusing the distance information measured by multiple edge sensing nodes, and calculating the location of the fault point based on the distance information of at least three edge sensing nodes based on the triangulation method, specifically: The triangulation method solves the coordinates of the fault point through the following equations: : ； In the formula, are the coordinates of the three edge sensing nodes, The distance from the fault point measured by the edge sensing node; S404, determine the fault point, and calculate the coordinates of the fault point according to the equations solved by triangulation method , that is, the actual fault location in the distribution network line. The coordinates of the fault point are determined by solving the intersection of multiple edge sensing node distance equations. The calculation result will be directly output as the geographical location of the fault point; S405. Verify the fault location result. After the fault location is determined, the result is fed back to the control center and compared with the actual topology of the distribution network line to verify the accuracy of the fault location. The location result will be cross-compared with the topology database of the distribution network line. If there is an error, it will be recalculated.

7. The distribution network line fault location method based on edge computing and multi-region collaboration as claimed in claim 6, characterized in that: Verify the fault location results, including: S405.1, result feedback, when the coordinates of the fault point After confirmation, the result is fed back to the control center and a preliminary fault location report is generated. After receiving the result, the control center starts the verification process and compares it with the topology of the distribution network line; S405.2, cross comparison, in the actual topology database of the distribution network line, the nearest neighbor matching method is used to As a benchmark, find the edge sensing node closest to it and calculate the distance: ； In the formula, represents the coordinates of the fault point, Indicates the first The coordinates of the edge sensing nodes, Indicates the minimum distance between the fault point and the nearest edge sensing node. The specific location of the fault point in the distribution network line is determined by the nearest neighbor matching method, and its matching with the actual edge sensing node is checked; S405.

3. Error calculation and threshold judgment, calculation of fault point With the nearest edge sensing node Positioning error between and with the preset threshold Compare and determine whether the positioning is accurate, specifically: ； like , the positioning result is considered accurate and the fault location is confirmed. , it is considered that there is an error in positioning; S405.4, position correction, when positioning error When the preset threshold is exceeded, the recalculation process is started. The recalculation is based on the new number of edge sensing nodes to improve the accuracy. Specifically: Increase the number of edge sensing nodes participating in the collaboration , and repeat the verification process from S405.2 to S405.3: ； In the formula, Indicates the number of edge sensing nodes participating after recalculation, and Respectively represent the updated Normalized weights of edge sensing nodes and fault feature matching, Represents the recalculated weighted total score.

Images